Reference Guide

Introduction

Overview

Spring XD is a unified, distributed, and extensible service for data ingestion, real time analytics, batch processing, and data export. The Spring XD project is an open source Apache 2 License licenced project whose goal is to tackle big data complexity. Much of the complexity in building real-world big data applications is related to integrating many disparate systems into one cohesive solution across a range of use-cases. Common use-cases encountered in creating a comprehensive big data solution are

  • High throughput distributed data ingestion from a variety of input sources into big data store such as HDFS or Splunk

  • Real-time analytics at ingestion time, e.g. gathering metrics and counting values.

  • Workflow management via batch jobs. The jobs combine interactions with standard enterprise systems (e.g. RDBMS) as well as Hadoop operations (e.g. MapReduce, HDFS, Pig, Hive or HBase).

  • High throughput data export, e.g. from HDFS to a RDBMS or NoSQL database.

The Spring XD project aims to provide a one stop shop solution for these use-cases.

Getting Started

Requirements

To get started, make sure your system has as a minimum Java JDK 7 or newer installed. Java JDK 7 is recommended.

Download Spring XD

To download the current GA release, you can download the distribution spring-xd-1.3.2.RELEASE-dist.zip and its accompanying documentation.

If you want to try out the latest build of Spring XD, You can download the snapshot distribution from the spring snapshots repository. You can also build the project from source if you wish. The wiki content should also be kept up to date with the current snapshot so if you are reading this on the github website, things may have changed since the last stable release.

Unzip the distribution which will unpack to a single installation directory. All the commands below are executed from this directory, so change into it before proceeding.

$ cd spring-xd-1.3.3.BUILD-SNAPSHOT

Install Spring XD

Spring XD can be run in two different modes. There’s a single-node runtime option for testing and development, and there’s a distributed runtime which supports distribution of processing tasks across multiple nodes. This document will get you up and running quickly with a single-node runtime. See Running Distributed Mode for details on setting up a distributed runtime.

You can also install Spring XD using homebrew on OSX and RPM on RedHat/CentOS.

Start the Runtime and the XD Shell

The single node option is the easiest to get started with. It runs everything you need in a single process. To start it, you just need to cd to the xd directory and run the following command

xd/bin>$ ./xd-singlenode

In a separate terminal, cd into the shell directory and start the XD shell, which you can use to issue commands.

shell/bin>$ ./xd-shell
 _____                           __   _______
/  ___|          (-)             \ \ / /  _  \
\ `--. _ __  _ __ _ _ __   __ _   \ V /| | | |
 `--. \ '_ \| '__| | '_ \ / _` |  / ^ \| | | |
/\__/ / |_) | |  | | | | | (_| | / / \ \ |/ /
\____/| .__/|_|  |_|_| |_|\__, | \/   \/___/
      | |                  __/ |
      |_|                 |___/
eXtreme Data
1.3.3.BUILD-SNAPSHOT | Admin Server Target: http://localhost:9393
Welcome to the Spring XD shell. For assistance hit TAB or type "help".
xd:>

The shell is a more user-friendly front end to the REST API which Spring XD exposes to clients. The URL of the currently targeted Spring XD server is shown at startup.

Note

If the server could not be reached, the prompt will read

server-unknown:>

You can then use the admin config server <url> to attempt to reconnect to the admin REST endpoint once you’ve figured out what went wrong:

admin config server http://localhost:9393

You should now be able to start using Spring XD.

Tip

Spring XD uses ZooKeeper internally which typically runs as an external process. XD singlenode runs with an embedded ZooKeeper server and assigns a random available port. This keeps things very simple. However if you already have a ZooKeeper ensemble set up and want to connect to it, you can edit xd/config/servers.yml:

#Zookeeper properties
# client connect string: host1:port1,host2:port2,...,hostN:portN
zk:
  client:
     connect: localhost:2181

Also, sometimes it is useful in troubleshooting to connect the ZooKeeper CLI to the embedded server. The assigned server port is listed in the console log, but you can also set the port directly by setting the property zk.embedded.server.port in servers.yml or set JAVA_OPTS before starting xd-singlenode.

$export JAVA_OPTS=-Dzk.embedded.server.port=<port>

Create a Stream

In Spring XD, a basic stream defines the ingestion of event driven data from a source to a sink that passes through any number of processors. You can create a new stream by issuing a stream create command from the XD shell. Stream definitions are built from a simple DSL. For example, execute:

xd:> stream create --name ticktock --definition "time | log" --deploy

This defines a stream named ticktock based off the DSL expression time | log. The DSL uses the "pipe" symbol |, to connect a source to a sink. The stream server finds the time and log definitions in the modules directory and uses them to setup the stream. In this simple example, the time source simply sends the current time as a message each second, and the log sink outputs it using the logging framework at the WARN logging level. Since the --deploy flag was provided, this stream will be deployed immediately. In the console where you started the server, you will see log output similar to that listed below

13:09:53,812  INFO http-bio-8080-exec-1 module.SimpleModule:109 - started module: Module [name=log, type=sink]
13:09:53,813  INFO http-bio-8080-exec-1 module.ModuleDeployer:111 - launched sink module: ticktock:log:1
13:09:53,911  INFO http-bio-8080-exec-1 module.SimpleModule:109 - started module: Module [name=time, type=source]
13:09:53,912  INFO http-bio-8080-exec-1 module.ModuleDeployer:111 - launched source module: ticktock:time:0
13:09:53,945  WARN task-scheduler-1 logger.ticktock:141 - 2013-06-11 13:09:53
13:09:54,948  WARN task-scheduler-1 logger.ticktock:141 - 2013-06-11 13:09:54
13:09:55,949  WARN task-scheduler-2 logger.ticktock:141 - 2013-06-11 13:09:55

To stop the stream, and remove the definition completely, you can use the stream destroy command:

xd:>stream destroy --name ticktock

It is also possible to stop and restart the stream instead, using the undeploy and deploy commands. The shell supports command completion so you can hit the tab key to see which commands and options are available.

Explore Spring XD

Learn about the modules available in Spring XD in the Sources, Processors, and Sinks sections of the documentation.

Don’t see what you’re looking for? Create a custom module: source, processor or sink (and then consider contributing it back to Spring XD).

Want to add some analytics to your stream? Check out the Taps and Analytics sections.

OSX Homebrew installation

If you are on a Mac and using homebrew, all you need to do to install Spring XD is:

$ brew tap pivotal/tap
$ brew install springxd

Homebrew will install springxd to /usr/local/bin. Now you can jump straight into using Spring XD:

$ xd-singlenode

Brew install also allows you to run Spring XD in distributed mode on you OSx. See Running Distributed Mode for details on setting up a distributed runtime.

RedHat/CentOS Installation

If you are using RHEL or CentOS v. 6.x you can install Spring XD using our RPM package. See the wiki page for instructions.

Running in Distributed Mode

Introduction

The Spring XD distributed runtime (DIRT) supports distribution of processing tasks across multiple nodes. See Getting Started for information on running Spring XD as a single node.

The XD distributed runtime architecture consists of the following distributed components:

  • Admin - Manages Stream and Job deployments and other end user operations and provides REST services to access runtime state, system metrics, and analytics

  • Container - Hosts deployed Modules (stream processing tasks) and batch jobs

  • ZooKeeper - Provides all runtime information for the XD cluster. Tracks running containers, in which containers modules and jobs are deployed, stream definitions, deployment manifests, and the like, see XD Distributed Runtime for an overview on how XD uses ZooKeeper.

  • Spring Batch Job Repository Database - An RDBMS is required for jobs. The XD distribution comes with HSQLDB, but this is not appropriate for a production installation. XD supports any JDBC compliant database.

  • A Message Broker - Used for data transport. XD data transport is designed to be pluggable. Currently XD supports Rabbit MQ and Redis for messaging during stream and job processing, and Kafka for messaging during stream processing only. Please note that support for job processing using Kafka as transport is not currently available. A production installation must configure one of these transport options.

  • Analytics Repository - XD currently uses Redis to store the counters and gauges provided Analytics

In addition, XD provides a Command Line Interface (CLI), XD Shell as well as a web application, XD-UI to interact with the XD runtime.

Spring XD Distributed Runtime
XD CommandLine Options

The XD distribution provides shell scripts to start its runtime components under the xd directory of the XD installation:

Whether you are running _xd-admin, xd-container or even xd-singlenode you can always get help by typing the command followed by --help. For example:

xd/bin/xd-admin --help

 _____                           __   _______
/  ___|          (-)             \ \ / /  _  \
\ `--. _ __  _ __ _ _ __   __ _   \ V /| | | |
 `--. \ '_ \| '__| | '_ \ / _` |  / ^ \| | | |
/\__/ / |_) | |  | | | | | (_| | / / \ \ |/ /
\____/| .__/|_|  |_|_| |_|\__, | \/   \/___/
      | |                  __/ |
      |_|                 |___/
1.3.3.BUILD-SNAPSHOT             eXtreme Data


Started : AdminServerApplication
Documentation: https://github.com/spring-projects/spring-xd/wiki

Usage:
 --analytics [redis]   : How to persist analytics such as counters and gauges
 --help (-?, -h)       : Show this help screen
 --httpPort <httpPort> : Http port for the REST API server
 --mgmtPort <mgmtPort> : The port for the management server
xd-admin command line args:
  • analytics - The data store that will be used to store the analytics data. The default is redis

  • help - Displays help for the command args. Help information may be accessed with a -? or -h.

  • httpPort - The http port for the REST API server. Defaults to 9393.

  • mgmtPort - The port for the management server. Defaults to the admin server port.

Also, note that it is recommended to use fixed http port for XDAdmin(s). This makes it easy to know the admin server addresses the REST clients (shell, webUI) can point to. If a random port is chosen (with server.port or $PORT set to 0), then one needs to go through the log and find which port admin server’s tomcat starts at.

xd-container command line args:
  • analytics - How to persist analytics such as counters and gauges. The default is redis

  • groups - The assigned group membership for this container as a comma delimited list

  • hadoopDistro - The Hadoop distribution to be used for HDFS access. HDFS is not available if not set.

  • help - Displays help for the command args. Help information may be accessed with a -? or -h.

  • mgmtPort - The port for the management server. Defaults to the container server port.

Setting up a RDBMS

The distributed runtime requires an RDBMS. The XD distrubution comes with an HSQLDB in memory database for testing purposes, but an alternate is expected. To start HSQLDB:

$ cd hsqldb/bin
$ ./hsqldb-server

To configure XD to connect to a different RDBMS, have a look at xd/config/servers.yml in the spring:datasource section for details. Note that spring.batch.initializer.enabled is set to true by default which will initialize the Spring Batch schema if it is not already set up. However, if those tables have already been created, they will be unaffected.

If the provided schemas are customized, other values may need to be customized. In the xd/config/servers.yml the following block exposes database specific values for the batch job repository.

  spring:
    batch:
      isolationLevel:             ISOLATION_SERIALIZABLE  (1)
      clobType:                                           (2)
      dbType:                                             (3)
      maxVarcharLength:           2500                    (4)
      tablePrefix:                BATCH_                  (5)
      validateTransactionState:   true                    (6)
      initializer:
        enabled:                  false                   (7)
  1. Transaction isolation level for the job repository.

  2. A special handler for large objects. The default is usually fine, except for some (usually older) versions of Oracle. The default is determined from the data base type.

  3. Used to determine what id incremented to use. The default is usually fine, except when the type returned by the datasource should be overridden (GemfireXD for example).

  4. Configures how large the maximum message can be stored in a VARCHAR type field.

  5. Prefix for repository tables.

  6. Flag to determine whether to check for an existing transaction when a JobExecution is created. Defaults to true because it is usually a mistake, and leads to problems with restartability and also to deadlocks in multi-threaded steps.

  7. Flag that indicates if the database tables should be created on startup.

Setting up ZooKeeper

Currently XD does not ship with ZooKeeper. At the time of this writing, the compliant version is 3.4.6 and you can download it from here. Please refer to the ZooKeeper Getting Started Guide for more information. A ZooKeeper ensemble consisting of at least three members is recommended for production installations, but a single server is all that is needed to have XD up and running.

You can configure the root path in Zookeeper where an XD cluster’s top level nodes will be created. This allows you to run multiple independent clusters of XD that share a single ZK instance. Add the following to servers.yml to configure. You can also set as an environment variable, system property in the standard manner.

Additionally, various time related settings may be optionally configured for ZooKeeper:

zk:
  namespace: xd
  client:
     connect: localhost:2181
     sessionTimeout: 60000
     connectionTimeout: 30000
     initialRetryWait: 1000
     retryMaxAttempts: 3

Setting up Redis

Redis is the default transport when running in distributed mode.

Installing Redis

If you already have a running instance of Redis it can be used for Spring XD. By default Spring XD will try to use a Redis instance running on localhost using port 6379. You can change that in the servers.yml file residing in the config/ directory.

If you don’t have a pre-existing installation of Redis, you can use the Spring XD provided instance (For Linux and Mac) which is included in the .zip download. If you are installing using brew or rpm you should install Redis using those installers or download the source tarball and compile Redis yourself. If you used the .zip download then inside the Spring XD installation directory (spring-xd) do:

$ cd redis/bin
$ ./install-redis

This will compile the Redis source tar and add the Redis executables under redis/bin:

  • redis-check-dump

  • redis-sentinel

  • redis-benchmark

  • redis-cli

  • redis-server

You are now ready to start Redis by executing

$ ./redis-server
Tip
For further information on installing Redis in general, please checkout the Redis Quick Start guide. If you are using Mac OS, you can also install Redis via Homebrew
Troubleshooting
Redis on Windows

Presently, Spring XD does not ship Windows binaries for Redis (See XD-151). However, Microsoft is actively working on supporting Redis on Windows. You can download Windows Redis binaries from:

Redis is not running

If you try to run Spring XD and Redis is NOT running, you will see the following exception:

11:26:37,830 ERROR main launcher.RedisContainerLauncher:85 - Unable to connect to Redis on localhost:6379; nested exception is com.lambdaworks.redis.RedisException: Unable to connect
Redis does not seem to be running. Did you install and start Redis? Please see the Getting Started section of the guide for instructions.
Starting Redis
$ redis-server

You should see something like this:

11655:C 08 Jun 14:45:04.167 # Warning: no config file specified, using the default config. In order to specify a config file use ./redis-server /path/to/redis.conf
                _._
           _.-``__ ''-._
      _.-``    `.  `_.  ''-._           Redis 3.0.1 (bddc928a/0) 64 bit
  .-`` .-```.  ```\/    _.,_ ''-._
 (    '      ,       .-`  | `,    )     Running in standalone mode
 |`-._`-...-` __...-.``-._|'` _.-'|     Port: 6379
 |    `-._   `._    /     _.-'    |     PID: 11655
  `-._    `-._  `-./  _.-'    _.-'
 |`-._`-._    `-.__.-'    _.-'_.-'|
 |    `-._`-._        _.-'_.-'    |           http://redis.io
  `-._    `-._`-.__.-'_.-'    _.-'
 |`-._`-._    `-.__.-'    _.-'_.-'|
 |    `-._`-._        _.-'_.-'    |
  `-._    `-._`-.__.-'_.-'    _.-'
      `-._    `-.__.-'    _.-'
          `-._        _.-'
              `-.__.-'

11655:M 08 Jun 14:45:04.169 # Server started, Redis version 3.0.1
11655:M 08 Jun 14:45:04.169 * DB loaded from disk: 0.000 seconds
11655:M 08 Jun 14:45:04.169 * The server is now ready to accept connections on port 6379

Using RabbitMQ

Installing RabbitMQ

If you already have a running instance of RabbitMQ it can be used for Spring XD. By default Spring XD will try to use a Rabbit instance running on localhost using port 5672. The default account credentials of guest/guest are assumed. You can change that in the servers.yml file residing in the config/ directory.

If you don’t have a RabbitMQ installation already, head over to http://www.rabbitmq.com and follow the instructions. Packages are provided for Windows, Mac and various flavor of unix/linux.

Launching RabbitMQ

Start the RabbitMQ broker by running the rabbitmq-server script:

$ rabbitmq-server

You should see something similar to this:

              RabbitMQ 3.3.0. Copyright (C) 2007-2013 GoPivotal, Inc.
  ##  ##      Licensed under the MPL.  See http://www.rabbitmq.com/
  ##  ##
  ##########  Logs: /usr/local/var/log/rabbitmq/[email protected]
  ######  ##        /usr/local/var/log/rabbitmq/[email protected]
  ##########
              Starting broker... completed with 10 plugins.

Starting Spring XD in Distributed Mode

Spring XD consists of two servers

  • XDAdmin - controls deployment of modules into containers

  • XDContainer - executes modules

You can start the xd-container and xd-admin servers individually as follows:

xd/bin>$ ./xd-admin
xd/bin>$ ./xd-container
Choosing a Transport

Spring XD uses data transport for sending data from the output of one module to the input of the next module. In general, this requires remote transport between container nodes. The Admin server also uses the data bus to launch batch jobs by sending a message to the job’s launch channel. Since the same transport must be shared by the Admin and all Containers, the transport configuration is centrally configured in xd/config/servers.yml. The default transport is redis. Open servers.yml with a text editor and you will see the transport configuration near the top. To change the transport, you can uncomment this section and change the transport to rabbit or any other supported transport. Any changes to the transport configuration must be replicated to every XD node in the cluster.

Note
XD singlenode also supports a --transport command line argument, useful for testing streams under alternate transports.
#xd:
#  transport: redis
Note

If you have multiple XD instances running share a single RabbitMQ server for transport, you may encounter issues if each system contains streams of the same name. We recommend using a different RabbitMQ virtual host for each system. Update the spring.rabbitmq.virtual_host property in $XD_HOME/config/servers.yml to point XD at the correct virtual host.

Choosing an Analytics provider

By default, the xd-container will store Analytics data in redis. At the time of writing, this is the only supported option (when running in distributed mode). Use the --analytics option to specify another backing store for Analytics data.

xd/bin>$ ./xd-container --analytics redis

You can configure the following settings for redis analytics

xd:
  analytics:
    redis:
         backOffInitialInterval:    1000  (1)
         backOffMaxInterval:        10000 (2)
         backOffMultiplier:         2.0   (3)
         maxAttempts:               3     (4)
  1. The time in milliseconds before retrying a failed redis operation

  2. The maximum time (ms) to wait between retries

  3. The back off multiplier (previous interval x multiplier = next interval)

  4. The maximum number of retry attempts

Other Options

There are additional configuration options available for these scripts:

To specify the location of the Spring XD install other than the default configured in the script

export XD_HOME=<Specific XD install directory>

To specify the http port of the XDAdmin server,

xd/bin>$ ./xd-admin --httpPort <httpPort>

The XDContainer nodes by default start up with server.port 0 (which means they will scan for an available HTTP port). You can disable the HTTP endpoints for the XDContainer by setting server.port=-1. Note that in this case HTTP source support will not work in a PaaS environment because typically it would require XD to bind to a specific port. Both the XDAdmin and XDContainer processes bind to server.port $PORT (i.e. an environment variable if one is available, as is typical in a PaaS).

Using Hadoop

Spring XD supports the following Hadoop distributions:

  • hadoop27 - Apache Hadoop 2.7.1 (default)

  • phd21 - Pivotal HD 2.1 and 2.0

  • phd30 - Pivotal HD 3.0

  • cdh5 - Cloudera CDH 5.3.0

  • hdp22 - Hortonworks Data Platform 2.2

To specify the distribution libraries to use for Hadoop client connections, use the option
--hadoopDistro for the xd-container and xd-shell commands:

xd/bin>$ ./xd-shell --hadoopDistro <distribution>
xd/bin>$ ./xd-admin
xd/bin>$ ./xd-container --hadoopDistro <distribution>

Pass in the --help option to see other configuration properties.

XD-Shell in Distributed Mode

If you wish to use a XD-Shell that is on a different machine than where you deployed your admin server.

1) Open your shell

shell/bin>$ ./xd-shell

2) From the xd shell use the "admin config server" command i.e.

admin config server <yourhost>:9393

Running on YARN

Introduction

The Spring XD distributed runtime (DIRT) supports distribution of processing tasks across multiple nodes. See Running Distributed Mode for information on running Spring XD in distributed mode. One option is to run these nodes on a Hadoop YARN cluster rather than on VMs or physical servers managed by you.

What do you need?

To begin with, you need to have access to a Hadoop cluster running a version based on Apache Hadoop version 2. This includes Apache Hadoop 2.7.1, Pivotal HD 2.1 or 3.0, Hortonworks HDP 2.3 and Cloudera CDH5.

Important

Running YARN on some Ubuntu distributions and Mac OS X has shown to have issues when YARN applications are killed. It seems that the kill command doesn’t always succesfully kill the corresponding OS process and you end up with application processes still running. See HADOOP-9752 for more details.

You need a supported transport, see Running Distributed Mode for installation of Redis or Rabbit MQ. Spring XD on YARN currently uses Redis as the default data transport.

You also need Zookeeper running. If your Hadoop cluster doesn’t have Zookeeper installed you need to install and run it specifically for Spring XD. See the Setting up ZooKeeper section of the "Running Distributed Mode" chapter.

Lastly, you need an RDBMs to support batch jobs and JDBC operations.

Download Spring XD on YARN binaries

In addition to the regular spring-xd-<version>-dist.zip files we also distribute a zip file that includes all you need to deploy on YARN. The name of this zip file is spring-xd-<version>-yarn.zip. You can download the zip file for the current release from Spring release repo or a milestone build from the Spring milestone repo. Unzip the downloaded file and you should see a spring-xd-<version>-yarn directory.

Configure your deployment

Configuration options are contained in a config/servers.yml file in the Spring XD YARN install directory. You need to configure the hadoop settings, the transport choice plus redis/rabbit settings, the zookeeper settings and the JDBC datasource properties.

Depending on the distribution used you might need to change the siteYarnAppClasspath and siteMapreduceAppClasspath. We have provided basic settings for the supported distros, you just need to uncomment the ones for the distro you use.

These are the settings used for Hadoop 2.7.1:

spring:
    yarn:
        siteYarnAppClasspath: "$HADOOP_CONF_DIR,$HADOOP_COMMON_HOME/share/hadoop/common/*,$HADOOP_COMMON_HOME/share/hadoop/common/lib/*,$HADOOP_HDFS_HOME/share/hadoop/hdfs/*,$HADOOP_HDFS_HOME/share/hadoop/hdfs/lib/*,$HADOOP_YARN_HOME/share/hadoop/yarn/*,$HADOOP_YARN_HOME/share/hadoop/yarn/lib/*"
        siteMapreduceAppClasspath: "$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/*,$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/lib/*"
XD options

For Spring XD you need to define how many admin servers and containers you need using properties spring.xd.adminServers and spring.xd.containers respectively. You also need to define the HDFS location using property spring.yarn.applicationDir where the Spring XD binary and config files will be stored.

spring:
  xd:
    appmasterMemory: 512M
    adminServers: 1
    adminMemory: 512M
    adminLocality: false
    containers: 3
    containerMemory: 512M
    containerLocality: false
    container:
      groups: yarn
  yarn:
    applicationDir: /xd/app/

More about memory settings in above configuration, see section Configuring YARN memory reservations.

Hadoop settings

You need to specify the host where the YARN Resource Manager is running using spring.hadoop.resourceManagerHost as well as the HDFS URL using spring.hadoop.fsUri.

# Hadoop properties
spring:
  hadoop:
    fsUri: hdfs://localhost:8020
    resourceManagerHost: localhost
    config:
      topology.script.file.name: /path/to/topology-script.sh
Note

Setting hadoop topology.script.file.name property is mandatory if more sophisticated container placement is used to allocate XD admins or containers from a spesific hosts or racks. If this property is not set to match a one used in a hadoop cluster, allocations using hosts and racks will simply fail.

Zookeeper settings

You should specify the Zookeeper connection settings

#Zookeeper properties
#client connect string: host1:port1,host2:port2,...,hostN:portN
zk:
  client:
     connect: localhost:2181
Transport options

You should choose either redis (default) or rabbit as the transport and include the host and port in the properties for the choice you made.

# Transport used
transport: redis
# Redis properties
spring:
  redis:
   port: 6379
   host: localhost
JDBC datasource properties

You should specify the JDBC connection properties based on the RDBMs that you use for the batch jobs and JDBC sink

#Config for use with MySQL - uncomment and edit with relevant values for your environment
spring:
  datasource:
    url: jdbc:mysql://yourDBhost:3306/yourDB
    username: yourUsername
    password: yourPassword
    driverClassName: com.mysql.jdbc.Driver
XD Admin port

On default the property server.port which defines the used port for embedded server is set to 9393 but it can be overridden by changing the value in servers.yml.

#Port that admin-ui is listening on
#server:
#  port: 9393

On YARN it is recommended that you simply set the port to 0 meaning that server will automatically choose a random port. This is advisable simply because it will prevent port collission which are usually a little difficult to track down from a cluster. See more instructions in the section Connect xd-shell to YARN runtime managed admins for how to connect xd-shell to admins managed by YARN.

#Port that admin-ui is listening on
server:
  port: 0

Adding custom modules

The recommended approach for custom modules is to define the module registry location as a directory in HDFS. This will allow the most flexibility and the modules will automatically be available to all XD containers running in the Hadoop cluster. The xd.customModule.home property is by default set to the value ${spring.hadoop.fsUri}/xd/yarn/custom-modules for YARN deployments. This can be modified, but we recommend keeping it to a location on HDFS within the same Hadoop cluster.

xd:
  customModule:
    home: ${spring.hadoop.fsUri}/xd/yarn/custom-modules

See the Modules section for more on custom modules.

Customizing module configurations

The configurations for all standard XD modules can be customized by modifying the file modules.yml in the config directory and then adding it to the modules-config.zip archive in the same directory.

You can run the following command from the config directory to achieve this:

jar -uf modules-config.zip modules.yml

Modify container logging

Logging configuration for XD admins and containers are defined in files config/xd-admin-logger.properties and config/xd-container-logger.properties respectively. These two files are copied over to hdfs during the deployment. If you want to modify logging configuration either modify source files and do a deployment again or modify files in hdfs directly.

Control XD YARN application lifecycle

Change current directory to be the directory that was unzipped spring-xd-<version>-yarn. To read about runtime configuration and more sophisticated features see section Working with container groups.

Push the Spring XD application binaries and config to HDFS

Run the command

$ bin/xd-yarn push
New version installed
List installed application versions

Run the command

$ bin/xd-yarn pushed
  NAME  PATH
  ----  --------------------
  app   hdfs://node1:8020/xd
Submit the Spring XD YARN application

Run the command

$ bin/xd-yarn submit
New instance submitted with id application_1420911708637_0001
Check the status of YARN apps

You can use the regular yarn command to check the status. Simply run:

$ bin/xd-yarn submitted
  APPLICATION ID                  USER          NAME    QUEUE    TYPE STARTTIME       FINISHTIME  STATE    FINALSTATUS  ORIGINAL TRACKING URL
  ------------------------------  ------------  ------  ------- ----  --------------  ----------  -------  ----------- ---------------------------
  application_1420911708637_0001  jvalkealahti  xd-app  default XD    09/01/15 14:25  N/A         RUNNING  UNDEFINED http://172.16.101.106:49792

You should see one application running named xd-app.

Note

Pay attention to APPLICATION ID listed in output because that is an id used in most of the control commands to communicate to a specific application instance. For example you may have multiple XD YARN runtime instances running.

Kill application

Application can be killed using a kill command.

$ bin/xd-yarn kill -a application_1420905836797_0001
Kill request for application_1420905836797_0001 sent
Using a built-in shell

To get a better and faster command usage a build-in shell can be used to run control commands:

$ bin/xd-yarn shell
Spring YARN Cli (v2.1.0.M3)
Hit TAB to complete. Type 'help' and hit RETURN for help, and 'exit' to quit.
$
clear            clustercreate    clusterdestroy   clusterinfo      clustermodify
clustersinfo     clusterstart     clusterstop      exit             help
kill             prompt           pushed           submit           submitted
$

Connect xd-shell to YARN runtime managed admins

XD admins will register its runtime information into zookeeper and you can use the admininfo command to query this information:

$ bin/xd-yarn admininfo
Admins: [http://hadoop.localdomain:43740]

Then connect xd-shell to this instance:

server-unknown:>admin config server --uri http://hadoop.localdomain:43740
Successfully targeted http://hadoop.localdomain:43740

xd:>runtime containers
  Container Id                          Host               IP Address      PID    Groups  Custom Attributes
  ------------------------------------  -----------------  --------------  -----  ------  --------------------------------------------------
 6324a9ae-205b-44b9-b851-f0edd7245286  node2.localdomain   172.16.101.102  12284  yarn    {virtualCores=1, memory=512, managementPort=54694}

Configuring YARN memory reservations

YARN Nodemanager is continously tracking how much memory is used by individual YARN containers. If containers are using more memory than what the configuration allows, containers are simply killed by a Nodemanager. Application master controlling the app lifecycle is given a little more freedom meaning that Nodemanager is not that aggressive when making a desicion when a container should be killed.

Lets take a quick look of memory related settings in YARN cluster and in YARN applications. Below xml config is what a default vanilla Apache Hadoop uses for memory related settings. Other distributions may have different defaults.

yarn-site.xml
<configuration>

  <property>
    <name>yarn.nodemanager.pmem-check-enabled</name>
    <value>true</value>
  </property>

  <property>
    <name>yarn.nodemanager.vmem-check-enabled</name>
    <value>true</value>
  </property>

  <property>
    <name>yarn.nodemanager.vmem-pmem-ratio</name>
    <value>2.1</value>
  </property>

  <property>
    <name>yarn.scheduler.minimum-allocation-mb</name>
    <value>1024</value>
  </property>

  <property>
    <name>yarn.scheduler.maximum-allocation-mb</name>
    <value>8192</value>
  </property>

  <property>
    <name>yarn.nodemanager.resource.memory-mb</name>
    <value>8192</value>
  </property>

</configuration>
yarn.nodemanager.pmem-check-enabled

Enables a check for physical memory of a process. This check if enabled is directly tracking amount of memory requested for a YARN container.

yarn.nodemanager.vmem-check-enabled

Enables a check for virtual memory of a process. This setting is one which is usually causing containers of a custom YARN applications to get killed by a node manager. Usually the actual ratio between physical and virtual memory is higher than a default 2.1 or bugs in a OS is causing wrong calculation of a used virtual memory.

yarn.nodemanager.vmem-pmem-ratio

Defines a ratio of allowed virtual memory compared to physical memory. This ratio simply defines how much virtual memory a process can use but the actual tracked size is always calculated from a physical memory limit.

yarn.scheduler.minimum-allocation-mb

Defines a minimum allocated memory for container.

Note

This setting also indirectly defines what is the actual physical memory limit requested during a container allocation. Actual physical memory limit is always going to be multiple of this setting rounded to upper bound. For example if this setting is left to default 1024 and container is requested with 512M, 1024M is going to be used. However if requested size is 1100M, actual size is set to 2048M.

yarn.scheduler.maximum-allocation-mb

Defines a maximum allocated memory for container.

yarn.nodemanager.resource.memory-mb

Defines how much memory a node controlled by a node manager is allowed to allocate. This setting should be set to amount of which OS is able give to YARN managed processes in a way which doesn’t cause OS to swap, etc.

Tip

If testing XD YARN runtime on a single computer with a multiple VM based hadoop cluster a pro tip is to set both yarn.nodemanager.pmem-check-enabled and yarn.nodemanager.vmem-check-enabled to false, set yarn.scheduler.minimum-allocation-mb much lower to either 256 or 512 and yarn.nodemanager.resource.memory-mb 15%-20% below a defined VM memory.

We have three memory settings for components participating XD YARN runtime. You can use configuration properties spring.xd.appmasterMemory, spring.xd.adminMemory and spring.xd.containerMemory respectively.

spring:
  xd:
    appmasterMemory: 512M
    adminMemory: 512M
    containerMemory: 512M

Working with container groups

Container grouping and clustering is more sophisticated feature which allows better control of XD admins and containers at runtime. Basic features are:

  • Control members in a groups.

  • Control lifecycle state for group as whole.

  • Create groups dynamically.

  • Re-start failed containers.

XD YARN Runtime has a few built-in groups to get you started. There are two groups admin and container created by default which both are lauching exactly one container chosen randomly from YARN cluster.

List existing groups

Run the command:

$ bin/xd-yarn clustersinfo -a application_1420911708637_0001
  CLUSTER ID
  ----------
  container
  admin
Get status of a group

Run the command:

bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c admin
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        1

Or to get verbose output:

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c admin -v
  CLUSTER STATE  MEMBER COUNT  ANY PROJECTION  HOSTS PROJECTION  RACKS PROJECTION  ANY SATISFY  HOSTS SATISFY  RACKS SATISFY
  -------------  ------------  --------------  ----------------  ----------------  -----------  -------------  -------------
  RUNNING        1             1               {}                {}                0            {}             {}
Control group state

Run the commands to stop group, list its status, start group and finally list status again:

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        1

$ bin/xd-yarn clusterstop -a application_1420911708637_0001 -c container
Cluster container stopped.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  STOPPED        0

$ bin/xd-yarn clusterstart -a application_1420911708637_0001 -c container
Cluster container started.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        1
Modify group configuration

In these commans we first ramp up container count and then ramp it down:

18:19 $ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        1

$ bin/xd-yarn clustermodify -a application_1420911708637_0001 -c container -w 3
Cluster container modified.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        3

$ bin/xd-yarn clustermodify -a application_1420911708637_0001 -c container -w 2
Cluster container modified.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        2
Note

In above example we used option -w which is a shortcut for defining YARN allocation which uses a wildcard requests allowing containers to be requested from any host.

Create a new group

When you want to create a new group that is because you need to add new XD admin or container nodes to a current system with a different settings. These setting usually differ by a colocation of containers. More about built-in group configuration refer to section Built-in group configurations.

Run the command:

$ bin/xd-yarn clustercreate -a application_1420911708637_0001 -c custom -i container-nolocality-template -p default -w 2
Cluster custom created.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c custom
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  INITIAL        0

$ bin/xd-yarn clusterstart -a application_1420911708637_0001 -c custom
Cluster custom started.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c custom
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        2

To create group with two containers on node5 and one on node6 run command:

$ bin/xd-yarn clustercreate -a application_1420911708637_0001 -c custom -i container-locality-template -p default -y "{hosts:{node6: 1,node5: 2}}"
Cluster custom created.

$ bin/xd-yarn -a application_1420911708637_0001 -c custom -v
  CLUSTER STATE  MEMBER COUNT  ANY PROJECTION  HOSTS PROJECTION    RACKS PROJECTION  ANY SATISFY  HOSTS SATISFY       RACKS SATISFY
  -------------  ------------  --------------  ------------------  ----------------  -----------  ------------------  -------------
  INITIAL        0             0               {node5=2, node6=1}  {}                0            {node5=2, node6=1}  {}
Destroy a group

Run the commands:

$ bin/xd-yarn clustersinfo -a application_1420911708637_0001
  CLUSTER ID
  ----------
  container
  admin

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  RUNNING        1

$ bin/xd-yarn clusterstop -a application_1420911708637_0001 -c container
Cluster container stopped.

$ bin/xd-yarn clusterinfo -a application_1420911708637_0001 -c container
  CLUSTER STATE  MEMBER COUNT
  -------------  ------------
  STOPPED        0

$ bin/xd-yarn clusterdestroy -a application_1420911708637_0001 -c container
Cluster container destroyed.

$ bin/xd-yarn clustersinfo -a application_1420911708637_0001
  CLUSTER ID
  ----------
  admin
Note

Group can only destroyed if its status is STOPPED or INITIAL.

Built-in group configurations

Few groups are already defined where admin and container are enabled automatically. Other groups are disabled and thus working as a blueprints which can be used to create groups manually.

admin

Default group definition for XD admins.

container

Default group definition for XD containers.

admin-nolocality-template

Blueprint with relax localization. Use this to create a groups if you plan to use any matching.

admin-locality-template

Blueprint with no relax localization. Use this to create a groups if you plan to use hosts or racks matching.

container-nolocality-template

Blueprint with relax localization. Use this to create a groups if you plan to use any matching.

container-locality-template

Blueprint with no relax localization. Use this to create a groups if you plan to use hosts or racks matching.

Configuration examples

This section contains examples of usual use cases for custom configurations.

Run containers on a specific hosts

Below configuration sets default XD container to exist on node1 and node2.

xd:
  containerLocality: true
spring:
  yarn:
    appmaster:
      containercluster:
        clusters:
          container:
            projection:
              data: {any: 0, hosts: {node1: 1, node2: 1}}
Run admins on a specific racks

Below configuration sets default XD admins to exist on /rack1 and /rack2.

xd:
  adminLocality: true
spring:
  yarn:
    appmaster:
      containercluster:
        clusters:
          admin:
            projection:
              data: {any: 0, racks: {/rack1: 1, /rack2: 1}}
Disable default admin and container groups

Existing built-in groups admin and container can be disabled by setting their projection types to null.

spring:
  yarn:
    appmaster:
      containercluster:
        clusters:
          admin:
            projection:
              type: null
          container:
            projection:
              type: null

xd-yarn command synopsis

push
xd-yarn push - Push new application version

usage: xd-yarn push [options]

Option                     Description
------                     -----------
-v, --application-version  Application version (default: app)
pushed
xd-yarn pushed - List pushed applications

usage: xd-yarn pushed [options]

No options specified
submit
xd-yarn submit - Submit application

usage: xd-yarn submit [options]

Option                     Description
------                     -----------
-v, --application-version  Application version (default: app)
submitted
xd-yarn submitted - List submitted applications

usage: xd-yarn submitted [options]

Option                   Description
------                   -----------
-t, --application-type   Application type (default: XD)
-v, --verbose [Boolean]  Verbose output (default: true)
kill
xd-yarn kill - Kill application

usage: xd-yarn kill [options]

Option                Description
------                -----------
-a, --application-id  Specify YARN application id
clustersinfo
xd-yarn clustersinfo - List clusters

usage: xd-yarn clustersinfo [options]

Option                Description
------                -----------
-a, --application-id  Specify YARN application id
clusterinfo
xd-yarn clusterinfo - List cluster info

usage: xd-yarn clusterinfo [options]

Option                   Description
------                   -----------
-a, --application-id     Specify YARN application id
-c, --cluster-id         Specify cluster id
-v, --verbose [Boolean]  Verbose output (default: true)
clustercreate
xd-yarn clustercreate - Create cluster

usage: xd-yarn clustercreate [options]

Option                  Description
------                  -----------
-a, --application-id    Specify YARN application id
-c, --cluster-id        Specify cluster id
-g, --container-groups  Container groups
-h, --projection-hosts  Projection hosts counts
-i, --cluster-def       Specify cluster def id
-p, --projection-type   Projection type
-r, --projection-racks  Projection racks counts
-w, --projection-any    Projection any count
-y, --projection-data   Raw projection data
clusterdestroy
xd-yarn clusterdestroy - Destroy cluster

usage: xd-yarn clusterdestroy [options]

Option                Description
------                -----------
-a, --application-id  Specify YARN application id
-c, --cluster-id      Specify cluster id
clustermodify
xd-yarn clustermodify - Modify cluster

usage: xd-yarn clustermodify [options]

Option                  Description
------                  -----------
-a, --application-id    Specify YARN application id
-c, --cluster-id        Specify cluster id
-h, --projection-hosts  Projection hosts counts
-r, --projection-racks  Projection racks counts
-w, --projection-any    Projection any count
-y, --projection-data   Raw projection data
clusterstart
xd-yarn clusterstart - Start cluster

usage: xd-yarn clusterstart [options]

Option                Description
------                -----------
-a, --application-id  Specify YARN application id
-c, --cluster-id      Specify cluster id
clusterstop
xd-yarn clusterstop - Stop cluster

usage: xd-yarn clusterstop [options]

Option                Description
------                -----------
-a, --application-id  Specify YARN application id
-c, --cluster-id      Specify cluster id

Introduction to YARN resource allocation

This section describes some background of how YARN resource allocation works, what are the limitations of it and more importantly how it reflects into XD YARN runtime.

Note

More detailed info of resource allocation can be found from a Spring for Apache Hadoop reference documentation.

YARN as having a strong roots from original MapReduce framework is imposing relatively strange concepts of where containers are about to be executed. In a MapReduce world every map and reduce tasks are executed in its own container where colocation is usually determined by a physical location of a HDFS file block map or reduce tasks are accessing. This is introducing a concepts of allocating containers on any hosts, specific hosts or specific racks. Usually YARN is trying to place container as close as possible to a physical location to minimize network IO so i.e. if host cannot be chosen, rack is chosen instead assuming a whole rack is connected together with a fast switch.

For custom YARN applications like XD YARN runtime this doesn’t necessarily make that much sense because we’re not hard-tied to HDFS file blocks. What makes sense is that we can still place containers on different racks to get better high availability in case whole rack goes down or if specific containers needs to exist on specific hosts to access either custom physical or network resources. Good example of having a need to execute something on a specific host is either a disk access or outbound internet access if cluster is highly secured.

One other YARN resource allocation concept worth mentioning is relaxation of container locality. This simply means that if resources are requested from hosts or racks, YARN will relax those requests if resources cannot be allocated immediately. Turning relax flag off guarantees that containers will be allocated from hosts or racks. Though these requests will then wait forever if allocation cannot be done.

Application Configuration

Introduction

There are two main parts of Spring XD that can be configured, servers and modules.

The servers (xd-singlenode, xd-admin, xd-container) are Spring Boot applications and are configured as described in the Spring Boot Reference documentation. In the most simple case this means editing values in the YAML based configuration file servers.yml. The values in this configuration file will overwrite the values in the default application.yml file that is embedded in the XD jar.

Note
The use of YAML is an alternative to using property files. YAML is a superset of JSON, and as such is a very convenient format for specifying hierarchical configuration data.

For modules, each module has its own configuration file located in its own directory, for example source/http/http.properties. Shared configuration values for modules can be placed in a common modules.yml file.

For both server and module configuration, you can have environment specific settings through the use of application profiles and the ability to override values in files by setting OS environment variables.

In this section we will walk though how to configure servers and modules.

Server Configuration

The startup scripts for xd-singlenode, xd-admin, and xd-container will by default look for the file $XD_HOME\config\servers.yml as a source of externalized configuration information.

The location and name of this resourse can be changed by using the environment variables XD_CONFIG_LOCATION and XD_CONFIG_NAME. The start up script takes the value of these environment variables to set the Spring Boot properties spring.config.location and spring.config.name. Note, that for XD_CONFIG_LOCATION you can reference any Spring Resource implementation, most commonly denoted using the prefixes classpath:, file: and http:.

It is common to keep your server configuration separate form the installation directory of XD itself. To do this, here is an example environment variable setting

export XD_CONFIG_LOCATION=file:/xd/config/
export XD_CONFIG_NAME=region1-servers

Note: the file path separator ("/") at the end of XD_CONFIG_LOCATION is necessary.

Profile support

Profiles provide a way to segregate parts of your application configuration and change their availability and/or values based on the environment. This lets you have different configuration settings for qa and prod environments and to easily switch between them.

To activate a profile, set the OS environment variable SPRING_PROFILES_ACTIVE to a comma delimited list of profile names. The server looks to load profile specific variants of the servers.yml file based on the naming convention servers-{profile}.yml. For example, if SPRING_PROFILES_ACTIVE=prod the following files would be searched for in the following order.

  1. XD_CONFIG_LOCATION/servers-prod.yml

  2. XD_CONFIG_LOCATION/servers.yml

You may also put multiple profile specific configuration in a single servers.yml file by using the key spring.profiles in different sections of the configuration file. See Multi-profile YAML documents for more information.

Database Configuration

Spring XD saves the state of the batch job workflows in a relational database. When running xd-singlenode an embedded HSQLDB database instance is started automatically. When running in distributed mode a standalone HSQLDB instance can be used. A startup script hsqldb-server is provided in the installation directory under the folder hsqldb/bin. It is recommended to use HSQLDB only for development and learning.

When deploying in a production environment, you will need to select another database. Spring XD is primarily tested on HSQLDB, MySql and Postgres but Apache Derby and Oracle are supported as well. All batch workflow tables are automatically created, if they do not exist, when you use HSQLDB, MySQL, Postgres, Apache Derby or Oracle. The JDBC driver jars for HSQLDB and Postgres are already on the XD classpath. If you use MySQL, Apache Derby or Oracle for your batch repository database, then you would need to copy the corresponding JDBC jar into the lib directory under $XD_HOME ($XD_HOME/lib) before starting Spring XD.

Note
If you access any database other than HSQLDB or Postgres in a stream module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.

The provided configuration file servers.yml located in $XD_HOME/config has commented out configuration for some commonly used databases. You can use these as a basis to support your database environment. XD also utilizes the Tomcat jdbc connection pool and these settings can be configured in the servers.yml.

Note
Until full schema support is added for Sybase and other databases, you will need to put a .jar file in the xd/lib directory that contains the equivalent functionality as these DDL scripts.
Note
There was a schema change in version 1.0 RC1. Use or adapt the the sample migration class to update your schema.
HSQLDB

When in distributed mode and you want to use HSQLDB, you need to change the value of spring.datasource properties. As an example,

hsql:
 server:
   host: localhost
   port: 9102
   dbname: xdjob
spring:
   datasource:
   url: jdbc:hsqldb:hsql://${hsql.server.host:localhost}:${hsql.server.port:9101}/${hsql.server.dbname:xdjob};sql.enforce_strict_size=true;hsqldb.tx=mvcc
   username: sa
   password:
   driverClassName: org.hsqldb.jdbc.JDBCDriver

The properties under hsql.server are substituted in the spring.datasource.url property value. This lets you create short variants of existing Spring Boot properties. Using this style, you can override the value of these configuration variables by setting an OS environment variable, such as xd_server_host. Alternatively, you can not use any placeholders and set spring.datasource.url directly to known values.

MySQL

When in distributed mode and you want to use MySQL, you need to change the value of spring.datasource.* properties. As an example,

spring:
 datasource:
 url: jdbc:mysql://yourDBhost:3306/yourDB
 username: yourUsername
 password: yourPassword
 driverClassName: com.mysql.jdbc.Driver

To override these settings set an OS environment variable such as spring_datasource_url to the value you require.

PostgreSQL

When in distributed mode and you want to use PostgreSQL, you need to change the value of spring.datasource.* properties. As an example,

spring:
  datasource:
    url: jdbc:postgresql://yourDBhost:5432/yourDB
    username: yourUsername
    password: yourPassword
    driverClassName: org.postgresql.Driver

To override these settings set an OS environment variable such as spring_datasource_url to the value you require.

Oracle database

When in distributed mode and you want to use Oracle database, you need to change the value of spring.datasource.* properties. As an example,

spring:
  datasource:
    url: jdbc:oracle:thin:@//yourDBhost:1521/yourDB
    username: scott
    password: tiger
    driverClassName: oracle.jdbc.driver.OracleDriver
    validationQuery: select 1 from dual

To override these settings set an OS environment variable such as spring_datasource_url to the value you require.

Redis

If you want to use Redis for analytics or data transport you should set the host and port of the Redis server.

spring:
  redis:
   port: 6379
   host: localhost
   pool:
     maxIdle: 8 # max idle connections in the pool
     minIdle: 0 # min idle connections in the pool
     maxActive: -1 # no limit to the number of active connections
     maxWait: 30000 # time limit to get a connection - only applies if maxActive is finite

To override these settings set an OS environment variable such as spring_redis_port to the value you require.

You can also configure redis to use Sentinel.

spring:
  redis:
   port: 6379
   host: host1
   pool:
     maxIdle: 8 # max idle connections in the pool
     minIdle: 0 # min idle connections in the pool
     maxActive: -1 # no limit to the number of active connections
     maxWait: 30000 # time limit to get a connection - only applies if maxActive is finite
   sentinel:
     master: mymaster
     nodes: host2:26379,host3:26380,host4:26381

In addition, the following default settings for the redis message bus can be modified in servers.yml…​

    redis:
      headers:				 (1)
      default:
        backOffInitialInterval:    1000  (2)
        backOffMaxInterval:        10000 (3)
        backOffMultiplier:         2.0   (4)
        concurrency:               1     (5)
        maxAttempts:               32    (6)
  1. comma-delimited list of additional (string-valued) header names to transport

  2. The time in milliseconds before retrying a failed message delivery

  3. The maximum time (ms) to wait between retries

  4. The back off multiplier (previous interval x multiplier = next interval)

  5. The minimum number of consumer threads receiving messages for a module

  6. The maximum number of delivery attempts

RabbitMQ

If you want to use RabbitMQ as a data transport use the following configuration settings

spring:
  rabbitmq:
   addresses: localhost:5672 (1)
   adminAddresses: http://localhost:15672 (2)
   nodes: [email protected] (3)
   username: guest         (4)
   password: guest         (5)
   virtual_host: /         (6)
   useSSL: false           (7)
   sslProperties:          (8)
   ssl:                    (9)
     keyStore:
     keyStorePassphrase:
     trustStore:
     trustStorePassphrase:
   channelCacheSize: 100   (10)
  1. A comma-separated list of RabbitMQ server addresses (a single entry when not clustering).

  2. A comma-separated list of RabbitMQ management plugin URLs - only used when nodes contains more than one entry. Entries in this list must correspond to the corresponding entry in addresses.

  3. A comma-separated list of RabbitMQ node names; when more than one entry, used to locate the server address where a queue is located. Entries in this list must correspond to the corresponding entry in addresses.

  4. The user name.

  5. The password.

  6. The virtual host.

  7. True to use SSL for the AMQP protocol.

  8. The location of the SSL properties file, when certificate exchange is used.

  9. Discrete SSL configuration properties as an alternative to a properties file.

  10. The channel cache size, this should be set to a large enough value to avoid opening/closing channels at a high rate. Examine the channel creation via the RabbitMQ Admin UI to determine a suitable value. Defaults to 100.

To override these settings set an OS environment variable such as spring_rabbitmq_host to the value you require.

See Message Bus regarding SSL configuration.

When configuring a clustered environment, with High Availability Queues, it is possible to configure the bus so that it consumes from the node where the queue is located. This is facilitated by the LocalizedQueueConnectionFactory which determines the node for a queue. To enable this feature, add the list of nodes to the spring.rabbitmq.nodes property. These nodes correspond to the broker addresses in the corresponding place in the spring.rabbitmq.addresses property. The size of these lists must be identical (when the nodes property has more than one entry). The spring.rabbitmq.adminAddresses property contains the corresponding URLs for the admins on those same nodes. Again, the property list must be the same length.

In addition, the following default settings for the rabbit message bus can be modified in servers.yml…​

  messagebus:
    rabbit:
      compressionLevel:            1     (1)
      longStringLimit:             8192  (2)
      default:
        ackMode:                   AUTO  (3)
        autoBindDLQ:               false (4)
        backOffInitialInterval:    1000  (5)
        backOffMaxInterval:        10000 (6)
        backOffMultiplier:         2.0   (7)
        batchBufferLimit:          10000 (8)
        batchingEnabled:           false (9)
        batchSize:                 100   (10)
        batchTimeout:              5000  (11)
        compress:                  false (12)
        concurrency:               1     (13)
        durableSubscription:       false (14)
        maxAttempts:               3     (15)
        maxConcurrency:            1     (16)
        prefix:                    xdbus. (17)
        prefetch:                  1     (18)
        replyHeaderPatterns:       STANDARD_REPLY_HEADERS,*   (19)
        republishToDLQ:            false (20)
        requestHeaderPatterns:     STANDARD_REQUEST_HEADERS,* (21)
        requeue:                   true  (22)
        transacted:                false (23)
        txSize:                    1     (24)
  1. When the bus (or a stream module deployment) is configured to compress messages, specifies the compression level. See java.uti.zip.Deflater for available values; defaults to 1 (BEST_SPEED)

  2. RabbitMQ headers longer than this value are not converted to String; instead they are made available as a DataInputStream; these are currently not properly re-converted during output conversion. If you expect headers longer than this, increase this setting appropriately if you wish them to pass to downstream modules.

  3. AUTO (container acks), NONE (broker acks), MANUAL (consumer acks). Upper case only. Note: MANUAL requires specialized code in the consuming module and is unlikely to be used in an XD application. For more information, see http://docs.spring.io/spring-integration/reference/html/amqp.html#amqp-inbound-ack

  4. When true, the bus will automatically declare dead letter queues and binding for each bus queue. The user is responsible for setting a policy on the broker to enable dead-lettering; see Message Bus Configuration for more information. The bus will configure a dead-letter-exchange (<prefix>DLX) and bind a queue with the name <original queue name>.dlq and route using the original queue name

  5. The time in milliseconds before retrying a failed message delivery

  6. The maximum time (ms) to wait between retries

  7. The back off multiplier (previous interval x multiplier = next interval)

  8. When batching is enabled, the size of the buffer that will cause a batch to be released (overrides batchSize)

  9. True to enable message batching by producers

  10. The number of messages in a batch (may be preempted by batchBufferLimit or batchTimeout)

  11. The idle time to wait before sending a partial batch

  12. True to enable message compression - also see (1. bus compressionLevel)

  13. The minimum number of consumer threads receiving messages for a module

  14. When true queues for subscriptions to publish/subscribe named channels (tap:, topic:) will be declared as durable and are eligible for dead-letter configuration according to the autoBindDLQ setting.

  15. The maximum number of delivery attempts. Setting this to 1 disables the retry mechanism and requeue must be set to false if you wish failed messages to be rejected or routed to a DLQ. Otherwise deliveries will be attempted repeatedly, with no termination. Also see republishToDLQ

  16. The maximum number of consumer threads receiving messages for a module

  17. A prefix applied to all queues, exchanges so that policies (HA etc) can be applied

  18. The number of messages to prefetch for each consumer

  19. Determines which reply headers will be transported

  20. By default, failed messages after retries are exhausted are rejected. If a dead-letter queue (DLQ) is configured, rabbitmq will route the failed message (unchanged) to the DLQ. Setting this property to true instructs the bus to republish failed messages to the DLQ, with additional headers, including the exception message and stack trace from the cause of the final failure. Note that the republish will occur even if maxAttempts is only set to 1. Also see autoBindDLQ

  21. Determines which request headers will be transported

  22. Whether rejected messages will be requeued by default

  23. Whether the channel is to be transacted

  24. The number of messages to process between acks (when ack mode is AUTO).

Kafka

If you want to use Kafka as a data transport, the following connection settings, as well as defaults for the kafka message bus can be modified in servers.yml. Starting with release 1.2, Spring XD only supports Kafka 0.8.2 or higher.

Note
To ensure the proper functioning of the Kafka Message Bus, you must eanble log cleaning in your Kafka configuration. This is set using the configuration variable log.cleaner.enable=true. See the Kafka documentation for additional configuration options for log cleaning.
Note
At this time, the Kafka message bus does not support job processing.
Note
The Kafka message bus does not support count=0 for module deployments, and therefore, it does not support direct binding of modules. This feature will be available in a future release. In the meantime, if direct communication between modules is necessary for Kafka deployments, composite modules should be used instead.
  messagebus:
    kafka:
      # connection properties
      brokers:                                localhost:9092  (1)
      zkAddress:                              localhost:2181  (2)
      socketBufferSize:                       2097152         (3)
      # operating mode
      mode:                                   embeddedHeaders (4)
      offsetManagement:                       kafkaTopic      (5)
      headers:
         # comma-delimited list of additional header names to transport (6)
      # offset topic settings
      offsetStoreTopic:                       SpringXdOffsets (7)
      offsetStoreSegmentSize:                 25000000        (8)
      offsetStoreRetentionTime:               60000           (9)
      offsetStoreRequiredAcks:                1               (10)
      offsetStoreMaxFetchSize:                1048576         (11)
      offsetStoreBatchBytes:                  16384           (12)
      offsetStoreBatchTime:                   1000            (13)
      offsetUpdateTimeWindow:                 10000           (14)
      offsetUpdateCount:                      0               (15)
      offsetUpdateShutdownTimeout:            2000            (16)
      # defaults for the bus
      default:
        batchSize:                 16384                      (17)
        batchTimeout:              0                          (18)
        replicationFactor:         1                          (19)
        concurrency:               1                          (20)
        requiredAcks:              1                          (21)
        compressionCodec:          none                       (22)
        queueSize:                 8192                       (23)
        maxWait:                   100                        (24)
        fetchSize:                 1048576                    (25)
        minPartitionCount:         1                          (26)
        syncProducer:              false                      (27)
        syncProducerTimeout:       5000                       (28)
  1. A list of Kafka broker addresses, for sending messages

  2. A list of ZooKeeper addresses, for receiving messages

  3. The size of the socket buffer, for the consumer

  4. How the bus handles headers and serialization: embeddedHeaders supports Spring Integration header embedding and bus-managed serialization based on embedded content type headers, whereas raw mode will operate only with byte array data, will not embed headers and will leave the handling of serialization to the user.

  5. Where the bus stores offsets: kafkaTopic is similar to pre-1.3 behaviour of using a dedicated Kafka topic, and kafkaNative relies on the native topic-based Kafka offset storage support compatible with Kafka 0.8.2 and later.

  6. A list of custom headers to be transported by the bus.

  7. The name of the topic where the Kafka Message Bus will store offsets (must be a compacted topic - Spring XD will attempt to create a compacted topic by default).

  8. The segment size for the offset topic if offsetManagement is kafkaTopic.

  9. The retention time for the offset topic if offsetManagement is kafkaTopic.

  10. The number of required acks for the offset topic if offsetManagement is kafkaTopic.

  11. The maximum fetch size when reading from the offset topic if offsetManagement is kafkaTopic.

  12. The batch size (in bytes) for the producers writing to the offset topic if offsetManagement is kafkaTopic.

  13. Upper bound for the offset topic producer delay for batching

  14. The frequency (in milliseconds) with which offsets are saved (mutually exclusive with offsetUpdateCount)

  15. The frequency (in message counts) with which offsets are saved (mutually exclusive with offsetUpdateTimeWindow)

  16. The timeout for shutting down offset management and ensuring that the latest offset updates have been pushed.

  17. The amount of data (in bytes) that the producer will try to buffer before sending data to brokers.

  18. Timeout (in milliseconds) for batching data on the producer side. A value of zero (default) means that data will be sent out immediately as available.

  19. The replication factor of the topics created by the message bus. At least as many brokers must be in the cluster when the topic is being created.

  20. The maximum number of consumer threads receiving messages for a module. The total number of threads actively consuming partitions across all the instances of a specific module cannot be larger than the partition count of a transport topic - therefore, if such a situation occurs, some modules instances will, in fact, use less consumer threads.

  21. The number of required acks when producing messages, i.e. how many brokers have committed data to the logs and acknowledged this to the leader. Special values are -1, meaning all in-sync replicas, and 0 indicating that no acks are necessary.

  22. Enables compression for the bus and sets the compression codec.

  23. The maximum size of the internal message queue (in messages), per consumer processing thread. It must be a power of 2.

  24. The maximum amount of time that the consumers will wait to fetch data from a broker (if less than fetchSize is available)

  25. The maximum amount of data that the consumers will try to fetch, per broker, in one polling cycle.

  26. The minimum number of partitions that will be used by a bus topic.

  27. When true, a synchronous producer is used

  28. If <27> is true, set the timeout to wait for Kafka delivery (in ms). Is ⇐ 0, wait forever.

Admin Server HTTP Port

The default HTTP port of the xd-admin server is 9393. To change the value use the following configuration setting

server:
  port: 9876

Management Port

The XD servers provide general health and JMX exported management endpoints via Jolokia.

By default the management and health endpoints are available on port 9393. To change the value of the port use the following configuration setting to servers.yml.

management:
  port: 9876

You can also disable http management endpoints by setting the port value to -1.

By default JMX MBeans are exported. You can disable JMX by setting spring.jmx.enabled=false.

The section on Monitoring and management over HTTP provides details on how to configure these endpoint.

Admin Server Security

By default, the Spring XD admin server is unsecured and runs on an unencrypted HTTP connection. You can secure your administration REST endpoints, as well as the Admin UI by enabling HTTPS and requiring clients to authenticate.

Enabling HTTPS

By default, the administration, management, and health endpoints, as well as the Admin UI use HTTP as a transport. You can switch to HTTPS easily, by adding a certificate to your configuration in servers.yml

spring:
  profiles: admin                                    (1)
server:
  ssl:
    key-alias: yourKeyAlias                          (2)
    key-store: path/to/keystore                      (3)
    key-store-password: yourKeyStorePassword         (4)
    key-password: yourKeyPassword                    (5)
    trust-store: path/to/trust-store                 (6)
    trust-store-password: yourTrustStorePassword     (7)
  1. The settings are applicable only to the admin server (regardless whether it’s started in single-node mode or as a separate instance).

  2. The alias (or name) under which the key is stored in the keystore.

  3. The path to the keystore file. Classpath resources may also be specified, by using the classpath prefix: classpath:path/to/keystore

  4. The password of the keystore.

  5. The password of the key.

  6. The path to the truststore file. Classpath resources may also be specified, by using the classpath prefix: classpath:path/to/trust-store

  7. The password of the trust store.

Note
If HTTPS is enabled, it will completely replace HTTP as the protocol over which the REST endpoints and the Admin UI interact. Plain HTTP requests will fail - therefore, make sure that you configure your Shell accordingly.
Enabling authentication

By default, the REST endpoints (administration, management and health), as well as the Admin UI do not require authenticated access. By turning on authentication on the admin server:

  • the REST endpoints will require Basic authentication for access;

  • the Admin UI will be accessible after signing in through a web form.

Note
When authentication is set up, it is strongly recommended to enable HTTPS as well, especially in production environments.

You can turn on authentication by adding the following to the configuration in servers.yml:

spring:
  profiles: admin                                                     (1)
security:
  basic:
    enabled: true                                                     (2)
    realm: SpringXD                                                   (3)
  user:
    name: yourAdminUsername
    password: yourAdminPassword
    role: ADMIN, VIEW, CREATE                                         (4)
  1. The settings are applicable only to the admin server (regardless whether it’s started in single node mode or as a separate instance).

  2. Must be set to true for security to be enabled.

  3. (Optional) The realm for Basic authentication. Will default to SpringXD if not explicitly set.

  4. Must set with appropriate roles (ADMIN, VIEW and CREATE) to enable. Note: the prefix ROLE_ isn’t required here.

Additionally, you must specify an authentication method, out of the following that Spring XD supports:

  • single user mode (the default made available by Spring Boot)

  • integration with an existing LDAP server

  • file based configuration

The options above are mutually exclusive, and they are described below.

Single user authentication

This option uses a single username/password pair is created for the server. This option is turned on by default, if security is enabled and LDAP is not configured.

You can configure this option by adding the following to the configuration in servers.yml, once security is enabled.

spring:
  profiles: admin
security:
  basic:
    enabled: true
    realm: SpringXD
  user:
    name: yourAdminUsername                                           (1)
    password: yourAdminPassword                                       (2)
  1. The username for authentication (must be used by REST clients and in the Admin UI). Will default to user if not explicitly set.

  2. The password for authentication (must be used by REST clients and in the Admin UI). If not explicitly set, it will be auto-generated, as described in the Spring Boot documentation.

LDAP authentication

Spring XD also supports authentication against an LDAP server, in both direct bind and "search and bind" modes. When the LDAP authentication option is activated, the default single user mode is turned off.

In direct bind mode, a pattern is defined for the user’s distinguished name (DN), using a placeholder for the username. The authentication process derive the distinguished name of the user by replacing the placeholder and use it to authenticate a user against the LDAP server, along with the supplied password. You can set up LDAP direct bind as follows:

spring:
  profiles: admin
security:
  basic:
    enabled: true
    realm: SpringXD
xd:
  security:
    authentication:
      ldap:
        enabled: true                                        (1)
        url: ldap://ldap.example.com:3309                    (2)
        userDnPattern: uid={0},ou=people,dc=example,dc=com   (3)
  1. Enables LDAP integration

  2. The URL for the LDAP server

  3. The distinguished name (DN) pattern for authenticating against the server.

The "search and bind" mode involves connecting to an LDAP server, either anonymously or with a fixed account, and searching for the distinguished name of the authenticating user based on its username, and then using the resulting value and the supplied password for binding to the LDAP server. This option is configured as follows:

spring:
  profiles: admin
security:
  basic:
    enabled: true
    realm: SpringXD
xd:
  security:
    authentication:
      ldap:
        enabled: true                                         (1)
        url: ldap://ldap.example.com:3309                     (2)
        managerDn: uid=bob,ou=managers,dc=example,dc=com      (3)
        managerPassword: managerPassword                      (4)
        userSearchBase: ou=otherpeople,dc=example,dc=com      (5)
        userSearchFilter: uid={0}                             (6)
  1. Enables LDAP integration

  2. The URL of the LDAP server

  3. A DN for to authenticate to the LDAP server, if anonymous searches are not supported (optional, required together with next option)

  4. A password to authenticate to the LDAP server, if anonymous searches are not supported (optional, required together with previous option)

  5. The base for searching the DN of the authenticating user (serves to restrict the scope of the search)

  6. The search filter for the DN of the authenticating user

File based authentication

Spring XD supports listing users in a configuration file, as described below. Each user must be assigned a password and one or more roles:

spring:
  profiles: admin
security:
  basic:
    enabled: true
    realm: SpringXD
xd:
  security:
    authentication:
      file:
        enabled: true                    (1)
        users:                           (2)
          bob: bobspassword, ROLE_VIEW   (3)
          alice: alicepwd, ROLE_ADMIN
  1. Enables file based integration

  2. This is a yaml map of username to (password and roles)

  3. Each map "value" is made of a password and one or more roles, comma separated

Customizing authorization

All of the above deals with authentication, i.e. how to assess the identity of the user. Irrespective of the option chosen, you can also customize authorization i.e. who can do what.

The default scheme uses three roles to protect the REST endpoints that Spring XD exposes:

  • ROLE_VIEW for anything that relates to retrieving state

  • ROLE_CREATE for anything that involves creating, deleting or mutating the state of the system

  • ROLE_ADMIN for boot management endpoints.

All of those defaults are written out in application.yml, which you can choose to override via servers.yml. This takes the form of a YAML list (as some rules may have precedence over others) and so you’ll need to copy/paste the whole list and tailor it to your needs (as there is no way to merge lists). Always refer to your version of application.yml, as the snippet reproduced below may be outdated. The default rules are as such:

  security:
    authorization:
      rules:
        # Streams
        - GET    /streams/definitions            => hasRole('ROLE_VIEW')
        - DELETE /streams/definitions            => hasRole('ROLE_CREATE')
        - GET    /streams/definitions/*          => hasRole('ROLE_VIEW')
        - POST   /streams/definitions            => hasRole('ROLE_CREATE')
        - DELETE /streams/definitions/*          => hasRole('ROLE_CREATE')
        # Stream Deployments
        - GET    /streams/deployments/           => hasRole('ROLE_VIEW')
        - DELETE /streams/deployments/           => hasRole('ROLE_CREATE')
        - GET    /streams/deployments/*          => hasRole('ROLE_VIEW')
        - POST   /streams/deployments/*          => hasRole('ROLE_CREATE')
        - DELETE /streams/deployments/*          => hasRole('ROLE_CREATE')
        # Job Definitions
        - GET    /jobs/definitions               => hasRole('ROLE_VIEW')
        - DELETE /jobs/definitions               => hasRole('ROLE_CREATE')
        - GET    /jobs/definitions/*             => hasRole('ROLE_VIEW')
        - POST   /jobs/definitions               => hasRole('ROLE_CREATE')
        - DELETE /jobs/definitions/*             => hasRole('ROLE_CREATE')
        # Job Deployments
        - GET    /jobs/deployments/              => hasRole('ROLE_VIEW')
        - DELETE /jobs/deployments/              => hasRole('ROLE_CREATE')
        - GET    /jobs/deployments/*             => hasRole('ROLE_VIEW')
        - POST   /jobs/deployments/*             => hasRole('ROLE_CREATE')
        - DELETE /jobs/deployments/*             => hasRole('ROLE_CREATE')
        # Batch Job Configurations
        - GET    /jobs/configurations            => hasRole('ROLE_VIEW')
        - GET    /jobs/configurations/*          => hasRole('ROLE_VIEW')
        # Batch Job Executions
        - GET    /jobs/executions                => hasRole('ROLE_VIEW')
        - PUT    /jobs/executions?stop=true      => hasRole('ROLE_CREATE')
        - GET    /jobs/executions?jobname=*      => hasRole('ROLE_VIEW')
        - POST   /jobs/executions?jobname=*      => hasRole('ROLE_CREATE')
        - GET    /jobs/executions/*              => hasRole('ROLE_VIEW')
        - PUT    /jobs/executions/*?restart=true => hasRole('ROLE_CREATE')
        - PUT    /jobs/executions/*?stop=true    => hasRole('ROLE_CREATE')
        - GET    /jobs/executions/*/steps        => hasRole('ROLE_VIEW')
        - GET    /jobs/executions/*/steps/*      => hasRole('ROLE_VIEW')
        - GET    /jobs/executions/*/steps/*/progress => hasRole('ROLE_VIEW')
        # Batch Job Instances
        - GET    /jobs/instances?jobname=*       => hasRole('ROLE_VIEW')
        - GET    /jobs/instances/*               => hasRole('ROLE_VIEW')
        # Module Definitions
        - GET    /modules                        => hasRole('ROLE_VIEW')
        - POST   /modules                        => hasRole('ROLE_CREATE')
        - GET    /modules/*/*                    => hasRole('ROLE_VIEW')
        - DELETE /modules/*/*                    => hasRole('ROLE_CREATE')
        # Deployed Modules
        - GET    /runtime/modules                => hasRole('ROLE_VIEW')
        # Containers
        - GET    /runtime/containers             => hasRole('ROLE_VIEW')
        # Counters
        - GET    /metrics/counters               => hasRole('ROLE_VIEW')
        - GET    /metrics/counters/*             => hasRole('ROLE_VIEW')
        - DELETE /metrics/counters/*             => hasRole('ROLE_CREATE')
        # Field Value Counters
        - GET    /metrics/field-value-counters   => hasRole('ROLE_VIEW')
        - GET    /metrics/field-value-counters/* => hasRole('ROLE_VIEW')
        - DELETE /metrics/field-value-counters/* => hasRole('ROLE_CREATE')
        # Aggregate Counters
        - GET    /metrics/aggregate-counters     => hasRole('ROLE_VIEW')
        - GET    /metrics/aggregate-counters/*   => hasRole('ROLE_VIEW')
        - DELETE /metrics/aggregate-counters/*   => hasRole('ROLE_CREATE')
        # Gauges
        - GET    /metrics/gauges                 => hasRole('ROLE_VIEW')
        - GET    /metrics/gauges/*               => hasRole('ROLE_VIEW')
        - DELETE /metrics/gauges/*               => hasRole('ROLE_CREATE')
        # Rich Gauges
        - GET    /metrics/rich-gauges            => hasRole('ROLE_VIEW')
        - GET    /metrics/rich-gauges/*          => hasRole('ROLE_VIEW')
        - DELETE /metrics/rich-gauges/*          => hasRole('ROLE_CREATE')
        # Tab Completions
        - GET    /completions/stream?start=*     => hasRole('ROLE_VIEW')
        - GET    /completions/job?start=*        => hasRole('ROLE_VIEW')
        - GET    /completions/module?start=*     => hasRole('ROLE_VIEW')
        # Boot Endpoints
        - GET    /management/**                  => hasRole('ROLE_ADMIN')

The format of each line is the following:

HTTP_METHOD URL_PATTERN '=>' SECURITY_ATTRIBUTE

where

Be mindful that the above is indeed a YAML list, not a map (thus the use of - dashes at the start of each line) that lives under the security.authorization.rules key.

Cross-origin resource sharing (CORS)

Cross-origin resource sharing (CORS) is a mechanism that allows restricted resources (e.g. fonts) on a web page to be requested from another domain outside the domain from which the resource originated.

— Wikipedia
https://en.wikipedia.org/wiki/Cross-origin_resource_sharing

We do set a default value of http://localhost:9889 in the internal application.yml file that is embedded inside the Spring XD jars.

application.yml
xd:
  …
  ui:
    …
    allow_origin: "http://localhost:9889"
    …

In order to customize this, set the xd.ui.allow_origin property in your server.yml file for the admin server profile by adding the following section:

server.yml
---
spring:
  profiles: admin
xd:
  ui:
    allow_origin: "*"
---

For example, if you set the value to "*" (asterisk), Spring XD should accept requests from any domain. Please make sure to wrap the asterisk with double quotes.

Under the hood the value will set the CORS Access-Control-Allow-Origin header in the AccessControlInterceptor via the RestConfiguration class.

Local transport

Local transport uses a QueueChannel to pass data between modules. There are a few properties you can configure on the QueueChannel

  • xd.local.transport.named.queueSize - The capacity of the queue, the default value is Integer.MAX_VALUE

  • xd.local.transport.named.polling - Messages that are buffered in a QueueChannel need to be polled to be consumed. This property controls the fixed rate at which polling occurs. The default value is 1000 ms.

Serialization

Serialization is used by remote transport. Please see the section on Optimizing Serialization for a detailed discussion of configuration options.

Module Configuration

Modules are configured by placing property files in a nested directory structure based on their type and name. The root of the nested directory structure is by default XD_HOME/config/modules. This location can be customized by setting the OS environment variable XD_MODULE_CONFIG_LOCATION, similar to how the environment variable XD_CONFIG_LOCATION is used for configuring the server. If XD_MODULE_CONFIG_LOCATION is set explicitly, then it is necessary to add the file path separator ("/") at the end of the path.

Note
If XD_MODULE_CONFIG_LOCATION is set to use explicit location, make sure to copy entire directory structure from the default module config location xd/config/modules into the new module config location. The XD_MODULE_CONFIG_LOCATION can reference any Spring Resource implementation, most commonly denoted using the prefixes classpath:, file: and http:.

As an example, if you wanted to configure the twittersearch module, you would create a file

XD_MODULE_CONFIG_LOCATION\source\twittersearch\twittersearch.properties

and the contents of that file would be property names such as consumerKey and consumerSecret.

Note
You do not need to prefix these property names with a source.twittersearch prefix.

You can override the values in the module property file in various ways. The following sources of properties are considered in the following order.

  1. Properties specified in the stream or job DSL definition

  2. Java System Properties (e.g. source.http.port=9454)

  3. OS environment variables. (e.g. source_http_port=9454)

  4. XD_MODULE_CONFIG_LOCATION\<type>\<name>\<name>.properties (including profile variants)

  5. Default values specified in module metadata (if available).

Values in XD_MODULE_CONFIG_LOCATION\<type>\<name>\<name>.properties can be property placeholder references to keys defined in another resource location. By default the resource is the file XD_MODULE_CONFIG_LOCATION\modules.yml. You can customize the name of the resource by using setting the OS environment variable XD_MODULE_CONFIG_NAME before running a server startup script.

The modules.yml file can be used to specify the values of keys that should be shared across different modules. For example, it is common to use the same twitter developer credentials in both the twittersearch and twitterstream modules. To avoid repeating the same credentials in two property files, you can use the following setup.

modules.yml contains

sharedConsumerKey: alsdjfqwopieur
sharedConsumerSecret: pqwieouralsdjkqwpo
sharedAccessToken: llixzchvpiawued
sharedAccessTokenSecret: ewoqirudhdsldke

and XD_MODULE_CONFIG_LOCATION\source\twitterstream\twitterstream.properties contains

consumerKey=${sharedConsumerKey}
consumerSecret=${sharedConsumerSecret}
accessToken=${sharedAccessToken}
accessTokenSecret=${sharedAccessTokenSecret}

and XD_MODULE_CONFIG_LOCATION\source\twittersearch\twittersearch.properties contains

consumerKey=${sharedConsumerKey}
consumerSecret=${sharedConsumerSecret}

Profiles

When resolving property file names, the server will look to load profile specific variants based on the naming convention <name>-{profile}.properties. For example, if given the OS environment variable spring_profiles_active=default,qa the following configuration file names for the twittersearch module would be searched in this order

  1. XD_MODULE_CONFIG_LOCATION\source\twittersearch\twittersearch.properties

  2. XD_MODULE_CONFIG_LOCATION\source\twittersearch\twittersearch-default.properties

  3. XD_MODULE_CONFIG_LOCATION\source\twittersearch\twittersearch-qa.properties

Also, the shared module configuration file is refernced using profile variants, so given the OS environment variable spring_profiles_active=default,qa the following shared module configuration files would be searched for in this order

  1. XD_MODULE_CONFIG_LOCATION\modules.yml

  2. XD_MODULE_CONFIG_LOCATION\modules-default.yml

  3. XD_MODULE_CONFIG_LOCATION\modules-qa.yml

Batch Jobs or modules accessing JDBC

Another common case is access to a relational database from a job or the JDBC Sink module.

As an example, to provide the properties for the batch job jdbchdfs the file XD_MODULE_CONFIG_LOCATION\job\jdbchdfs\jdbchdfs.properties should contain

driverClass=org.hsqldb.jdbc.JDBCDriver
url=jdbc:hsqldb:mem:xd
username=sa
password=

A property file with the same keys, but likely different values would be located in XD_MODULE_CONFIG_LOCATION\sink\jdbc\jdbc.properties.

Encrypted Properties

If you wish encrypt passwords and other secret values stored in application configuration file, you must provide a Spring bean that implements TextEncryptor and implement the decrypt method. The bean, annotated with @Component, or a @Configuration class providing the bean definition, must be present under the base package spring.xd.ext.encryption.

This introduces a compile time dependency on spring-security which offers some standard implementations (see Spring Security Crypto Module). There is also a compatible RSA implementation.

Package the implementation of your choice in a jar file and install it in xd/lib along with any required dependencies (spring-security-core is already in the runtime classpath).

Spring XD will use the TextEncryptor you provide to decrypt any properties that contain the prefix {cipher}. This conforms to the convention used by the Spring Cloud Config Server.

For example:

spring:
  datasource:
    username: dbuser
    password: '{cipher}FKSAJDFGYOS8F7GLHAKERGFHLSAJ'

If a TextEncrypor bean is present, Spring XD will detect the encrypted password FKSAJDFGYOS8F7GLHAKERGFHLSAJ and decrypt it prior to creating any beans that reference this value.

Note
Single quotes, as shown above, are required in yaml files. Do not include them in plain properties files.

Also this will work with module options included in the stream definition, if you are so inclined:

xd:> stream create --name test --definition "jdbc --username=dbuser --password='{cipher}FKSAJDFGYOS8F7GLHAKERGFHLSAJ'..."

DSL Guide

Introduction

Spring XD provides a DSL for defining a stream. Over time the DSL is likely to evolve significantly as it gains the ability to define more and more sophisticated streams as well as the steps of a batch job.

Pipes and filters

A simple linear stream consists of a sequence of modules. Typically an Input Source, (optional) Processing Steps, and an Output Sink. As a simple example consider the collection of data from an HTTP Source writing to a File Sink. Using the DSL the stream description is:

http | file

A stream that involves some processing:

http | filter | transform | file

The modules in a stream definition are connected together using the pipe symbol |.

Module parameters

Each module may take parameters. The parameters supported by a module are defined by the module implementation. As an example the http source module exposes port setting which allows the data ingestion port to be changed from the default value.

http --port=1337

It is only necessary to quote parameter values if they contain spaces or the | character. Here the transform processor module is being passed a SpEL expression that will be applied to any data it encounters:

transform --expression='new StringBuilder(payload).reverse()'

If the parameter value needs to embed a single quote, use two single quotes:

// Query is: Select * from /Customers where name='Smith'
scan --query='Select * from /Customers where name=''Smith'''

Named channels

Instead of a source or sink it is possible to use a named channel. Normally the modules in a stream are connected by anonymous internal channels (represented by the pipes), but by using explicitly named channels it becomes possible to construct more sophisticated flows. In keeping with the unix theme, sourcing/sinking data from/to a particular channel uses the > character. A named channel is specified by using a channel type, followed by a : followed by a name. The channel types available are:

queue - this type of channel has point-to-point (p2p) semantics
topic - this type of channel has pub/sub semantics

Here is an example that shows how you can use a named channel to share a data pipeline driven by different input sources.

queue:foo > file
http > queue:foo
time > queue:foo

Now if you post data to the http source, you will see that data intermingled with the time value in the file.

The opposite case, the fanout of a message to multiple streams, is planned for a future release. However, taps are a specialization of named channels that do allow publishing data to multiple sinks. For example:

tap:stream:mystream > file
tap:stream:mystream > log

Once data is received on mystream, it will be written to both file and log.

Support for routing messages to different streams based on message content is also planned for a future release.

Labels

Labels provide a means to alias or group modules. Labels are simply a name followed by a : When used as an alias a label can provide a more descriptive name for a particular configuration of a module and possibly something easier to refer to in other streams.

mystream = http | obfuscator: transform --expression=payload.replaceAll('password','*') | file

Labels are especially useful (and required) for disambiguating when multiple modules of the same name are used:

mystream = http | uppercaser: transform --expression=payload.toUpperCase() | exclaimer: transform --expression=payload+'!' | file

Refer to this section of the Taps chapter to see how labels facilitate the creation of taps in these cases where a stream contains ambiguous modules.

Single quotes, Double quotes, Escaping

Spring XD is a complex runtime that involves a lot of systems when you look at the complete picture. There is a Spring Shell based client that talks to the admin that is responsible for parsing. In turn, modules may themselves rely on embedded languages (like the Spring Expression Language) to accomplish their behavior.

Those three components (shell, XD parser and SpEL) have rules about how they handle quotes and how syntax escaping works, and when stacked with each other, confusion may arise. This section explains the rules that apply and provides examples to the most common situations.

Note
It’s not always that complicated

This section focuses on the most complicated cases, when all 3 layers are involved. Of course, if you don’t use the XD shell (for example if you’re using the REST API directly) or if module option values are not SpEL expressions, then escaping rules can be much simpler

Spring Shell

Arguably, the most complex component when it comes to quotes is the shell. The rules can be laid out quite simply, though:

  • a shell command is made of keys (--foo) and corresponding values. There is a special, key-less mapping though, see below

  • a value can not normally contain spaces, as space is the default delimiter for commands

  • spaces can be added though, by surrounding the value with quotes (either single ['] or double ["] quotes)

  • if surrounded with quotes, a value can embed a literal quote of the same kind by prefixing it with a backslash (\)

  • Other escapes are available, such as \t, \n, \r, \f and unicode escapes of the form \uxxxx

  • Lastly, the key-less mapping is handled in a special way in the sense that if does not need quoting to contain spaces

For example, the XD shell supports the ! command to execute native shell commands. The ! accepts a single, key-less argument. This is why the following works:

xd:>! rm foo

The argument here is the whole rm foo string, which is passed as is to the underlying shell.

As another example, the following commands are strictly equivalent, and the argument value is foo (without the quotes):

xd:>stream destroy foo
xd:>stream destroy --name foo
xd:>stream destroy "foo"
xd:>stream destroy --name "foo"

XD Syntax

At the XD parser level (that is, inside the body of a stream or job definition) the rules are the following:

  • option values are normally parsed until the first space character

  • they can be made of literal strings though, surrounded by single or double quotes

  • To embed such a quote, use two consecutive quotes of the desired kind

As such, the values of the --expression option to the filter module are semantically equivalent in the following examples:

filter --expression=payload>5
filter --expression="payload>5"
filter --expression='payload>5'
filter --expression='payload > 5'

Arguably, the last one is more readable. It is made possible thanks to the surrounding quotes. The actual expression is payload > 5 (without quotes).

Now, let’s imagine we want to test against string messages. If we’d like to compare the payload to the SpEL literal string, "foo", this is how we could do:

filter --expression=payload=='foo'           (1)
filter --expression='payload == ''foo'''     (2)
filter --expression='payload == "foo"'       (3)
  1. This works because there are no spaces. Not very legible though

  2. This uses single quotes to protect the whole argument, hence actual single quotes need to be doubled

  3. But SpEL recognizes String literals with either single or double quotes, so this last method is arguably the best

Please note that the examples above are to be considered outside of the Spring XD shell. When entered inside the shell, chances are that the whole stream definition will itself be inside double quotes, which would need escaping. The whole example then becomes:

xd:>stream create foo --definition "http | filter --expression=payload='foo' | log"
xd:>stream create foo --definition "htpp | filter --expression='payload == ''foo''' | log"
xd:>stream create foo --definition "http | filter --expression='payload == \"foo\"' | log"

SpEL syntax and SpEL literals

The last piece of the puzzle is about SpEL expressions. Many modules accept options that are to be interpreted as SpEL expressions, and as seen above, String literals are handled in a special way there too. Basically,

  • literals can be enclosed in either single or double quotes

  • quotes need to be doubled to embed a literal quote. Single quotes inside double quotes need no special treatment, and vice versa

As a last example, assume you want to use the transform module. That module accepts an expression option which is a SpEL expression. It is to be evaluated against the incoming message, with a default of payload (which forwards the message payload untouched).

It is important to understand that the following are equivalent:

transform --expression=payload
transform --expression='payload'

but very different from the following:

transform --expression="'payload'"
transform --expression='''payload'''

and other variations.

The first series will simply evaluate to the message payload, while the latter examples will evaluate to the actual literal string payload (again, without quotes).

Putting it all together

As a last, complete example, let’s review how one could force the transformation of all messages to the string literal hello world, by creating a stream in the context of the XD shell:

stream create foo --definition "http | transform --expression='''hello world''' | log" (1)
stream create foo --definition "http | transform --expression='\"hello world\"' | log" (2)
stream create foo --definition "http | transform --expression=\"'hello world'\" | log" (2)
  1. This uses single quotes around the string (at the XD parser level), but they need to be doubled because we’re inside a string literal (very first single quote after the equals sign)

  2. use single and double quotes respectively to encompass the whole string at the XD parser level. Hence, the other kind of quote can be used inside the string. The whole thing is inside the --definition argument to the shell though, which uses double quotes. So double quotes are escaped (at the shell level) == Interactive Shell

Introduction

Spring XD includes an interactive shell that you can use create, deploy, destroy and query streams and jobs. There are also commands to help with common tasks such as interacting with HDFS, the UNIX shell, and sending HTTP requests. In this section we will introduce the main commands and features of the shell.

Using the Shell

When you start the shell you can type help to show all the commands that are available. Note, that since the XD shell is based on Spring Shell you can contribute you own commands into the shell. The general groups of commands are related to the management of

  • Modules

  • Streams

  • Jobs

  • Analytics (Counters, Aggregate Counters, Gauges, etc.)

  • HDFS

For example to see what modules are available issue the command

xd:>module list
Tip

The list of all Spring XD specific commands can be found in the Shell Reference

The shell also provides extensive command completion capabilities. For example, if you type mod and hit TAB, you will be presented with all the matching commands.

xd:>module
module compose    module delete     module display    module info
module list
Note
Tab completion works for module options as well as for the DSL used within the --definition option for stream and module commands.

The command module list shows all the modules available

xd:>module list
      Source              Processor           Sink                     Job
  ------------------  ------------------  -----------------------  ----------------
      gemfire-cq          aggregator          counter                  hdfsjdbc
      post                http-client         log                      jdbchdfs
      twitterstream       splitter            field-value-counter      hdfsmongodb
      http                filter              rich-gauge               filejdbc
      reactor-syslog      json-to-tuple       mqtt                     ftphdfs
      reactor-ip          transform           file                     filepollhdfs
      jms                 bridge              splunk
      tcp-client          object-to-json      mail
      mqtt                script              tcp
      file                                    hdfs
      twittersearch                           gauge
      gemfire                                 jdbc
      mail                                    gemfire-server
      trigger                                 throughput-sampler
      tcp                                     gemfire-json-server
      tail                                    router
      syslog-tcp                              aggregate-counter
      syslog-udp                              rabbit
      rabbit                                  hdfs-dataset
      time

Suppose we want to create a stream that uses the http source and file sink. How do we know what options are available to use? There are two ways to find out. The first is to use the command module info. Pressing TAB after typiing moudle info will complete the command with the --name option and then present all the modules prefixed by their type.

xd:>module info --name
job:filejdbc                job:filepollhdfs            job:ftphdfs
job:hdfsjdbc                job:hdfsmongodb             job:jdbchdfs
processor:aggregator        processor:bridge            processor:filter
processor:http-client       processor:json-to-tuple     processor:object-to-json
processor:script            processor:splitter          processor:transform
sink:aggregate-counter      sink:counter                sink:field-value-counter
sink:file                   sink:gauge                  sink:gemfire-json-server
sink:gemfire-server         sink:hdfs                   sink:hdfs-dataset
sink:jdbc                   sink:log                    sink:mail
sink:mqtt                   sink:rabbit                 sink:rich-gauge
sink:router                 sink:splunk                 sink:tcp
sink:throughput-sampler     source:file                 source:gemfire
source:gemfire-cq           source:http                 source:jms
source:mail                 source:mqtt                 source:post
source:rabbit               source:reactor-ip           source:reactor-syslog
source:syslog-tcp           source:syslog-udp           source:tail
source:tcp                  source:tcp-client           source:time
source:trigger              source:twittersearch        source:twitterstream
xd:>module info --name

The module info command for the http source shows the option names, a brief description, and default values.

xd:>module info --name source:http
Information about source module 'http':

  Option Name  Description                                       Default  Type
  -----------  ------------------------------------------------  -------  --------
  port         the port to listen to                             9000     int
  outputType   how this module should emit messages it produces  <none>   MimeType

For the file sink the options are

xd:>module info --name sink:file
Information about sink module 'file':

  Option Name  Description                                                        Default            Type
  -----------  -----------------------------------------------------------------  -----------------  --------
  binary       if false, will append a newline character at the end of each line  false              boolean
  charset      the charset to use when writing a String payload                   UTF-8              String
  dir          the directory in which files will be created                       /tmp/xd/output/    String
  mode         what to do if the file already exists                              APPEND             Mode
  name         filename pattern to use                                            ${xd.stream.name}  String
  suffix       filename extension to use                                          <none>             String
  inputType    how this module should interpret messages it consumes              <none>             MimeType

Note that the default value ${xd.stream.name} will be resolved to the name of the stream that contains the module.

Tab completion for Job and Stream DSL definitions

When creating a stream defintion tab completion after --definition will enable you to see all the options that are available for a given module as well as a list of candidate modules for the subsequent module in the stream. For example, hitting TAB after http as shown below

xd:>stream create --name test --definition "http
http --outputType=           http --port=                 http | aggregate-counter     http | aggregator
http | bridge                http | counter               http | field-value-counter   http | file
http | filter                http | gauge                 http | gemfire-json-server   http | gemfire-server
http | hdfs                  http | hdfs-dataset          http | http-client           http | jdbc
http | json-to-tuple         http | log                   http | mail                  http | mqtt
http | object-to-json        http | rabbit                http | rich-gauge            http | router
http | script                http | splitter              http | splunk                http | tcp
http | throughput-sampler    http | transform

shows the options outputType and port in addition to any processors and sinks. Hitting TAB after entering -- after the http module will provide a list of only the http options

xd:>stream create --name test --definition "http --
http --outputType=   http --port=

Entering the port number and also the pipel | symbol and hitting tab will show completions for candidate processor and sink modules. The same process of tab completion for module options applies to each module in the chain.

Executing a script

You can execute a script by either passing in the --cmdfile argument when starting the shell or by executing the script command inside the shell. When using scripts it is common to add comments using either // or ; characters at the start of the line for one line comments or use /* and */ for multiline comments

Single quotes, Double quotes, Escaping

There are often three layers of parsing when passing entering commands to the shell. The shell parses the command to recognize -- options, inside the body of a stream/job definition the values are parsed until the first space character, and inside some command options SpEL is used (e.g. router). Understanding the interaction between these layers can cause some confusion. The DSL Guide section on quotes and escaping will help you if you run into any issues.

Admin UI

Introduction

Spring XD provides a browser-based GUI which currently has 4 sections:

  • Containers Provides the XD cluster view with the list of all running containers

  • Streams Deploy/undeploy Stream Definitions

  • Jobs Perform Batch Job related tasks

  • Analytics Create data visualizations for the various analytics modules

Upon starting Spring XD, the Admin UI is available at:

"http://<adminHost>:<adminPort>/admin-ui"

If you have enabled https, then it will be located at https://localhost:9393/admin-ui

If you have enabled security, a login form is available at http://localhost:9393/admin-ui/#/login

Note: Default admin server port is 9393

The Spring XD Admin UI
Figure 1. The Spring XD Admin UI

Containers

The Containers section of the admin UI shows the containers that are in the XD cluster. For each container the group properties and deployed modules are shown. More information on the container (hostname, pid, ip address) and for the module (module options and deployment properties) is available by clicking on the respective links. You can also shutdown a container (in distributed mode) by clicking on the shutdown button. You will be asked for confirmation if you select to shutdown.

List of Containers
Figure 2. List of Containers

Streams

The Streams section of the admin UI provides the Definitions tab that provides a listing of Stream definitions. There you have the option to deploy or undeploy those streams. Additionally you can remove the definition by clicking on destroy.

List of Stream Definitions
Figure 3. List of Stream Definitions

Jobs

The Jobs section of the admin UI currently has four tabs specific for Batch Jobs

  • Modules

  • Definitions

  • Deployments

  • Executions

Modules

Modules encapsulate a unit of work into a reusable component. Within the XD runtime environment Modules allow users to create definitions for Streams as well as Batch Jobs. Consequently, the Modules tab within the Jobs section allows users to create Batch Job definitions. In order to learn more about Modules, please see the chapter on Modules.

List available batch job modules

This page lists the available batch job modules.

List Job Modules
Figure 4. List Job Modules

On this screen you can perform the following actions:

View Details

View details such as the job module options.

Create Definition

Create a Job Definition from the respective Module.

Create a Job Definition from a selected Job Module

On this screen you can create a new Job Definition. As a minimum you must provide a name for the new definition. Optionally you can select wether the new definition shall be automatically deployed. Depending on the selected module, you will also have the option to specify various parameters that are used during the deployment of the definition.

Create a Job Definition
Figure 5. Create a Job Definition
View Job Module Details
View Job Module Details
Figure 6. View Job Module Details

On this page you can view the details of a selected job module. The pages lists the available options (properties) of the modules.

List job definitions

This page lists the XD batch job definitions and provides actions to deploy, un-deploy or destroy those jobs.

List Job Definitions
Figure 7. List Job Definitions

List job deployments

This page lists all the deployed jobs and provides option to launch or schedule the deployed job.

List Job Deployments
Figure 8. List Job Deployments
Launching a batch Job

Once the job is deployed, they can be launched through the Admin UI as well. Navigate to the Deployments tab. Select the job you want to launch and press Launch. The following modal dialog should appear:

Launch a Batch Job with parameters
Figure 9. Launch a Batch Job with parameters

Using this screen, you can define one or more job parameters. Job parameters can be typed and the following data types are available:

  • String (The default)

  • Date (The default date format is: yyyy/MM/dd)

  • Long

  • Double

Schedule Batch Job Execution
Schedule a Batch Job
Figure 10. Schedule a Batch Job

When clicking on Schedule, you have the option to run the job:

  • using a fixed delay interval (specified in seconds)

  • on a specific data/time

  • using a valid CRON expression

Job Deployment Details

On this screen, you can view additional deployment details. Besides viewing the stream definition, the available Module Metadata is shown as well, e.g. on which Container the definition has been deployed to.

Job Deployment Details
Figure 11. Job Deployment Details

List job executions

This page lists the Batch Job Executions and provides the option to restart or stop a specific job execution, provided the operation is available. Furthermore, you have the option to view the Job execution details.

List Job Executions
Figure 12. List Job Executions

The list of Job Executions also shows the state of the underlying Job Definition. The following states can be shown.

Create Definition

The underlying Job Definition was undeployed.

Create Definition

The underlying Job Definition was deleted.

View Details

This is a composed Job Definition.

As an example, we have created the following composed job using the Spring XD shell:

spring> job create composed1 --definition "myjob" --deploy
spring> job create composed2 --definition "myjob" --deploy
spring> job create composedDemoJob --definition "composed1 || composed2" --deploy
spring> job launch composedDemoJob

Once executed, you should see the composedDemoJob on the Job Execution page:

Composed Job Execution
Figure 13. Job Executions with Composed Job Execution

By clicking on the magnifying glass you will be able to see the composed job’s details, including the associated child Job Executions for the jobs composed1 and composed2.

Job execution details
Composed Job Execution Details
Figure 14. Job Execution Details of Composed Job

The list of Job Executions on the Job execution details page will only be visible for composed jobs. In the case of a non-composed Job, you would see the following:

Job Execution Details - Steps
Figure 15. Job Execution Details - Steps

The Job Execution Details screen also contains a list of the executed steps. You can further drill into the Step Execution Details by clicking onto the magnifying glass.

Step execution details

On the top of the page, you will see progress indicator the respective step, with the option to refresh the indicator. Furthermore, a link is provided to view the step execution history.

Step Execution Details
Figure 16. Step Execution Details

The Step Execution details screen provides a complete list of all Step Execution Context key/value pairs. For example, the Spring for Apache Hadoop steps provides exhaustive detail information.

Step Execution Context
Figure 17. Step Execution Context

This includes a link back to the Job History UI of the Hadoop Cluster.

Job History UI
Figure 18. Job History UI
Important
In case of exceptions, the Exit Description field will contain additional error information. Please be aware, though, that this field can only have a maximum of 2500 characters. Therefore, in case of long exception stacktraces, trimming of error messages may occur. In that case, please refer to the server log files for further details.
Step execution history
Step Execution History
Figure 19. Step Execution History

On this screen, you can view various metrics associated with the selected step such as duration, read counts, write counts etc.

Analytics

The Analytics section of the admin UI provides dashboarding and data visualization capabilities for the various analytics modules available in Spring XD:

  • Counters

  • Aggregate Counters

  • Field-Value Counters

  • Gauges

  • Rich Gauges

For example, if you have created the springtweets stream and the corresponding counter in the Counter chapter, you can now easily create the corresponding graph from within the Dashboard tab:

  1. Under Metric Type, select Counters from the select box

  2. Under Stream, select tweetcount

  3. Under Visualization, select the desired chart option, Bar Chart

You will see a visualization similar to the following:

Analytics Bar Chart
Figure 20. Counter Bar Chart

Under Visualization, select the desired chart option, Graph Chart

You will see a visualization similar to the following:

Analytics Counter Graph
Figure 21. Counter Graph

Using the icons to the right, you can add additional charts to the dashboard, re-arange the order of created dashboards or remove data visualizations.

The remaining 3 tabs Counters, Gauges and Rich-Gauges, provide default visualizations for all Spring XD counters within the system. For example, if you have created the Simple Tap Example (Rich Gauge) in the Counter Chapter, you will see a visualization similar to the following under the Rich-Gauges tab:

Analytics Rich Gauge
Figure 22. Rich Gauge

Architecture

Introduction

Spring XD is a unified, distributed, and extensible service for data ingestion, real time analytics, batch processing, and data export. The foundations of XD’s architecture are based on the over 100+ man years of work that have gone into the Spring Batch, Integration and Data projects. Building upon these projects, Spring XD provides servers and a configuration DSL that you can immediately use to start processing data.  You do not need to build an application yourself from a collection of jars to start using Spring XD.

Spring XD has two modes of operation - single and multi-node. The first is a single process that is responsible for all processing and administration. This mode helps you get started easily and simplifies the development and testing of your application. The second is a distributed mode, where processing tasks can be spread across a cluster of machines and an administrative server reacts to user commands and runtime events managed within a shared runtime state to coordinate processing tasks executing on the cluster.

Runtime Architecture

The key components in Spring XD are the XD Admin and XD Container Servers. Using a high-level DSL, you post the description of the required processing tasks to the Admin server over HTTP. The Admin server then maps the processing tasks into processing modules. A module is a unit of execution and is implemented as a Spring ApplicationContext. A distributed runtime is provided that will assign modules to execute across multiple XD Container servers. A single XD Container server can run multiple modules. When using the single node runtime, all modules are run in a single XD Container and the XD Admin server is run in the same process.

DIRT Runtime

A distributed runtime, called Distributed Integration Runtime, aka DIRT, will distribute the processing tasks across multiple XD Container instances. The XD Admin server breaks up a processing task into individual module definitions and assigns each module to a container instance using ZooKeeper (see XD Distributed Runtime). Each container listens for module definitions to which it has been assigned and deploys the module, creating a Spring ApplicationContext to run it.

Modules share data by passing messages using a configured messaging middleware (Rabbit, Redis, or Local for single node). To reduce the number of hops across messaging middleware between them, multiple modules may be composed into larger deployment units that act as a single module. To learn more about that feature, refer to the Composing Modules section.

distributed node
Figure 23. The XD Admin Server sending module definitions to each XD Container

How the processing task is broken down into modules is discussed in the section Container Server Architecture.

Support for other distributed runtimes

In the 1.0 release, You can run Spring XD natively, in which case you are responsible for starting up the XD Admin and XD Container instances. Alternately you can run Spring XD on Hadoop’s YARN, see Running XD on YARN. Pivotal Cloud Foundry support is planned for a future release. If you are feeling a adventurous, you can also take a look at our scripts for deploying Spring XD to EC2. These are used as part of our system integration tests.

Single Node Runtime

A single node runtime is provided that runs the Admin and Container servers, ZooKeeper, and HSQLDB in the same process. the single node runtime is primarily intended for testing and development purposes but it may also appropriate to use in small production use-cases. The communication to the XD Admin server is over HTTP and the XD Admin server communicates to an in-process XD Container using an embedded ZooKeeper server.

local mode
Figure 24. Single Node Runtime

Admin Server Architecture

The Admin Server uses an embedded servlet container and exposes REST endpoints for creating, deploying, undeploying, and destroying streams and jobs, querying runtime state, analytics, and the like. The Admin Server is implemented using Spring’s MVC framework and the Spring HATEOAS library to create REST representations that follow the HATEOAS principle. The Admin Server and Container Servers monitor and update runtime state using ZooKeeper (see XD Distributed Runtime).

Container Server Architecture

The key components of data processing in Spring XD are

  • Streams

  • Jobs

  • Taps

Streams define how event driven data is collected, processed, and stored or forwarded. For example, a stream might collect syslog data, filter, and store it in HDFS.

Jobs define how coarse grained and time consuming batch processing steps are orchestrated, for example a job could be be defined to coordinate performing HDFS operations and the subsequent execution of multiple MapReduce processing tasks.

Taps are used to process data in a non-invasive way as data is being processed by a Stream or a Job. Much like wiretaps used on telephones, a Tap on a Stream lets you consume data at any point along the Stream’s processing pipeline. The behavior of the original stream is unaffected by the presence of the Tap.

tap jobs streams
Figure 25. Taps, Jobs, and Streams

Streams

The programming model for processing event streams in Spring XD is based on the well known Enterprise Integration Patterns as implemented by components in the Spring Integration project. The programming model was designed so that it is easy to test components.

A Stream consist of the following types of modules: * An Input source * Processing steps * An Output sink

An Input source produces messages from an external source. XD supports a variety of sources, e.g. syslog, tcp, http. The output from a module is a Spring Message containing a payload of data and a collection of key-value headers. Messages flow through message channels from the source, through optional processing steps, to the output sink. The output sink delivers the message to an external resource. For example, it is common to write the message to a file system, such as HDFS, but you may also configure the sink to forward the message over tcp, http, or another type of middleware, or route the message to another stream.

A stream that consists of a input source and a output sink is shown below

SourceSinkMessageChannel
Figure 26. Foundational components of the Stream processing model

A stream that incorporates processing steps is shown below

MultipleProcessingSteps
Figure 27. Stream processing with multiple steps

For simple linear processing streams, an analogy can be made with the UNIX pipes and filters model. Filters represent any component that produces, processes or consumes events. This corresponds to the modules (source, processing steps, and sink) in a stream. Pipes represent the way data is transported between the Filters. This corresponds to the Message Channel that moves data through a stream.

A simple stream definition using UNIX pipes and filters syntax that takes data sent via a HTTP post and writes it to a file (with no processing done in between) can be expressed as

http | file

The pipe symbol represents a message channel that passes data from the HTTP source to the File sink. The message channel implementation can either be backed with a local in-memory transport, Redis queues, or RabbitMQ. The message channel abstraction and the XD architecture are designed to support a pluggable data transport. Future releases will support other transports such as JMS.

Note that the UNIX pipes and filter syntax is the basis for the DSL that Spring XD uses to describe simple linear flows. Non-linear processing is partially supported using named channels which can be combined with a router sink to effectively split a single stream into multiple streams (see Dynamic Router Sink). Additional capabilities for non-linear processing are planned for future releases.

The programming model for processing steps in a stream originates from the Spring Integration project and is included in the core Spring Framework as of version 4. The central concept is one of a Message Handler class, which relies on simple coding conventions to Map incoming messages to processing methods. For example, using an http source you can process the body of an HTTP POST request using the following class

public class SimpleProcessor {

  public String process(String payload) {
    return payload.toUpperCase();
  }

}

The payload of the incoming Message is passed as a string to the method process. The contents of the payload is the body of the http request as we are using a http source. The non-void return value is used as the payload of the Message passed to the next step. These programming conventions make it very easy to test your Processor component in isolation. There are several processing components provided in Spring XD that do not require you to write any code, such as a filter and transformer that use the Spring Expression Language or Groovy. For example, adding a processing step, such as a transformer, in a stream processing definition can be as simple as

http | transformer --expression=payload.toUpperCase() | file

For more information on processing modules, refer to the Processors section.

Stream Deployment

The Container Server listens for module deployment events initiated from the Admin Server via ZooKeeper. When the container node handles a module deployment event, it connects the module’s input and output channels to the data bus used to transport messages during stream processing. In a single node configuration, the data bus uses in-memory direct channels. In a distributed configuration, the data bus communications are backed by the configured transport middleware. Redis and Rabbit are both provided with the Spring XD distribution, but other transports are envisioned for future releases.

anatomyOfAStreamSingleNode
Figure 28. A Stream Deployed in a single node server
anatomyOfAStreamV3
Figure 29. A Stream Deployed in a distributed runtime

In the http | file example, the Admin assigns each module to a separate Container instance, provided there are at least two Containers available. The file module is deployed to one container and the http module to another. The definition of a module is stored in a Module Registry. A module definition consists of a Spring XML configuration file, some classes used to validate and handle options defined by the module, and dependent jars. The module definition contains variable placeholders, corresponding to DSL parameters (called options) that allow you to customize the behavior of the module. For example, setting the http listening port would be done by passing in the option --port, e.g. http --port=8090 | file, which is in turn used to substitute a placeholder value in the module definition.

The Module Registry is backed by the filesystem and corresponds to the directory <xd-install-directory>/modules. When a module deployment is handled by the Container, the module definition is loaded from the registry and a new Spring ApplicationContext is created in the Container process to run the module. Dependent classes are loaded via the Module Classloader which first looks at jars in the modules /lib directory before delegating to the parent classloader.

Using the DIRT runtime, the http | file example would map onto the following runtime architecture

http2file
Figure 30. Distributed HTTP to File Stream

Data produced by the HTTP module is sent over a Redis Queue and is consumed by the File module. If there was a filter processing module in the stream definition, e.g http | filter | file that would map onto the following DIRT runtime architecture.

http2filter2file
Figure 31. Distributed HTTP to Filter to File Stream

Jobs

The creation and execution of Batch jobs builds upon the functionality available in the Spring Batch and Spring for Apache Hadoop projects. See the Batch Jobs section for more information.

Taps

Taps provide a non-invasive way to consume the data that is being processed by either a Stream or a Job, much like a real time telephone wire tap lets you eavesdrop on telephone conversations. Taps are recommended as way to collect metrics and perform analytics on a Stream of data. See the section Taps for more information.

Distributed Runtime

Introduction

This document describes what’s happening "under the hood" of the Spring XD Distributed Runtime (DIRT) and in particular, how the runtime architecture achieves high availability and failover in a clustered production environment. See Running in Distributed Mode for more information on installing and running Spring XD in distributed mode.

This discussion focuses on Spring XD’s core runtime components and the role of ZooKeeper in managing the state of the Spring XD cluster and enabling automatic recovery from failures.

Configuring Spring XD for High Availabilty(HA)

A production Spring XD environment is typically distributed among multiple hosts in a clustered environment. Spring XD scales horizontally when you add container instances. In the simplest case, all containers are replicas, that is each instance is running on an identically configured host and modules are deployed to any available container in a round-robin fashion. However, this simplifying assumption does not address real production scenarios in which more control is requred in order to optimize resource utilization. To this end, Spring XD supports a flexible algorithm which allows you to match module deployments to specific container configurations. The container matching algorithm will be covered in more detail later, but for now, let’s assume the simple case. Running multiple containers not only enables horizontal scalability, but enables failure recovery. If a container becomes unavailable due to an unrecoverable connection loss, any modules currently deployed to that container will be deployed automatically to the other available instances.

Spring XD requires that a single active Admin server handle interactions with the containers, such as stream deployment requests, as these types of operations must be processed serially in the order received. Without a backup, the Admin server becomes single point of failure. Therefore, two (or more for the risk averse) Admin servers are recommended for a production environment. Note that every Admin server can handle all requests via REST endpoints but only one instance, the "Leader", will actually perform requests that update the runtime state. If the Leader goes down, another available Admin server will assume the leader role. Leader Election is an example of a common feature for distributed systems provided by the Curator Framework which sits on top of ZooKeeper.

An HA Spring XD installation also requires that external servers - ZooKeeper, messaging middleware, and data stores needed for running Spring XD in distributed mode must be configured for HA as well. Please consult the product documentation for specific recommendations regarding each of these external components. Also see Message Bus Configuration for tips on configuring the MessageBus for HA, error handling, etc.

ZooKeeper Overview

In the previous section, we claimed that if a container goes down, Spring XD will redeploy any modules deployed on that instance to another available container. We also claimed that if the Admin Leader goes down, another Admin server will assume that role. ZooKeeper is what makes this all possible. ZooKeeper is a widely used Apache project designed primarily for distributed system management and coordination. This section will cover some basic concepts necessary to understand its role in Spring XD. See The ZooKeeper Wiki for a more complete overview.

ZooKeeper is based on a simple hierarchical data structure, formally a tree, and conceptually and semantically similar to a file directory structure. As such, data is stored in nodes. A node is referenced via a path, for example, /xd/streams/mystream. Each node can store additional data, serialized as a byte array. In Spring XD, all data is a java.util.Map serialized as JSON. The following figure shows the Spring XD schema:

ZooKeeper XD Schema

A ZooKeeper node is either ephemeral or persistent. An ephemeral node exists only as long as the session that created it remains active. A persistent node is, well, persistent. Ephemeral nodes are appropriate for registering Container instances. When an Spring XD container starts up it creates an ephemeral node, /xd/containers/<container-id>, using an internally generated container id. When the container’s session is closed due to a connection loss, for example, the container process terminates, its node is removed. The ephemeral container node also holds metadata such as its hostname and IP address, runtime metrics, and user defined container attributes. Persistent nodes maintain state needed for normal operation and recovery. This includes data such as stream definitions, job definitions, deployment manifests, module deployments, and deployment state for streams and jobs.

Obviously ZooKeeper is a critical piece of the Spring XD runtime and must itself be HA. ZooKeeper itself supports a distributed architecture, called an ensemble. The details are beyond the scope of this document but for the sake of this discussion it is worth mentioning that there should be at least three ZooKeeper server instances running (an odd number is always recommended) on dedicated hosts. The Container and Admin nodes are clients to the ZooKeeper ensemble and must connect to ZooKeeper at startup. Spring XD components are configured with a zk.client.connect property which may designate a single <host>:<port> or a comma separated list. The ZooKeeper client will attempt to connect to each server in order until it succeeds. If it is unable to connect, it will keep trying. If a connection is lost, the ZooKeeper client will attempt to reconnect to one of the servers. The ZooKeeper cluster guarantees consistent replication of data across the ensemble. Specifically, ZooKeeper guarantees:

  • Sequential Consistency - Updates from a client will be applied in the order that they were sent.

  • Atomicity - Updates either succeed or fail. No partial results.

  • Single System Image - A client will see the same view of the service regardless of the server that it connects to.

  • Reliability - Once an update has been applied, it will persist from that time forward until a client overwrites the update.

  • Timeliness - The clients view of the system is guaranteed to be up-to-date within a certain time bound.

ZooKeeper maintains data primarily in memory backed by a disk cache. Updates are logged to disk for recoverability, and writes are serialized to disk before they are applied to the in-memory database.

In addition to performing basic CRUD operations on nodes, A ZooKeeper client can register a callback on a node to respond to any events or state changes to that node or any of its children. Such node operations and callbacks are the mechanism that control the Spring XD runtime.

Spring XD and ZK Ensemble

The Admin Server Internals

Assuming more than one Admin instance is running, Each instance requests leadership at start up. If there is already a designated leader, the instance will watch the xd/admin node to be notified if the Leader goes away. The instance designated as the "Leader", using the Leader Selector recipe provided by Curator, a ZooKeeper client library that implements some common patterns. Curator also provides some Listener callback interfaces that the client can register on a node. The AdminServer creates the top level nodes, depicted in the figure above:

  • /xd/admins - children are ephemeral nodes for each available Admin instance and used for Leader Selector

  • /xd/containers - children are ephemeral nodes containing runtime attributes including hostname,process id, ip address, and user defined attributes for each container instance.

  • /xd/streams - children are persistent nodes containing the definition (DSL) for each stream.

  • /xd/jobs - children are persistent nodes containing the definition (DSL) for each job.

  • /xd/taps - children are persistent nodes describing each deployed tap.

  • /xd/deployments/streams - children are nodes containing stream deployment status (leaf nodes are ephemeral).

  • /xd/deployments/jobs - children are nodes containing job deployment status (leaf nodes are ephemeral).

  • /xd/deployments/modules/requested - stores module deployment requests including deployment criteria.

  • /xd/deployments/modules/allocated - stores information describing currently deployed modules.

The admin leader creates a DeploymentSupervisor which registers listeners on /xd/deployments/modules/requested to handle module deployment requests related to stream and job deployments, and xd/containers/ to be notified when containers are added and removed from the cluster. Note that any Admin instance can handle user requests. For example, if you enter the following commands via XD shell,

xd>stream create ticktock --definition "time | log"

This command will invoke a REST service on its connected Admin instance to create a new node /xd/streams/ticktock

xd>stream deploy ticktock

Assuming the deployment is successful, This will result in the creation of several nodes used to manage deployed resources, for example, /xd/deployments/streams/ticktock. The details are discussed in the example below.

If the Admin instance connected to the shell is not the Leader, it will perform no further action. The Leader’s DeploymentSupervisor will attempt to deploy each module in the stream definition, in accordance with the deployment manifest, to an available container, and update the runtime state.

XD Admin Internals

Example

Let’s walk through the simple example above. If you don’t have a Spring XD cluster set up, this example can be easily executed running Spring XD in a single node configuration. The single node application includes an embedded ZooKeeper server by default and allocates a random unused port. The embedded ZooKeeper connect string is reported in the console log for the single node application:

...
13:04:27,016  INFO main util.XdConfigLoggingInitializer - Transport: local
13:04:27,016  INFO main util.XdConfigLoggingInitializer - Hadoop Distro: hadoop22
13:04:27,019  INFO main util.XdConfigLoggingInitializer - Hadoop version detected from classpath: 2.2.0
13:04:27,019  INFO main util.XdConfigLoggingInitializer - Zookeeper at: localhost:31316
...

For our purposes, we will use the ZooKeeper CLI tool to inspect the contents of ZooKeeper nodes reflecting the current state of Spring XD. First, we need to know the port to connect the CLI tool to the embedded server. For convenience, we will assign the ZooKeeper port (5555 in this example) when starting the single node application. From the XD install directory:

$export JAVA_OPTS="-Dzk.embedded.server.port=5555"
$xd/bin/xd-singlenode

In another terminal session, start the ZooKeeper CLI included with ZooKeeper to connect to the embedded server and inspect the contents of the nodes (NOTE: tab completion works) :

$zkCli.sh -server localhost:5555

After some console output, you should see a prompt:

WatchedEvent state:SyncConnected type:None path:null
[zk: localhost:5555(CONNECTED) 0]

navigate using the ls command. This will reflect the schema shown in the figure above, the unique container ID will be different for you.

[[zk: localhost:5555(CONNECTED) 0] ls /xd
[deployments, containers, admins, taps, streams, jobs]
[zk: localhost:5555(CONNECTED) 1] ls /xd/streams
[]
[zk: localhost:5555(CONNECTED) 2] ls /xd/deployments
[jobs, streams, modules]
[zk: localhost:5555(CONNECTED) 3] ls /xd/deployments/streams
[]
[zk: localhost:5555(CONNECTED) 4] ls /xd/deployments/modules
[requested, allocated]
[zk: localhost:5555(CONNECTED) 5] ls /xd/deployments/modules/allocated
[2ebbbc9b-63ac-4da4-aa32-e39d69eb546b]
[zk: localhost:5555(CONNECTED) 6] ls /xd/deployments/modules/2ebbbc9b-63ac-4da4-aa32-e39d69eb546b
[]
[zk: localhost:5555(CONNECTED) 7] ls /xd/containers
[2ebbbc9b-63ac-4da4-aa32-e39d69eb546b]
[zk: localhost:5555(CONNECTED) 8]

The above reflects the initial state of Spring XD with a running admin and container instance. Nothing is deployed yet and there are no existing stream or job definitions. Note that xd/deployments/modules/allocated has a persistent child corresponding to the id of the container at xd/containers. If you are running in a distributed configuration and connected to one of the ZooKeeper servers in the same ensemble that Spring XD is connected to, you might see multiple nodes under /xd/containers, and xd/admins. Because the external ensemble persists the state of the Spring XD cluster, you will also see any deployments that existed when the Spring XD cluster was shut down.

Start the XD Shell in a new terminal session and create a stream:

$ shell/bin/xd-shell
 _____                           __   _______
/  ___|          (-)             \ \ / /  _  \
\ `--. _ __  _ __ _ _ __   __ _   \ V /| | | |
 `--. \ '_ \| '__| | '_ \ / _` |  / ^ \| | | |
/\__/ / |_) | |  | | | | | (_| | / / \ \ |/ /
\____/| .__/|_|  |_|_| |_|\__, | \/   \/___/
      | |                  __/ |
      |_|                 |___/
eXtreme Data
1.3.3.BUILD-SNAPSHOT | Admin Server Target: http://localhost:9393
Welcome to the Spring XD shell. For assistance hit TAB or type "help".
xd:>stream create ticktock --definition "time | log"
Created new stream 'ticktock'
xd:>

Back to the ZK CLI session:

[zk: localhost:5555(CONNECTED) 8] ls /xd/streams
[ticktock]
[zk: localhost:5555(CONNECTED) 9] get /xd/streams/ticktock
{"definition":"time | log"}
cZxid = 0x31
ctime = Mon Jul 14 10:32:33 EDT 2014
mZxid = 0x31
mtime = Mon Jul 14 10:32:33 EDT 2014
pZxid = 0x31
cversion = 0
dataVersion = 0
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 27
numChildren = 0
[zk: localhost:5555(CONNECTED) 10]

using the get command on the new stream node, we can see the stream definition represented as JSON, along with some standard ZooKeeper metadata.

Note
ephemeralOwner = 0x0, indicating this is not an ephemeral node. At this point, nothing else should have changed from the initial state.

Now, Using the Spring XD shell, let’s deploy the stream,

xd>stream deploy ticktock
Deployed stream 'ticktock'

and verify with ZooKeeper:

[zk: localhost:5555(CONNECTED) 10] ls /xd/deployments/streams
[ticktock]
[zk: localhost:2181(CONNECTED) 11] ls /xd/streams/deployments/ticktock
[modules, status]
[[zk: localhost:2181(CONNECTED) 12] get /xd/deployments/streams/ticktock/status
{"state":"deployed"}
....
zk: localhost:2181(CONNECTED) 13] ls /xd/deployments/streams/ticktock/modules
[source.time.1.2ebbbc9b-63ac-4da4-aa32-e39d69eb546b, sink.log.1.2ebbbc9b-63ac-4da4-aa32-e39d69eb546b]

Note the deployment state shown for the stream’s status node is deployed, meaning the deployment request was satisfied. Deployment states are discussed in more detail here.

Spring XD decomposes stream deployment requests to individual module deployment requests. Hence, we see that each module in the stream is associated with a container instance. The container instance in this case is the same since there is only one instance in the single node configuration. In a distributed configuration with more than one instance, the stream source and sink will each be deployed to a separate container. The node name itself is of the form <module_type>.<module_name>.<module_sequence_number>.<container_id>, where the sequence number identifies a deployed instance of a module if multiple instances of that module are requested.

[zk: localhost:2181(CONNECTED) 14] ls /xd/deployments/modules/allocated/2ebbbc9b-63ac-4da4-aa32-e39d69eb546b/ticktock.source.time.1
[metadata, status]

The metadata and status nodes are ephemeral nodes which store details about the deployed module. This information is provided to XD shell queries. For example:

xd:>runtime modules
  Module                  Container Id                          Options                                          Deployment Properties
  ----------------------  ------------------------------------  -----------------------------------------------  ---------------------
  ticktock.sink.log.1     2ebbbc9b-63ac-4da4-aa32-e39d69eb546b  {name=ticktock, expression=payload, level=INFO}  {count=1, sequence=1}
  ticktock.source.time.1  2ebbbc9b-63ac-4da4-aa32-e39d69eb546b  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}       {count=1, sequence=1}

Module Deployment

This section describes how the Spring XD runtime manages deployment internally. For more details on how to deploy streams and jobs see Deployment.

To process a stream deployment request, the StreamDeploymentListener invokes its ContainerMatcher to select a container instance for each module and records the module’s deployment properties under /xd/deployments/modules/requested/. If a match is found, the StreamDeploymentListener creates a node for the module under /xd/deployments/modules/allocated/<container_id>. The Container includes a DeploymentListener that monitors the container node for new modules to deploy. If the deployment is successful, the Container writes the ephemeral nodes status and metadata under the new module node.

Module Deployment

When a container departs, the ephemeral nodes are deleted so its modules are now undeployed. The ContainerListener responds to the deleted nodes and attempts to redeploy any affected modules to another instance.

Example: Automatic Redeployment

For this example we start two container instances and deploy and simple stream:

xd:>runtime containers
  Container Id                          Host            IP Address   PID    Groups  Custom Attributes
  ------------------------------------  --------------  -----------  -----  ------  -----------------
  0ddf80b9-1e80-44b8-8c12-ecc5c8c32e11  ultrafox.local  192.168.1.6  19222
  6cac85f8-4c52-4861-a225-cdad3675f6c9  ultrafox.local  192.168.1.6  19244

xd:>stream create ticktock --definition "time | log"
Created new stream 'ticktock'
xd:>stream deploy ticktock
Deployed stream 'ticktock'
xd:>runtime modules
  Module                  Container Id                          Options                                          Deployment Properties
  ----------------------  ------------------------------------  -----------------------------------------------  ---------------------
  ticktock.sink.log.1     0ddf80b9-1e80-44b8-8c12-ecc5c8c32e11  {name=ticktock, expression=payload, level=INFO}  {count=1, sequence=1}
  ticktock.source.time.1  6cac85f8-4c52-4861-a225-cdad3675f6c9  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}       {count=1, sequence=1}

Now we will kill one of the container processes and observe that the affect module has been redeployed to the remaining container:

xd:>runtime containers
  Container Id                          Host            IP Address   PID    Groups  Custom Attributes
  ------------------------------------  --------------  -----------  -----  ------  -----------------
  6cac85f8-4c52-4861-a225-cdad3675f6c9  ultrafox.local  192.168.1.6  19244

xd:>runtime modules
  Module                  Container Id                          Options                                          Deployment Properties
  ----------------------  ------------------------------------  -----------------------------------------------  ---------------------
  ticktock.sink.log.1     6cac85f8-4c52-4861-a225-cdad3675f6c9  {name=ticktock, expression=payload, level=INFO}  {count=1, sequence=1}
  ticktock.source.time.1  6cac85f8-4c52-4861-a225-cdad3675f6c9  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}       {count=1, sequence=1}

Now if we kill the remaining container, we see warnings in the xd-admin log:

14:36:07,593  WARN DeploymentSupervisorCacheListener-0 server.DepartingContainerModuleRedeployer - No containers available for redeployment of log for stream ticktock
14:36:07,599  WARN DeploymentSupervisorCacheListener-0 server.DepartingContainerModuleRedeployer - No containers available for redeployment of time for stream ticktock

Batch Jobs

Introduction

One of the features that XD offers is the ability to launch and monitor batch jobs based on Spring Batch. The Spring Batch project was started in 2007 as a collaboration between SpringSource and Accenture to provide a comprehensive framework to support the development of robust batch applications. Batch jobs have their own set of best practices and domain concepts which have been incorporated into Spring Batch building upon Accenture’s consulting business. Since then Spring Batch has been used in thousands of enterprise applications and is the basis for the recent JSR standardization of batch processing, JSR-352.

Spring XD builds upon Spring Batch to simplify creating batch workflow solutions that span traditional use-cases such as moving data between flat files and relational databases as well as Hadoop use-cases where analysis logic is broken up into several steps that run on a Hadoop cluster. Steps specific to Hadoop in a workflow can be MapReduce jobs, executing Hive/Pig scripts or HDFS operations. Spring XD ships with a number of useful job modules out of the box. This section presents an overview for creating, deploying, and running batch jobs.

Workflow

The concept of a workflow translates to a Job, not to be confused with a MapReduce job. A Job is a directed graph, each node of the graph is a processing Step. Steps can be executed sequentially or in parallel, depending on the configuration. Jobs can be started, stopped, and restarted. Restarting jobs is possible since the progress of executed steps in a Job is persisted in a database via a JobRepository. The following figures shows the basic components of a workflow.

The Spring XD Workflow overview

A Job that has steps specific to Hadoop is shown below.

Steps in a workflow that execute Hadoop HDFS operations and run Pig

A JobLauncher is responsible for starting a job and is often triggered via a scheduler. Other options to launch a job are through Spring XD’s RESTful administration API, the XD web application, or in response to an external event from and XD stream definition, e.g. file polling using the file source.

Features

Spring XD allows you to create and launch jobs. The launching of a job can be triggered using a cron expression or in reaction to data on a stream. When jobs are executing, they are also a source of event data that can be subscribed to by a stream. There are several type of events sent during a job’s execution, the most common being the status of the job and the steps taken within the job. This bi-direction communication between stream processing and batch processing allows for more complex chains of processing to be developed.

As a starting point, jobs for the following cases are provided to use out of the box

  • Poll a Directory and import CSV files to HDFS

  • Import CSV files to JDBC

  • HDFS to JDBC Export

  • JDBC to HDFS Import

  • HDFS to MongoDB Export

These are described in the section below.

The purpose of this section is to show you how to create, schedule and monitor a job.

The Lifecycle of a Job in Spring XD

Before we dive deeper into the details of creating batch jobs with Spring XD, we need to understand the typical lifecycle for batch jobs in the context of Spring XD:

  1. Register a Job Module

  2. Create a Job Definition

  3. Deploy a Job

  4. Launch a Job

  5. Job Execution

  6. Un-deploy a Job

  7. Destroy a Job Definition

Register a Job Module

Register a Job Module with the Module Registry using the XD Shell module upload command. See registering a module for details.

Create a Job Definition

Create a Job Definition from a Job Module by providing a definition name as well as properties that apply to all Job Instances. At this point the job is not deployed, yet.

Deploy the Job

Deploy the Job Definition to one or more Spring XD containers. This will initialize the Job Definitions on those containers. The jobs are now "live" and a job can be created by sending a message to a job queue that contains optional runtime Job Parameters.

Launch a Job

Launch a job by sending a message to the job queue with Job Parameters. A Job Instance is created, representing a specific run of the job. A Job Instance is the Job Definition plus the runtime Job Parameters. You can query for the Job Instances associated with a given job name.

Job Execution

The job is executed creating a Job Execution object that captures the success or failure of the job. You can query for Job Executions associated with a given job name.

Un-deploy a Job

This removes the job from the Spring XD container(s) preventing the launching of any new Job Instances. For reporting purposes, you will still be able to view historic Job Executions associated with the the job.

Destroy a Job Definition

Destroying a Job Definition will not only un-deploy any still deployed Job Definitions but will also remove the Job Definition itself.

Creating Jobs - Additional Options

When creating jobs, the following options are available to all job definitions:

dateFormat

The optional date format for job parameters (default: yyyy-MM-dd)

numberFormat

Defines the number format when parsing numeric parameters (default: NumberFormat.getInstance(Locale.US))

makeUnique

Shall job parameters be made unique? (default: true)

Also, similar to the stream create command, the job create command has an optional --deploy option to create the job definition and deploy it. --deploy option is false by default.

Below is an example of some of these options combined:

job create myjob --definition "fooJob --makeUnique=false"

Remember that you can always find out about available options for a job by using the module info command.

Deployment manifest support for job

When deploying batch job you can provide a deployment manifest. Deployment manifest properties for jobs are the same as for streams, you can declare

  • The number of job modules to deploy

  • The criteria expression to use for matching the job to available containers

For example,

job create myjob --definition "fooJob --makeUnique=false"

job deploy myjob --properties "module.fooJob.count=3,module.fooJob.criteria=groups.contains('hdfs-containers-group')"

The above deployment manifest would deploy 3 number of fooJob modules into containers whose group name matches "hdfs-containers-group".

When a batch job is launched/scheduled, the job module that picks up the job launching request message executes the batch job. To support partitioning of the job across multiple containers, the job definition needs to define how the job will be partitioned. The type of partitioning depends on the type of the job, for example a job reading from JDBC would partition the data in a table by dividing up the number of rows and a job reading files form a directory would partition on the number of files available.

The FTP to HDFS and FILE to JDBC jobs support for partitioning. To add partitioning support for your own jobs you should import singlestep-partition-support.xml in your job definition. This provides the infrastructure so that the job module that processes the launch request can communicate as the master with the other job modules that have been deployed. You will also need to provide an implementation of the Partitioner interface.

For more information on the deployment manifest, please refer here

Launching a job

XD uses triggers as well as regular event flow to launch the batch jobs. So in this section we will cover how to:

  • Launch the Batch Job Ad-hoc

  • Launch the Batch Job using a named Cron-Trigger

  • Launch the Batch Job as sink.

Ad-hoc

To launch a job one time, use the launch option of the job command. So going back to our example above, we’ve created a job module instance named helloSpringXD. Launching that Job Module Instance would look like:

xd:> job launch helloSpringXD

In the logging output of the XDContainer you should see the following

16:45:40,127  INFO http-bio-9393-exec-1 job.JobPlugin:98 - Configuring module with the following properties: {numberFormat=, dateFormat=, makeUnique=true, xd.job.name=myjob}
16:45:40,185  INFO http-bio-9393-exec-1 module.SimpleModule:140 - initialized module: SimpleModule [name=job, type=job, group=myjob, index=0 @3a9ecb9d]
16:45:40,198  INFO http-bio-9393-exec-1 module.SimpleModule:161 - started module: SimpleModule [name=job, type=job, group=myjob, index=0 @3a9ecb9d]
16:45:40,199  INFO http-bio-9393-exec-1 module.ModuleDeployer:161 - deployed SimpleModule [name=job, type=job, group=myjob, index=0 @3a9ecb9d]
Hello Spring XD!

To re-launch the job just execute the launch command. For example:

xd:> job launch helloSpringXD

Launch the Batch using Cron-Trigger

To launch a batch job based on a cron scheduler is done by creating a stream using the trigger source.

xd:> stream create --name cronStream --definition "trigger --cron='0/5 * * * * *'  > queue:job:myCronJob" --deploy

A batch job can receive parameters from a source (in this case a trigger) or process. A trigger uses the --payload expression to declare its payload.

xd:> stream create --name cronStream --definition "trigger --cron='0/5 * * * * *'  --payload={\"param1\":\"Kenny\"} > queue:job:myCronJob" --deploy
Note
The payload content must be in a JSON-based map representation.

To pause/stop future scheduled jobs from running for this stream, the stream must be undeployed for example:

xd:> stream undeploy --name cronStream

Launch the Batch using a Fixed-Delay-Trigger

A fixed-delay-trigger is used to launch a Job on a regular interval. Using the --fixedDelay parameter you can set up the number of seconds between executions. In the example below we are running myXDJob every 10 seconds and passing it a payload containing a single attribute.

xd:> stream create --name fdStream --definition "trigger --payload={\"param1\":\"fixedDelayKenny\"} --fixedDelay=5 > queue:job:myXDJob" --deploy

To pause/stop future scheduled jobs from running for this stream, you must undeploy the stream for example:

xd:> stream undeploy --name fdStream

Launch job as a part of event flow

A batch job is always used as a sink, with that being said it can receive messages from sources (other than triggers) and processors. In the case below we see that the user has created an http source (http source receives http posts and passes the payload of the http message to the next module in the stream) that will pass the http payload to the "myHttpJob".

 stream create --name jobStream --definition "http > queue:job:myHttpJob" --deploy

To test the stream you can execute a http post, like the following:

xd:> http post --target http://localhost:9000 --data "{\"param1\":\"fixedDelayKenny\"}"

Retrieve job notifications

Spring XD offers the facilities to capture the notifications that are sent from the job as it is executing. When a batch job is deployed, by default it registers the following listeners along with pub/sub channels that these listeners send messages to.

  • Job Execution Listener

  • Chunk Listener

  • Item Listener

  • Step Execution Listener

  • Skip Listener

Along with the pub/sub channels for each of these listeners, there will also be a pub/sub channel that the aggregated events from all these listeners are published to.

In the following example, we setup a Batch Job called myHttpJob. Afterwards we create a stream that will tap into the pub/sub channels that were implicitly generated when the myHttpJob job was deployed.

To receive aggregated events

The stream receives aggregated event messages from all the default batch job listeners and sends those messages to the log.

xd>job create --name myHttpJob --definition "httpJob" --deploy
xd>stream create --name aggregatedEvents --definition "tap:job:myHttpJob >log" --deploy
xd>job launch myHttpJob

Note: The syntax for the tap that receives the aggregated events is: tap:job:<job-name>

In the logging output of the container you should see something like the following when the job completes (with the aggregated events

09:55:53,532  WARN SimpleAsyncTaskExecutor-1 logger.aggregatedEvents:150 - JobExecution: id=2, version=1, startTime=Sat Apr 12 09:55:53 PDT 2014, endTime=null, lastUpdated=Sat Apr 12 09:55:53 PDT 2014, status=STARTED, exitStatus=exitCode=UNKNOWN;exitDescription=, job=[JobInstance: id=2, version=0, Job=[myHttpJob]], jobParameters=[{random=0.07002785662707867}]
09:55:53,554  WARN SimpleAsyncTaskExecutor-1 logger.aggregatedEvents:150 - StepExecution: id=2, version=1, name=step1, status=STARTED, exitStatus=EXECUTING, readCount=0, filterCount=0, writeCount=0 readSkipCount=0, writeSkipCount=0, processSkipCount=0, commitCount=0, rollbackCount=0, exitDescription=
09:55:53,561  WARN SimpleAsyncTaskExecutor-1 logger.aggregatedEvents:150 - XdChunkContextInfo [complete=false, stepExecution=StepExecution: id=2, version=1, name=step1, status=STARTED, exitStatus=EXECUTING, readCount=0, filterCount=0, writeCount=0 readSkipCount=0, writeSkipCount=0, processSkipCount=0, commitCount=0, rollbackCount=0, exitDescription=, attributes={}]
09:55:53,567  WARN SimpleAsyncTaskExecutor-1 logger.aggregatedEvents:150 - XdChunkContextInfo [complete=false, stepExecution=StepExecution: id=2, version=2, name=step1, status=STARTED, exitStatus=EXECUTING, readCount=0, filterCount=0, writeCount=0 readSkipCount=0, writeSkipCount=0, processSkipCount=0, commitCount=1, rollbackCount=0, exitDescription=, attributes={}]
09:55:53,573  WARN SimpleAsyncTaskExecutor-1 logger.aggregatedEvents:150 - StepExecution: id=2, version=2, name=step1, status=COMPLETED, exitStatus=COMPLETED, readCount=0, filterCount=0, writeCount=0 readSkipCount=0, writeSkipCount=0, processSkipCount=0, commitCount=1, rollbackCount=0, exitDescription=
09:55:53,580  WARN SimpleAsyncTaskExecutor-1 logger.aggregatedEvents:150 - JobExecution: id=2, version=1, startTime=Sat Apr 12 09:55:53 PDT 2014, endTime=Sat Apr 12 09:55:53 PDT 2014, lastUpdated=Sat Apr 12 09:55:53 PDT 2014, status=COMPLETED, exitStatus=exitCode=COMPLETED;exitDescription=, job=[JobInstance: id=2, version=0, Job=[myHttpJob]], jobParameters=[{random=0.07002785662707867}]

To receive job execution events

xd>job create --name myHttpJob --definition "httpJob" --deploy
xd>stream create --name jobExecutionEvents --definition "tap:job:myHttpJob.job >log" --deploy
xd>job launch myHttpJob

Note: The syntax for the tap that receives the job execution events is: tap:job:<job-name>.job

In the logging output of the container you should see something like the following when the job completes

10:06:41,579  WARN SimpleAsyncTaskExecutor-1 logger.jobExecutionEvents:150 - JobExecution: id=3, version=1, startTime=Sat Apr 12 10:06:41 PDT 2014, endTime=null, lastUpdated=Sat Apr 12 10:06:41 PDT 2014, status=STARTED, exitStatus=exitCode=UNKNOWN;exitDescription=, job=[JobInstance: id=3, version=0, Job=[myHttpJob]], jobParameters=[{random=0.3774227747555795}]
10:06:41,626  INFO SimpleAsyncTaskExecutor-1 support.SimpleJobLauncher:136 - Job: [FlowJob: [name=myHttpJob]] completed with the following parameters: [{random=0.3774227747555795}] and the following status: [COMPLETED]
10:06:41,626  WARN SimpleAsyncTaskExecutor-1 logger.jobExecutionEvents:150 - JobExecution: id=3, version=1, startTime=Sat Apr 12 10:06:41 PDT 2014, endTime=Sat Apr 12 10:06:41 PDT 2014, lastUpdated=Sat Apr 12 10:06:41 PDT 2014, status=COMPLETED, exitStatus=exitCode=COMPLETED;exitDescription=, job=[JobInstance: id=3, version=0, Job=[myHttpJob]], jobParameters=[{random=0.3774227747555795}]

To receive step execution events

xd>job create --name myHttpJob --definition "httpJob" --deploy
xd>stream create --name stepExecutionEvents --definition "tap:job:myHttpJob.step >log" --deploy
xd>job launch myHttpJob

Note: The syntax for the tap that receives the step execution events is: tap:job:<job-name>.step

In the logging output of the container you should see something like the following when the job completes

10:13:16,072  WARN SimpleAsyncTaskExecutor-1 logger.stepExecutionEvents:150 - StepExecution: id=6, version=1, name=step1, status=STARTED, exitStatus=EXECUTING, readCount=0, filterCount=0, writeCount=0 readSkipCount=0, writeSkipCount=0, processSkipCount=0, commitCount=0, rollbackCount=0, exitDescription=
10:13:16,092  WARN SimpleAsyncTaskExecutor-1 logger.stepExecutionEvents:150 - StepExecution: id=6, version=2, name=step1, status=COMPLETED, exitStatus=COMPLETED, readCount=0, filterCount=0, writeCount=0 readSkipCount=0, writeSkipCount=0, processSkipCount=0, commitCount=1, rollbackCount=0, exitDescription=

To receive item, skip and chunk events

xd>job create --name myHttpJob --definition "httpJob" --deploy

xd>stream create --name itemEvents --definition "tap:job:myHttpJob.item >log" --deploy
xd>stream create --name skipEvents --definition "tap:job:myHttpJob.skip >log" --deploy
xd>stream create --name chunkEvents --definition "tap:job:myHttpJob.chunk >log" --deploy

xd>job launch myHttpJob

Note: The syntax for the tap that receives the item events: tap:job:<job-name>.item,for skip events: tap:job:<job-name>.skip and for chunk events: tap:job:<job-name>.chunk

To disable the default listeners

xd>job create --name myHttpJob --definition "httpJob --listeners=disable" --deploy

To select specific listeners

To select specific listeners, specify comma separated list in --listeners option. Following example illustrates the selection of job and step execution listeners only:

xd>job create --name myHttpJob --definition "httpJob --listeners=job,step" --deploy

Note: List of options are: job, step, item, chunk and skip The aggregated channel is registered if at least one of these default listeners are enabled.

For a complete example, please see the Batch Notifications Sample which is part of the Spring XD Samples repository.

Removing Batch Jobs

Batch Jobs can be deleted by executing:

xd:> job destroy helloSpringXD

Alternatively, one can just undeploy the job, keeping its definition for a future redeployment:

xd:> job undeploy helloSpringXD

Streams

Introduction

In Spring XD, a basic stream defines the ingestion of event driven data from a source to a sink that passes through any number of processors. Stream processing is performed inside the XD Containers and the deployment of stream definitions to containers is done via the XD Admin Server. The Getting Started section shows you how to start these servers and how to start and use the Spring XD shell

Sources, sinks and processors are predefined configurations of a module. Module definitions are found in the –0— directory. [1]. Modules definitions are standard Spring configuration files that use existing Spring classes, such as Input/Output adapters and Transformers from Spring Integration that support general Enterprise Integration Patterns.

A high level DSL is used to create stream definitions. The DSL to define a stream that has an http source and a file sink (with no processors) is shown below

http | file

The DSL mimics a UNIX pipes and filters syntax. Default values for ports and filenames are used in this example but can be overriden using -- options, such as

http --port=8091 | file --dir=/tmp/httpdata/

To create these stream definitions you make an HTTP POST request to the XD Admin Server. More details can be found in the sections below.

Creating a Simple Stream

The XD Admin server [2] exposes a full RESTful API for managing the lifecycle of stream definitions, but the easiest way to use the XD shell. Start the shell as described in the Getting Started section

New streams are created by posting stream definitions. The definitions are built from a simple DSL. For example, let’s walk through what happens if we execute the following shell command:

xd:> stream create --definition "time | log" --name ticktock

This defines a stream named ticktock based off the DSL expression time | log. The DSL uses the "pipe" symbol |, to connect a source to a sink.

Then to deploy the stream execute the following shell command (or alternatively add the --deploy flag when creating the stream so that this step is not needed):

xd:> stream deploy --name ticktock

The stream server finds the time and log definitions in the modules directory and uses them to setup the stream. In this simple example, the time source simply sends the current time as a message each second, and the log sink outputs it using the logging framework.

processing module 'Module [name=log, type=sink]' from group 'ticktock' with index: 1
processing module 'Module [name=time, type=source]' from group 'ticktock' with index: 0
17:26:18,774  WARN ThreadPoolTaskScheduler-1 logger.ticktock:141 - Thu May 23 17:26:18 EDT 2013

If you would like to have multiple instances of a module in the stream, you can include a property with the deploy command:

xd:> stream deploy --name ticktock --properties "module.time.count=3"

You can also include a SpEL Expression as a criteria property for any module. That will be evaluated against the attributes of each currently available Container. Instances of the module will only be deployed to Containers for which the expression evaluates to true.

xd:> stream deploy --name ticktock --properties "module.time.count=3,module.log.criteria=groups.contains('x')"
Important
See Module Labels.

Deleting a Stream

You can delete a stream by issuing the stream destroy command from the shell:

xd:> stream destroy --name ticktock

Deploying and Undeploying Streams

Often you will want to stop a stream, but retain the name and definition for future use. In that case you can undeploy the stream by name and issue the deploy command at a later time to restart it.

xd:> stream undeploy --name ticktock
xd:> stream deploy --name ticktock

Other Source and Sink Types

Let’s try something a bit more complicated and swap out the time source for something else. Another supported source type is http, which accepts data for ingestion over HTTP POSTs. Note that the http source accepts data on a different port (default 9000) from the Admin Server (default 8080).

To create a stream using an http source, but still using the same log sink, we would change the original command above to

xd:> stream create --definition "http | log" --name myhttpstream --deploy

which will produce the following output from the server

processing module 'Module [name=log, type=sink]' from group 'myhttpstream' with index: 1
processing module 'Module [name=http, type=source]' from group 'myhttpstream' with index: 0

Note that we don’t see any other output this time until we actually post some data (using shell command)

xd:> http post --target http://localhost:9000 --data "hello"
xd:> http post --target http://localhost:9000 --data "goodbye"

and the stream will then funnel the data from the http source to the output log implemented by the log sink

15:08:01,676  WARN ThreadPoolTaskScheduler-1 logger.myhttpstream:141 - hello
15:08:12,520  WARN ThreadPoolTaskScheduler-1 logger.myhttpstream:141 - goodbye

Of course, we could also change the sink implementation. You could pipe the output to a file (file), to hadoop (hdfs) or to any of the other sink modules which are provided. You can also define your own modules.

Simple Stream Processing

As an example of a simple processing step, we can transform the payload of the HTTP posted data to upper case using the stream definitions

http | transform --expression=payload.toUpperCase() | log

To create this stream enter the following command in the shell

xd:> stream create --definition "http | transform --expression=payload.toUpperCase() | log" --name myprocstrem --deploy

Posting some data (using shell command)

xd:> http post --target http://localhost:9000 --data "hello"

Will result in an uppercased hello in the log

15:18:21,345  WARN ThreadPoolTaskScheduler-1 logger.myprocstream:141 - HELLO

See the Processors section for more information.

DSL Syntax

In the examples above, we connected a source to a sink using the pipe symbol |. You can also pass parameters to the source and sink configurations. The parameter names will depend on the individual module implementations, but as an example, the http source module exposes a port setting which allows you to change the data ingestion port from the default value. To create the stream using port 8000, we would use

xd:> stream create --definition "http --port=8000 | log" --name myhttpstream

If you know a bit about Spring configuration files, you can inspect the module definition to see which properties it exposes. Alternatively, you can read more in the source and sink documentation.

Advanced Features

In the examples above, simple module definitions are used to construct each stream. However, modules may be grouped together in order to avoid duplication and/or reduce the amount of chattiness over the messaging middleware. To learn more about that feature, refer to the Composing Modules section.

If directed graphs are needed instead of the simple linear streams described above, two features are relevant. First, named channels may be used as a way to combine multiple flows upstream and/or downstream from the channel. The behavior of that channel may either be queue-based or topic-based depending on what prefix is used ("queue:myqueue" or "topic:mytopic", respectively). To learn more, refer to the Named Channels section. Second, you may need to determine the output channel of a stream based on some information that is only known at runtime. To learn about such content-based routing, refer to the Dynamic Router section.

Module Labels

When a stream is comprised of multiple modules with the same name, they must be qualified with labels. See Labels.

Modules

Introduction

Spring XD supports data ingestion by allowing users to define streams. Streams are composed of modules which encapsulate a unit of work into a reusable component. A job in Spring XD must also be implemented as a module.

Modules are categorized by type, typically representing the role or function of the module. Current Spring XD module types include source, sink, processor, and job. The type determines how the modules may be composed in a stream, or used to deploy a batch job. More precisely:

  • A source polls an external resource, or is triggered by an event and only provides output. The first module in a stream must be a source.

  • A processor performs some type of task, using a message as input and produces a new message, so it requires both input and output.

  • A sink consumes input messages and outputs data to an external resource to terminate the stream.

  • A job module implements a Spring Batch job enabled for Spring XD.

Available Modules

Spring XD ships with a number of pre-packaged modules ready to use for running batch jobs or assembling streams to perform common stream processing tasks using files, HDFS, Spark, Kafka, http, twitter, syslog, GemFire, and more. You can easily assemble these modules into streams to build complex big data applications declaratively, without having to write Java code or know the underlying Spring products on which Spring XD is built. You can use these modules out of the box or as a basis for building your own custom modules.

Modules Included with Spring XD

The following pages provide a detailed description, along with configuration options, and examples for all modules included in the Spring XD Distribution:

In addition to the standard modules included with the Spring XD distribution, you will find community-contributed modules in the spring-xd-modules GitHub repository.

If you are looking for a technical deep dive or want to develop your own modules, the sections below contain the relevant details. If you are interested in developing your own modules, some knowledge of Spring, Spring Integration or Spring Batch is essential. The remainder of this document assumes the reader has some familiarity with these topics.

Creating a Module

This section provides some general details on implementing and packaging custom modules. For a quick start, take a look at the si-dsl-module example, the community modules repository, or dive into the examples of creating source, processor, sink, and job modules.

Stream Modules

Sources, processors, and sinks are built using Spring Integration and are typically perform a single task that they may be easily reused in streams. Alternately, a custom module may be required to perform a specific function, such as integration with a legacy service. In Spring Integration terms:

  • A source is a valid message flow that contains a direct channel named output which is fed by an inbound adapter, either configured with a poller, or triggered by an event.

  • A processor is a valid message flow that contains a direct channel named input and a subscribable channel named output (direct or publish subscribe). It typically performs some type of transformation on the message, using its input channel’s message to create a new message on its output channel.

  • A sink is a valid message flow that contains a direct channel named input and an outbound adapter, or service activator used to provide the message to an external resource, HDFS for example.

For example, take a look at the file source which simply polls a directory using a file inbound adapter and file sink which appends an incoming message payload to a file using a file outbound adapter. On the surface, there is nothing special about these components. They are plain old Spring XML bean definition files.

Notice that modules adhere to an important convention: The input and output channels are always named input and output, in keeping with the KISS principle (let us know if you come up with some simpler names). The Spring XD runtime uses these names to bind these channels to the message transport.

Module Packaging

A module is a packaged component containing artifacts used to create a Spring application context. In general, a module is not aware of its runtime environment. Each module’s application context is configured and connected to other modules via Plugins in order to support distributed processing. In this respect, modules may potentially be applied to purposes other than stream processing. The module types described here (source, processor, sink, and job) are specific to Spring XD, but the Module type is designed to act as a core component of any micro-service architecture built with Spring.

Physically, a Module is somewhat analogous to a war file in Servlet container. The Spring XD container configures and starts a module when it is deployed. Deploying a module in Spring XD terms means activating an instance for processing, not to be confused with deploying a web application in Servlet container. Here we use the terms install or register to refer to uploading the module jar to make it available to the Spring XD runtime. Consistent with the war analogy, a module uses a separate class loader to load its resources, notably the files found in its config in lib directories. Another feature in common with a war file is that web applications are installed in a configured location and must conform to a standard layout. Artifacts are installed in a known location, either in expanded form or as a single archive. Spring XD modules work the same way. The module’s layout has evolved significantly as new features have been added to support custom module development. This evolution has generally led to increased flexibility with respect to individual artifacts. However, the module’s packaging structure is well defined:

<module_name>
      ├── <local class files and resources, e.g. com/acme/....>
      ├── config
      │   ├── <any-name>.properties
      │   └── <any-name>.[xml | groovy] (optional)
      ├── lib
      │   ├── <dependent libraries not provided by Spring XD (xd/lib)>
      │  

For historical reasons, all modules included with Spring XD distribution are provided in expanded form and are commonly configured using XML bean definition files (<module-name>.xml) and property files (<module-name>.properties>. This is subject to change as this convention is no longer required. Meanwhile the out-of-the-box modules provide copious examples of module configuration and packaging.

A module’s contents typically includes:

  • Application context configuration: If either config/<any_name>.xml or config/<any_name>.groovy are present, it will be used as the source for the module’s application context. At most one of these may be present. If using an @Configuration class, neither of these files should be present.

  • Module properties file: If the module declares options (e.g. property placeholders whose values must be supplied for each instance when creating a stream), the properties file config/<any_name>.properties may contain an options_class property referencing the fully qualified class name of a Module Options Metadata class. Alternately the properties file may provide in-line Module Option descriptors (see Module Options below). If using @Configuration, the properties file must include a base_packages property containing a comma delimited list of package names to enable Spring component scanning scoped to the module. Note that base_packages will be ignored if a configuration resource (config/*.xml or config/*.groovy) is present.

Note

As of Spring XD 1.1, the names of the module’s bean definition resource (xml or groovy) and properties file are arbitrary. This provides additional flexibilty over requiring a conventional file name, as has been the case in prior releases. Currently, the required top level config directory is the convention. This carries the constraint that no other matching file types may be present in config. Multiple xml, groovy, or properties files matching the pattern, for example, config/*.xml will result in an exception. If you want to combine bean definitions from multiple resources, you may use import declarations and the imported resources must be somewhere else in the module’s class path. This may be a subdirectory of config or any other arbitrary location.

  • Custom code: Any root level .class files packaged as in a typical jar file. This may include an @Configuration class, a Module Options Metadata class, and any dependent classes required by the module.

  • Dependent jar files: Any required runtime dependencies that are not provided by the Spring XD runtime (in $XD_INSTALL_DIR/xd/lib) are loaded from the module’s /lib directory.

As mentioned previously, a Spring XD module can be installed as an expanded directory tree or an archive. If the module requires dependent jars, which is the typical case, it may be packaged as an uber jar compatible with the Spring Boot layout, and conforming to the above structure. The next section describes Spring XD’s support for module packaging and development.

Creating a Module Project

Spring XD (1.1.x or later) provides build tools for creating a module project to test and package the module either with Maven or Gradle. As described in the above sections, the module jar must export any dependencies that are not provided by the Spring XD container. The build tools address these concerns, packaging your module as an uber jar by wrapping the Spring Boot Maven Plugin or the Spring Boot Gradle Plugin, respectively. The plugins are configured with The MODULE layout for Spring Boot packaging. This does not build an executable jar file, as is normally done, and ensures provided dependencies will not be included in the uber jar.

In addition, the build tools provide Spring XD dependencies necessary to compile and test the module. Specifically, spring-xd-dirt and spring-xd-test provide some useful features for module development. As you would expect, the Spring XD versions match the specified parent pom or plugin version. These provide support for:

  • Java defined Module Options Metadata

  • In-container testing - You can start an embedded single node container in a test class, create a stream designed to test your module using a Spring XD test framework, deploy the stream, and validate the results.

Note

If your module has no additional dependencies, a plain old jar file conforming to the module layout shown above will work. In this case, you may still benefit from using the build tools to simplify development and testing.

Module dependency management

Normally a module should export only the dependencies not provided by the Spring XD runtime. Runtime dependencies provided by the module are loaded using a separate module class loader when the module is deployed. This potentially can cause class version conflicts. Spring XD build tools are designed to prevent this and allow you to override the default exclusion rules if necessary. Generally we don’t recommend this unless you have a specific requirement for an alternate version. If your module introduces version conflicts, you will see errors such as ClassDefNotFoundException or NoSuchMethodError when you deploy the module. If you encounter such errors, you should manually check the contents of $XD_INSTALL_DIR/xd/lib against the contents of the module jar, or use the dependency analysis tools provided by Maven or Gradle, and make the necessary changes to your build script.

Porting to another Spring XD version

A module project’s build script is configured for a specific Spring XD version. With each new release of Spring XD, its runtime dependencies are subject to change and this directly affects which dependencies will be exported to the module jar. Deploying an existing module to a different Spring XD runtime version may result in version conflicts or unsatisfied dependencies. For this reason, we highly recommend that you rebuild any custom modules to match your target runtime environment. In many cases, this is simply a matter of updating the spring XD version in your build.

The details of dependency management depend on which build tool you are using. The sections on Building with Maven or Building with Gradle provide more specifics.

Building with Maven

Start by setting the parent to spring-xd-module-parent in your pom.xml:

<parent>
  <groupId>org.springframework.xd</groupId>
  <artifactId>spring-xd-module-parent</artifactId>
  <!-- 1.1.x or later -->
  <version>1.3.2.RELEASE</version>
</parent>
<dependencies>
...
</dependencies>
<repositories>
...
</repositories>
Note

Maven requires the parent pom version to be hard coded. Hence, you must edit the build script to target an alternate version of Spring XD.

To build the module:

$mvn clean package

The parent adds many of the transitive dependencies of spring-xd-dirt (provided) and spring-xd-test(test). Some transitive dependencies are excluded, such as Hadoop. Any easy way to determine what dependencies are included is to run a maven dependency goal, e.g.:

$mvn dependency:list

Any provided dependencies need not be declared as a module dependency. In any case, they will be excluded from the module jar by default.

If you must provide an alternate version of an existing Spring XD dependency, configure the Boot plugin explicitly in your pom to override the default, for example:

<parent>
  <groupId>org.springframework.xd</groupId>
  <artifactId>spring-xd-module-parent</artifactId>
  <version>...</version>
</parent>
<build>
  <plugins>
    <plugin>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-maven-plugin</artifactId>
        <configuration>
<!-- this is required to force the includes to come after the excludes and override -->
            <excludes>
                <exclude/>
            </excludes>
<!-- specify exactly what is included; transitive dependencies are not considered here -->
            <includes>
              <include>
                <groupId>xmlpull</groupId>
                <artifactId>xmlpull</artifactId>
              </include>
            </includes>
          </configuration>
        </plugin>
      </plugins>
     </build>

    <dependencies>
        <dependency>
          <groupId>xmlpull</groupId>
          <artifactId>xmlpull</artifactId>
          <version>1.1.3.4d_b4_min</version>
        </dependency>
    </dependencies>

Building with Gradle

Start by creating a gradle.properties file defining springXdVersion.

springXdVersion = [the Spring XD version]

This property is required by the spring-xd-module plugin to resolve Spring XD dependencies, and to configure dependent libraries that your module project will need. This property should be used to reference the Spring XD version where needed.

Note

Defining this value in gradle.properties helps with dependency-management when porting to other Spring XD versions. We also recommend adding any additional version references here.

Next, add the following to your build.gradle script:

buildscript {
    repositories {
     ...
    }
    // Add the path of the Spring XD Module plugin
    dependencies {
      classpath("org.springframework.xd:spring-xd-module-plugin:${springXdVersion}")
    }
}

apply plugin: 'spring-xd-module'

The plugin adds many of the transitive dependencies of spring-xd-dirt (provided) and spring-xd-test(test). Some transitive dependencies are excluded, such as Hadoop. Any easy way to determine what dependencies are included is to run one of Gradle’s dependency tools, e.g.:

$./gradlew dependencies

To build the module:

$./gradlew clean build

This configuration allows you to override springXdVersion on the command line:

$./gradlew clean build -PspringXdVersion=1.3.2.RELEASE
Note

Overriding the property on the Gradle command line does not work if the springXdVersion is hard coded in build.gradle itself, e.g., in an ext closure.

If you must provide an alternate version of an existing Spring XD dependency, override the exported configuration and the configureModule task in build.gradle, for example:

dependencies {
    ...
    exported "org.springframework.integration:spring-integration-xml:${springIntegrationVersion}"
    exported "org.springframework.ws:spring-xml:${springWsVersion}"
}

task configureModule(overwrite:true){
  configurations {
    exported {
        transitive = false
        exclude group:'org.springframework.xd'
    }
  }
}

Testing a Module Project

The sections Creating a Source Module, Creating a Processor Module, Creating a Sink Module, and Creating a Job Module each reference working examples of custom module projects including in-container tests.

See the si-dsl-module example for a complete working example.

Registering a Module

Registering a module requires you to install to the Spring XD Module Registry. A Module must be registered before it may be deployed as part of a stream or job. Once you have packaged your module, following the instructions in the above section, you can register it using the Spring XD Shell module upload command:

xd:>module upload --file [path-to]/mymodule-1.0.0.BUILD-SNAPSHOT.jar --name mymodule --type processor
Note

By default, the module upload command will fail if a module with the same name and type already exists. You may override this using the --force option. This is especially useful during development.

The Module Registry

A module definition requires the following attributes to uniquely define a module:

  • name - the name of the component, normally a single word representing the purpose of the module. Examples are file, http, syslog.

  • type - the module type, current Spring XD module types include source, sink, processor, and job

All modules included with Spring XD out-of-the-box are located in the xd/modules directory where Spring XD is installed. The Module Registry organizes modules by type in corresponding sub-directories, so a directory listing will look something like:

modules
      ├── job
      ├── processor
      ├── sink
      ├── source

Spring XD provides a strategy interface ModuleRegistry used to locate a module of a given name and type. Currently Spring XD implements a ResourceModuleRegistry which is configured to locate modules in the following locations in this order:

  • The file path given by xd.module.home (${xd.home}/modules by default)

  • classpath:/modules/ (Spring XD does not provide any module definitions here)

  • The file path given by xd.customModule.home (${xd.home}/custom-modules by default)

Custom Module Registry

Custom modules are located separately from out-of-the-box modules. The location is given by xd.customModule.home in servers.yml. The location defaults to ${xd.home}/custom-modules but we strongly recommend setting this to an external location on a network file system or using the replicating registry if you are using custom modules in production. There are two reasons for doing this. First, custom modules must be accessible to all nodes on the Spring XD cluster, including the XD Admin node. This allows any container instance to deploy the module. Second, if custom modules are registered within the Spring XD installation, they will not survive an upgrade to the Spring XD distribution and will need to be reinstalled. By default Spring XD expects MD5 hash files to be present next to the custom module jar. This is done to ensure the module upload has completed successfully before the module is used. Hash files are created automatically when installing modules vie the module upload command. If you wish to disable this requirement, set xd.customModule.requiresHashFiles to false in servers.yml. This will allow you to manually copy module jars to a file based custom module registry or reuse an existing custom module registry that may not include hash files. This setting does not apply to hdfs registries.

Note

An alternative way for specifying the location of custom modules via servers.yml is using the environment variable XD_CUSTOMMODULE_HOME that must point to the custom modules location.

In cases where you want to start e.g. a single-node runtime with a custom module location you can also define the environment variable right before the executable like this:
XD_CUSTOMMODULE_HOME=file\:/path/to/custom-modules bin/xd-singlenode

If you manually deploy your custom modules to XD_CUSTOMMODULE_HOME, since Spring XD (1.2.x or later), you will need to calculate md5 md5 -q mymodule.jar > mymodule.jar.md5 and put it together with your module. Otherwise the module will not be loaded and displayed using Spring XD shell command module list

custom-modules
      ├── job
      ├── processor
            ├── mymodule.jar
            ├── mymodule.jar.md5
      ├── sink
      ├── source

Replicating Module Registry

When running in distributed mode, an alternative to using a shared file system for custom modules is to use the replicating module registry.

If the value of xd.customModule.home does not use the file: protocol, then Spring XD will automatically set up a replicating registry that proxies that remote registry to the local filesystem. This is all done transparently and by default, files are copied down from the central repository only if their contents has changed.

At the time of writing, only the hdfs: protocol is supported. Setting this up is straightforward:

xd:
  customModule:
    home: hdfs://somehost/root/path/of/registry

Files will be replicated on the local filesystem in a temporary directory, on demand and loaded from there. The XD Admin process will need to have write access to that shared HDFS directory. Intermediary paths (/root/path/of/registry in the example above) are created at startup if they don’t exist yet.

Module Class Loading

Modules use a separate class loader that will first load classes from jars in the module’s /lib (and any class files located in the module’s root path). If not found, the class will be loaded from the parent ClassLoader that Spring XD normally uses (which includes everything under $XD_HOME/lib). Still, there are a couple of caveats to be aware of:

  • Avoid putting into the module’s lib/ directory any jar files that are already in Spring XD’s class path or you may end up with ClassCastExceptions or other class loading issues.

  • When not using local transport, any class that is directly or indirectly referenced from the payload type of your messages (i.e. any type in transit from module to module) must be referenced by both the producing and consuming modules and thus should be installed into xd/lib.

  • Occasionally, a class’s dependencies are not resolved correctly even though all the required jars appear to be on the module classpath. Consider a scenario in which class A depends on class B, and B depends on class C. If A and C are visible to the module class loader but only B is visible to the parent class loader, then you will get a ClassDefNotFoundException for class C if it has not already been loaded, because the parent class loader cannot resolve C. Unfortunately, an automated strategy to resolve this situation is difficult. A workaround is to install the jar containing class C into xd/lib.

Dynamic Module ClassLoader

Starting with Spring XD 1.2, a module can selectively add libraries from paths that are derived from module options. The aim is to support alternate implementations in the same module. This works like the following:

  1. In the module .properties file, specify a value for the module.classloader key. The default is /lib/*.jar,/lib/*.zip, which is consistent with what has been exposed earlier.

  2. The value for that key is a comma separated list of paths (most certainly with Ant-style patterns) that will be searched for additional libraries to add to the module ClassLoader (in addition to the module "Archive" itself, which is always included).

    1. paths that start with a / (as /lib/*.jar in the example above) are considered internal resources to the archives (e.g. nested jars in the über-jar)

    2. paths that do not start with a / (and in particular paths that start with a protocol, such as file:) are loaded with a regular Spring resource pattern resolver

  3. Those paths can contain placeholders of the form ${foo}. Those will be resolved against the visible module options (and other inherited properties). Paths containing unresolvable placeholders are silently ignored.

This allows constructions like this (assuming for example that we want to create a jpa module that supports several JPA providers):

jpa.jar
  +- config/
  |    jpa.properties
  +- lib/
       +- hibernate/
       |    hibernate-core-4.2.jar
       |    other-hibernate-specific.jar
       +- eclipse-link/
       |    eclipse-link-2.5.0.jar
       |    ...
       +- some-common.jar
       +- another-common.jar

And, in jpa.properties:

options_class = com.acme.jpa.JpaOptionsMetadata

module.classloader = /lib/*.jar, /lib/${provider}/*.jar, ${xd.home}/lib/jpa/${provider}/*.jpa

Where the metadata class includes a provider option (of type String) that will take e.g. the values hibernate or eclipse-link. Note the presence of a third ${xd.home}/lib/jpa/${provider}/*.jpa entry that can be used for unforeseen provider implementations.

Module Options

Each module instance is configured using property placeholders which are bound to the module’s options defined via Module Options Metadata. Options may be required or optional, where optional properties must provide a default value. Module Options Metadata may be provided within the module’s properties file or in a Java class provided by the module or one of its dependencies. In addition to binding module options to properties in the module’s application context, options may also be used to activate Spring environment profiles.

For example, here is part of the Spring configuration for the twittersearch source that runs a query against Twitter:

<beans>

  <bean class="org.springframework.integration.x.twitter.TwitterSearchChannelAdapter">
    <constructor-arg ref="twitterTemplate"/>
    <property name="readTimeout" value="${readTimeout}"/>
    <property name="connectTimeout" value="${connectTimeout}"/>
    <property name="autoStartup" value="false"/>
    <property name="outputChannel" ref="output"/>
    <property name="query" value="${query}" />
    <property name="language" value="${language}" />
    <property name="geocode" value="${geocode}" />
    <property name="resultType" value="${resultType}"/>
    <property name="includeEntities" value="${includeEntities}"/>
  </bean>

  <bean id="twitterTemplate" class="org.springframework.social.twitter.api.impl.TwitterTemplate">
    <constructor-arg value="${consumerKey}"/>
    <constructor-arg value="${consumerSecret}"/>
  </bean>

  <int:channel id="output"/>

</beans>

Note the Spring properties such as query, language, consumerKey and consumerSecret. Spring XD will bind values for all of these properties as provided as options for each module instance. The options exposed for this module are defined in TwitterSearchOptionsMetadata.java

For example, we can create two different streams, each using the twittersearch source providing different option values.

xd:> stream create --name tweettest --definition "twittersearch --query='java' | file"

and

xd:> stream create --name tweettest2 --definition "twittersearch --query='spring' --language=en --consumerKey='mykey' --consumerSecret='mysecret' | file"

In addition to options, modules may reference Spring beans such that each module instance may inject a different implementation of a bean. The ability to deploy the same module definition with different configurations is only possible because each module is created in its own application context. This results in some very useful features, such as the ability to use standard bean ids such as input and output and simple property names without having to worry about naming collisions.

Observe the use of property placeholders with sensible defaults where possible in the above example. Sometimes, a sensible default is derived from the stream name, module name, or some other runtime context. For example, the file source requires a directory. An appropriate strategy is to define a common root path for XD input files (At the time of this writing it is /tmp/xd/input/. This is subject to change, but illustrates the point). A stream definition using the file source may specify the the directory name by providing a value for the dir option. If not provided, it will default to the stream name, which is contained in the xd.stream.name property bound to the module by the Spring XD runtime, see file source metadata. The module info command illustrates this point:

xd:>module info --name source:file
Information about source module 'file':

  Option Name        Description                                                                  Default                          Type
  -----------------  ---------------------------------------------------------------------------  -------------------------------  --------
  dir                the absolute path to the directory to monitor for files                      /tmp/xd/input/${xd.stream.name}  String
  pattern            a filter expression (Ant style) to accept only files that match the pattern  *                                String
  preventDuplicates  whether to prevent the same file from being processed twice                  true                             boolean
  ref                set to true to output the File object itself                                 false                            boolean
  fixedDelay         the fixed delay polling interval specified in seconds                        5                                int
  outputType         how this module should emit messages it produces                             <none>                           MimeType

Placeholders available to all modules

By convention, Spring XD defined properties are prefixed with xd. Below is the list of all available ${xd.xxx} keys that module authors may use in their declaration.

Placeholder Context Meaning

${xd.stream.name}

streams

the name of the stream the module lives in

${xd.job.name}

jobs

the name of the job the module lives in

${xd.module.name}

streams, jobs

the technical name of the module

${xd.module.type}

streams, jobs

the type of the module

${xd.module.index}

streams

the 0-based position of the module inside the stream

${xd.container.id}

streams, jobs

the generated unique id of the container the module is deployed in

${xd.container.host}

streams, jobs

the hostname of the container the module is deployed in

${xd.container.pid}

streams, jobs

the process id of the container the module is deployed in

${xd.container.ip}

streams, jobs

the IP address of the container the module is deployed in

${xd.container.<foo>}

streams, jobs

the value of the custom attribute <foo> for the container

Note
Using placeholders in stream definitions

One can also use the ${xd.xxx} notation directly inside the DSL definition of a stream or a job. For example:

xd:>stream create foo --definition "http | filter --expression=\"'${xd.stream.name}'\" | log"

will only let messages that read "foo" pass through.

How module options are resolved

As we’ve seen so far, a module is a re-usable Spring Integration or Spring Batch application context that can be dynamically configured through the use of module options.

A module option is any value that the may be configured within a stream or job definition. Preferably, the module provides metadata to describe the available options. This section explains how default values are computed for each module option.

In a nutshell, actual values are resolved from the following sources, in order of precedence:

  1. values provided in the stream definition (e.g. --foo=bar)

  2. platform-wide defaults (appearing e.g. in .yml and .properties files, see below)

  3. defaults defined in the module’s metadata

Going into more detail, the platform-wide defaults will resolve like so, assuming option <optionname> of a module <modulename> which is of type <moduletype>:

  1. a system property named <moduletype>.<modulename>.<optionname>

  2. an environment variable named <moduletype>.<modulename>.<optionname> (or <MODULETYPE>_<MODULENAME>_<OPTIONNAME>)

  3. a key named <optionname> in the properties file <root>/<moduletype>/<modulename>/<modulename>.properties

  4. a key named <moduletype>.<modulename>.<optionname> in the YaML file <root>/<module-config>.yml

where

<root>

is the value of the xd.module.config.location system property (driven by the XD_MODULE_CONFIG_LOCATION env var when using the canonical Spring XD shell scripts). This property defaults to ${xd.config.home}/modules/

<module-config>

is the value of the xd.module.config.name system property (driven by the XD_MODULE_CONFIG_NAME env var). Defaults to xd-module-config

Note that YaML is particularly well suited for hierarchical configuration, so for example, instead of

source.file.dir: foo
source.file.pattern: *.txt

source.http.port: 1234

one can write

source:
  file:
    dir: foo
    pattern: *.txt
  http:
    port: 1234

Note that options in the .properties files can reference values that appear in the modules.yml file (this makes sharing common configuration easy). Also, the values that are used to configure the server runtimes (in servers.yml) are visible to modules.yml and .properties file (but the inverse is not true).

Composing Modules

As described above, a stream is defined as a sequence of modules, minimally a source module followed by a sink module. Sometimes streams may want share a common processing chain. For example, consider the following two streams:

stream1 = http | filter --expression=payload.contains('foo') | file
stream2 = file | filter --expression=payload.contains('foo') | file

Aside from the source, the two stream definitions are the same. Composite Modules provide a way to avoid this type of duplication by allowing the filter processor and file sink to be combined into a single composite module. Perhaps more importantly, composite modules may improve performance. Each module within a stream represents a unit of deployment. Therefore, stream1 and stream2, as defined above, are each comprised of three such units (a source, a processor, and a sink). In a singlenode runtime with local transport, creating a composite module won’t affect performance since the communication between modules in this case already uses in-memory channels. However, when deploying a stream to a distributed runtime environment, the communication between adjacent modules typically occurs via messaging middleware, as modules are, by default, distributed evenly among the available containers. Often a stream will perform better when adjacent modules are co-located and can avoid middleware "hops", and object marshalling. In such cases, composing modules allows the composite module to behave as a single "black box." In other words, if "foo | bar" are composed to create a new module named "baz", the input and/or output to "baz" will still go over the middleware, but foo and bar will be co-located in a single container instance and wired to communicate via local memory.

Working with Composite Modules

To create a composite module, use the module compose shell command:

xd:> module compose foo --definition "filter --expression=payload.contains('foo') | file"

Then, to verify the new module composition was successful, check if it exists:

xd:>module list
      Source              Processor           Sink                     Job
  ------------------  ------------------  -----------------------  ----------------
      file                aggregator          aggregate-counter        filejdbc
      gemfire             http-client         counter                  ftphdfs

                                   (....)

      trigger                                 splunk
      twittersearch                           tcp
      twitterstream                           throughput-sampler
      time                                (c) foo

Notice that the composed module shows up in the list of sink modules. That is because logically it acts as a sink: It provides an input channel (which is bridged to the filter processor’s input channel), but it provides no output channel (since the file sink has no output). Also notice that the module has a small (c) prefixed to it, to indicate that it is a composed module.

If a module were composed of two processors, it would be classified as a processor:

xd:> module compose myprocessor --definition "splitter | filter --expression=payload.contains('foo')"

If a module were composed of a source and a processor, it would be classified as a source:

xd:> module compose mysource --definition "http | filter --expression=payload.contains('foo')"

Based on the logical type of the composed module, it may be used in a stream as if it were a simple module instance. For example, to redefine the two streams from the first problem case above, now that the foo sink module has been composed, you can issue the following shell commands:

xd:> stream create httpfoo --definition "http | foo" --deploy
xd:> stream create filefoo --definition "file --outputType=text/plain | foo"  --deploy

To test the httpfoo stream, try the following:

xd:> http post --data hi
xd:> http post --data hifoo

The first message should have been ignored due to the filter, but the second one should exist in the file:

xd:> ! cat /tmp/xd/output/httpfoo.out
command is:cat /tmp/xd/output/httpfoo.out
hifoo

To test the filefoo stream, echo "foo" to a file in the /tmp/xd/input/filefoo directory, then verify:

xd:> ! cat /tmp/xd/output/filefoo.out
command is:cat /tmp/xd/output/filefoo.out
foo

When you no longer need a composed module, you may delete it with the module delete shell command. However, if that composed module is currently being used in one or more stream definitions, Spring XD will not allow you to delete it until those stream definitions are destroyed. In this case, module delete will fail as shown below:

xd:> module delete --name sink:foo
16:51:37,349  WARN Spring Shell client.RestTemplate:566 - DELETE request for "http://localhost:9393/modules/sink/foo" resulted in 500 (Internal Server Error); invoking error handler
Command failed org.springframework.xd.rest.client.impl.SpringXDException: Cannot delete module sink:foo because it is used by [stream:filefoo, stream:httpfoo]

As you can see, the failure message shows which stream(s) depend upon the composed module you are trying to delete.

If you destroy both of those streams and try again, it will work:

xd:> stream destroy --name filefoo
Destroyed stream 'filefoo'
xd:> stream destroy --name httpfoo
Destroyed stream 'httpfoo'
xd:> module delete --name sink:foo
Successfully destroyed module 'foo' with type sink

When creating a module, if you duplicate the name of an existing module for the same type, you will receive an error. In the example below the user tried to compose a tcp module, however one already exists:

xd:>module compose tcp --definition "filter --expression=payload.contains('foo') | file"
14:52:27,781  WARN Spring Shell client.RestTemplate:566 - POST request for "http://ec2-50-16-24-31.compute-1.amazonaws.com:9393/modules" resulted in 409 (Conflict); invoking error handler
Command failed org.springframework.xd.rest.client.impl.SpringXDException: There is already a module named 'tcp' with type 'sink'

However, you can create a module for a given type even though a module of that name exists but as a different type. For example: I can create a sink module named filter, even though filter already exists as a processor.

Finally, it’s worth mentioning that in some cases duplication may be avoided by reusing an actual stream rather than a composed module. This is possible when named channels are used in the source and/or sink position of a stream definition. For example, the same overall functionality as provided by the two streams above could also be achieved as follows:

xd:> stream create foofilteredfile --definition "queue:foo > filter --expression=payload.contains('foo') | file"
xd:> stream create httpfoo --definition "http > queue:foo"
xd:> stream create filefoo --definition "file > queue:foo"

This approach is more appropriate for use-cases where individual streams on either side of the named channel may need to be deployed or undeployed independently. Whereas the queue typed channel will load-balance across multiple downstream consumers, the topic: prefix may be used if broadcast behavior is needed instead. For more information about named channels, refer to the Named Channels section.

Getting Information about Modules

To view the available modules use the the module list command. Modules appearing with a (c) marker are composed modules. For example:

xd:>module list
      Source              Processor           Sink                     Job
  ------------------  ------------------  -----------------------  ----------------
      file                aggregator          aggregate-counter        filejdbc
      gemfire             analytic-pmml       counter                  ftphdfs
      gemfire-cq          http-client         field-value-counter      hdfsjdbc
      http                bridge              file                     hdfsmongodb
      jms                 filter              gauge                    jdbchdfs
      mail                json-to-tuple       gemfire-json-server      filepollhdfs
      mqtt                object-to-json      gemfire-server
      post                script              jdbc
      reactor-syslog      splitter            mail
      reactor-tcp         transform           mqtt
      syslog-tcp      (c) myfilter            rich-gauge
      syslog-udp                              splunk
      tail                                    tcp
      tcp                                     throughput-sampler
      tcp-client                              avro
      trigger                                 hdfs
      twittersearch                           log
      twitterstream                           rabbit
      rabbit                                  router
      time

To get information about a particular module (such as what options it accepts), use the module info --<module type>:<module name> command. For example:

xd:>module info --name source:file
Information about source module 'file':

  Option Name        Description                                                                  Default  Type
  -----------------  ---------------------------------------------------------------------------  -------  ---------
  dir                the absolute path to the directory to monitor for files                      <none>   String
  pattern            a filter expression (Ant style) to accept only files that match the pattern  *        String
  outputType         how this module should emit messages it produces                             <none>   MimeType
  preventDuplicates  whether to prevent the same file from being processed twice                  true     boolean
  ref                set to true to output the File object itself                                 false    boolean
  fixedDelay         the fixed delay polling interval specified in seconds                        5        int

Sources

This section describes the source modules included with Spring XD. A source implements a data provider to originate a stream. To run the examples shown here, start the XD Container as instructed in the Getting Started page.

The available Sources are

Future releases will provide support for other currently available Spring Integration Adapters. For information on how to adapt an existing Spring Integration Adapter for use in Spring XD see the section Creating a Source Module.

The following sections show a mix of Spring XD shell and plain Unix shell commands, so if you are trying them out, you should open two separate terminal prompts, one running the XD shell and one to enter the standard commands for sending HTTP data, creating directories, reading files and so on.

File

The file source provides the contents of a File as a byte array by default. However, this can be customized using the --mode option:

  • ref Provides a java.io.File reference

  • lines Will split files line-by-line and emit a new message for each line

  • contents The default. Provides the contents of a file as a byte array

When using --mode=lines, you can also provide the additional option --withMarkers=true. If set to true, the underlying FileSplitter will emit additional start-of-file and end-of-file marker messages before and after the actual data. The payload of these 2 additional marker messages is of type FileSplitter.FileMarker. The option withMarkers defaults to false if not explicitly set.

To log the contents of a file create a stream definition using the XD shell

xd:> stream create --name filetest --definition "file | log" --deploy

The file source by default will look into a directory named after the stream, in this case /tmp/xd/input/filetest

Note the above will log the raw bytes. For text files, it is normally desirable to output the contents as plain text. To do this, set the outputType parameter:

xd:> stream create --name filetest --definition "file --outputType=text/plain | log" --deploy

For more details on the use of the outputType parameter see Type Conversion

Copy a file into the directory /tmp/xd/input/filetest and observe its contents being logged in the XD Container.

File with options

The file source has the following options:

dir

the absolute path to the directory to monitor for files (String, default: /tmp/xd/input/<stream name>)

fixedDelay

the fixed delay polling interval specified in seconds (int, default: 5)

initialDelay

an initial delay when using a fixed delay trigger, expressed in TimeUnits (seconds by default) (int, default: 0)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: -1)

mode

specifies how the file is being read. By default the content of a file is provided as byte array (FileReadingMode, default: contents, possible values: ref,lines,contents)

pattern

a filter expression (Ant style) to accept only files that match the pattern (String, default: *)

preventDuplicates

whether to prevent the same file from being processed twice (boolean, default: true)

timeUnit

the time unit for the fixed and initial delays (String, default: SECONDS)

withMarkers

if true emits start of file/end of file marker messages before/after the data. Only valid with FileReadingMode 'lines' (Boolean, no default)

The ref option is useful in some cases in which the file contents are large and it would be more efficient to send the file path.

FTP

This source module supports transfer of files using the FTP protocol. Files are transferred from the remote directory to the local directory where the module is deployed. Messages emitted by the source are provided as a byte array by default. However, this can be customized using the --mode option:

  • ref Provides a java.io.File reference

  • lines Will split files line-by-line and emit a new message for each line

  • contents The default. Provides the contents of a file as a byte array

When using --mode=lines, you can also provide the additional option --withMarkers=true. If set to true, the underlying FileSplitter will emit additional start-of-file and end-of-file marker messages before and after the actual data. The payload of these 2 additional marker messages is of type FileSplitter.FileMarker. The option withMarkers defaults to false if not explicitly set.

Options

The ftp source has the following options:

autoCreateLocalDir

local directory must be auto created if it does not exist (boolean, default: true)

clientMode

client mode to use : 2 for passive mode and 0 for active mode (int, default: 0)

deleteRemoteFiles

delete remote files after transfer (boolean, default: false)

filenamePattern

simple filename pattern to apply to the filter (String, default: *)

fixedDelay

the rate at which to poll the remote directory (int, default: 1)

host

the host name for the FTP server (String, default: localhost)

initialDelay

an initial delay when using a fixed delay trigger, expressed in TimeUnits (seconds by default) (int, default: 0)

localDir

set the local directory the remote files are transferred to (String, default: /tmp/xd/ftp)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: -1)

mode

specifies how the file is being read. By default the content of a file is provided as byte array (FileReadingMode, default: contents, possible values: ref,lines,contents)

password

the password for the FTP connection (Password, no default)

port

the port for the FTP server (int, default: 21)

preserveTimestamp

whether to preserve the timestamp of files retrieved (boolean, default: true)

remoteDir

the remote directory to transfer the files from (String, default: /)

remoteFileSeparator

file separator to use on the remote side (String, default: /)

timeUnit

the time unit for the fixed and initial delays (String, default: SECONDS)

tmpFileSuffix

extension to use when downloading files (String, default: .tmp)

username

the username for the FTP connection (String, no default)

withMarkers

if true emits start of file/end of file marker messages before/after the data. Only valid with FileReadingMode 'lines' (Boolean, no default)

GemFire Continuous Query (gemfire-cq)

Continuous query allows client applications to create a GemFire query using Object Query Language(OQL) and register a CQ listener which subscribes to the query and is notified every time the query’s result set changes. The gemfire_cq source registers a CQ which will post CQEvent messages to the stream.

Options

The gemfire-cq source has the following options:

host

host name of the cache server or locator (if useLocator=true). May be a comma delimited list (String, no default)

port

port of the cache server or locator (if useLocator=true). May be a comma delimited list (String, no default)

query

the query string in Object Query Language (OQL) (String, no default)

useLocator

indicates whether a locator is used to access the cache server (boolean, default: false)

The example is similar to that presented for the gemfire source above, and requires an external cache server as described in the above section. In this case the query provides a finer filter on data events. In the example below, the cqtest stream will only receive events matching a single ticker symbol, whereas the gftest stream example above will receive updates to every entry in the region.

xd:> stream create --name stocks --definition "http --port=9090 | gemfire-json-server --regionName=Stocks --keyExpression=payload.getField('symbol')" --deploy
xd:> stream create --name cqtest --definition "gemfire-cq --query='Select * from /Stocks where symbol=''FAKE''' | file" --deploy

Now send some messages to the stocks stream.

xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":73}
xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":78}
xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":80}

The cqtest stream is now listening for any stock quote updates for the ticker symbol FAKE. As updates are posted to the cache you should see them captured in the output file:

$ cat /tmp/xd/output/cqtest.out

{"symbol":"FAKE","price":73}
{"symbol":"FAKE","price":78}
{"symbol":"FAKE","price":80}

GemFire Source (gemfire)

This source configures a client cache and client region, along with the necessary subscriptions enabled, in the XD container process along with a Spring Integration GemFire inbound channel adapter, backed by a CacheListener that outputs messages triggered by an external entry event on the region. By default the payload contains the updated entry value, but may be controlled by passing in a SpEL expression that uses the EntryEvent as the evaluation context.

Tip
If native gemfire properties are required to configure the client cache, e.g., for security, place a gemfire.properties file in $XD_HOME/config.

Options

The gemfire source has the following options:

cacheEventExpression

an optional SpEL expression referencing the event (String, default: newValue)

host

host name of the cache server or locator (if useLocator=true). May be a comma delimited list (String, no default)

port

port of the cache server or locator (if useLocator=true). May be a comma delimited list (String, no default)

regionName

the name of the region for which events are to be monitored (String, default: <stream name>)

useLocator

indicates whether a locator is used to access the cache server (boolean, default: false)

Example

Use of the gemfire source requires an external process (or a separate stream) that creates or updates entries in a GemFire region configured for a cache server. Such events may feed a Spring XD stream. To support such a stream, the Spring XD container must join a GemFire distributed client-server grid as a client, creating a client region corresponding to an existing region on a cache server. The client region registers a cache listener via the Spring Integration GemFire inbound channel adapter. The client region and pool are configured for a subscription on all keys in the region.

The following example creates two streams: One to write http messages to a Gemfire region named Stocks, and another to listen for cache events and record the updates to a file. This works with the Cache Server and sample configuration included with the Spring XD distribution:

xd:> stream create --name gftest --definition "gemfire --regionName=Stocks | file" --deploy
xd:> stream create --name stocks --definition "http --port=9090 | gemfire-json-server --regionName=Stocks --keyExpression=payload.getField('symbol')" --deploy

Now send some messages to the stocks stream.

xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":73}
xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":78}
xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":80}
Note
Avoid spaces in the JSON when using the shell to post data

As updates are posted to the cache you should see them captured in the output file:

$ cat /tmp/xd/output/gftest.out

{"symbol":"FAKE","price":73}
{"symbol":"FAKE","price":78}
{"symbol":"FAKE","price":80}
Note
The useLocator option is intended for integration with an existing GemFire installation in which the cache servers are configured to use locators in accordance with best practice. GemFire supports configuration of multiple locators (or direct server connections) and this is specified by supplying comma-delimited values for the host and port options. You may specify a single value for either of these options otherwise each value must contain the same size list. The following are examples are valid for multiple connection addresses:
gemfire --host=myhost --port=10334,10335
gemfire --host=myhost1,myhost2 --port=10334
gemfire --host=myhost1,myhost2,myhost3 --port=10334,10335,10336

The last example creates connections to myhost1:10334, myhost2:10335, myhost3:10336

Note
You may also configure default Gemfire connection settings for all gemfire modules in config/modules.yml:
gemfire:
   useLocator: true
   host: myhost1,myhost2
   port: 10334
Tip
If you are deploying on Java 7 or earlier and need to deploy more than 4 Gemfire modules be sure to increase the permsize of the singlenode or container. i.e. JAVA_OPTS="-XX:PermSize=256m"

Launching the XD GemFire Server

This source requires a cache server to be running in a separate process and its host and port, or a locator host and port must be configured. The XD distribution includes a GemFire server executable suitable for development and test purposes. This is a Java main class that runs with a Spring configured cache server. The configuration is passed as a command line argument to the server’s main method. The configuration includes a cache server port and one or more configured region. XD includes a sample cache configuration called cq-demo. This starts a server on port 40404 and creates a region named Stocks. A Logging cache listener is configured for the region to log region events.

Run Gemfire cache server by changing to the gemfire/bin directory and execute

$ ./gemfire-server ../config/cq-demo.xml

HTTP

To create a stream definition in the server using the XD shell

xd:> stream create --name httptest --definition "http | file" --deploy

Post some data to the http server on the default port of 9000

xd:> http post --target http://localhost:9000 --data "hello world"

See if the data ended up in the file

$ cat /tmp/xd/output/httptest

To send binary data, set the Content-Type header to application/octet-string

$ curl --data-binary @foo.zip -H'Content-Type: application-octet-string' http://localhost:9000

HTTP with options

The http source has the following options:

https

true for https:// (boolean, default: false)

keyStore

key store location (if sslPropertiesLocation not used) (String, no default)

keyStorePassphrase

key store passphrase (if sslPropertiesLocation not used) (String, no default)

maxContentLength

the maximum allowed content length (int, default: 1048576)

messageConverterClass

the name of a custom MessageConverter class, to convert HttpRequest to Message; must have a constructor with a 'MessageBuilderFactory' parameter (String, default: org.springframework.integration.x.http.NettyInboundMessageConverter)

port

the port to listen to (int, default: 9000)

sslPropertiesLocation

location (resource) of properties containing the location of the pkcs12 keyStore and pass phrase (String, no default)

Here is an example

xd:> stream create --name httptest9020 --definition "http --port=9020 | file" --deploy

Post some data to the new port

xd:> http post --target http://localhost:9020 --data "hello world"
$ cat /tmp/xd/output/httptest9020
hello world
Note
When using https, you may either provide a properties file that references a pkcs12 key store (containing the server certificate(s)) and its passphrase, or set keyStore and keyStorePassphrase explicitly. Setting --https=true enables https:// and the module uses SSL properties configured in config/modules/source/http/http.properties. By default, the resource classpath:httpSSL.properties is used. This location can be overridden in config/modules/source/http/http.properties or with the --sslPropertiesLocation property. For example:
xd:> stream create --name https9021 --definition "http --port=9021 --https=true --sslPropertiesLocation=file:/secret/ssl.properties | file" --deploy
$ cat /secret/ssl.properties
keyStore=file:/secret/httpSource.p12
keyStore.passPhrase=secret

Since this properties file contains sensitive information, it will typically be secured by the operating system with the XD container process having read access.

Note
If you set keyStore and keyStorePassphrase in config/modules/source/http/http.properties in lieue of using an external properties file, the passPhrase may be encrypted. See Encrypted Properties for more details.

JDBC Source (jdbc)

This source module supports the ability to ingest data directly from various databases. It does this by querying the database and sending the results as messages to the stream.

Configure a stream with a jdbc source using a query:

xd:> stream create foo --definition "jdbc --fixedDelay=1 --split=1 --url=jdbc:hsqldb:hsql://localhost:9101/mydb --query='select * from testfoo' |log" --deploy

In the example above the user will be polling the testfoo table to retrieve all the rows in the table once a second until the stream is undeployed or destroyed.

Configure a stream with a jdbc source using a query and update:

xd:> stream create foo --definition "jdbc --fixedDelay=1 --split=1 --url=jdbc:hsqldb:hsql://localhost:9101/mydb --query='select * from testfoo where tag = 0' --update='update testfoo set tag=1 where fooid in (:fooid)'|log" --deploy

In the example above the user will be polling the testfoo table to retrieve rows in the table that have a "tag" of zero. The update will set the value of tag to 1 for the rows that were retrieved, thus rows that have already been retrieved will not included in future queries.

Note
If you access any database other than HSQLDB or Postgres in a stream module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.

The jdbc source has the following options:

abandonWhenPercentageFull

connections that have timed out wont get closed and reported up unless the number of connections in use are above the percentage (int, default: 0)

alternateUsernameAllowed

uses an alternate user name if connection fails (boolean, default: false)

connectionProperties

connection properties that will be sent to our JDBC driver when establishing new connections (String, no default)

driverClassName

the JDBC driver to use (String, no default)

fairQueue

set to true if you wish that calls to getConnection should be treated fairly in a true FIFO fashion (boolean, default: true)

fixedDelay

how often to poll for new messages (s) (int, default: 5)

initSQL

custom query to be run when a connection is first created (String, no default)

initialSize

initial number of connections that are created when the pool is started (int, default: 0)

jdbcInterceptors

semicolon separated list of classnames extending org.apache.tomcat.jdbc.pool.JdbcInterceptor (String, no default)

jmxEnabled

register the pool with JMX or not (boolean, default: true)

logAbandoned

flag to log stack traces for application code which abandoned a Connection (boolean, default: false)

maxActive

maximum number of active connections that can be allocated from this pool at the same time (int, default: 100)

maxAge

time in milliseconds to keep this connection (int, default: 0)

maxIdle

maximum number of connections that should be kept in the pool at all times (int, default: 100)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: 1)

maxRowsPerPoll

max numbers of rows to process for each poll (int, default: 0)

maxWait

maximum number of milliseconds that the pool will wait for a connection (int, default: 30000)

minEvictableIdleTimeMillis

minimum amount of time an object may sit idle in the pool before it is eligible for eviction (int, default: 60000)

minIdle

minimum number of established connections that should be kept in the pool at all times (int, default: 10)

password

the JDBC password (Password, no default)

query

an SQL select query to execute to retrieve new messages when polling (String, no default)

removeAbandoned

flag to remove abandoned connections if they exceed the removeAbandonedTimout (boolean, default: false)

removeAbandonedTimeout

timeout in seconds before an abandoned connection can be removed (int, default: 60)

split

whether to split the SQL result as individual messages (boolean, default: true)

suspectTimeout

this simply logs the warning after timeout, connection remains (int, default: 0)

testOnBorrow

indication of whether objects will be validated before being borrowed from the pool (boolean, default: false)

testOnReturn

indication of whether objects will be validated before being returned to the pool (boolean, default: false)

testWhileIdle

indication of whether objects will be validated by the idle object evictor (boolean, default: false)

timeBetweenEvictionRunsMillis

number of milliseconds to sleep between runs of the idle connection validation/cleaner thread (int, default: 5000)

update

an SQL update statement to execute for marking polled messages as 'seen' (String, no default)

url

the JDBC URL for the database (String, no default)

useEquals

true if you wish the ProxyConnection class to use String.equals (boolean, default: true)

username

the JDBC username (String, no default)

validationInterval

avoid excess validation, only run validation at most at this frequency - time in milliseconds (long, default: 30000)

validationQuery

sql query that will be used to validate connections from this pool (String, no default)

validatorClassName

name of a class which implements the org.apache.tomcat.jdbc.pool.Validator (String, no default)

JMS

The "jms" source enables receiving messages from JMS.

The following example shows the default settings.

Configure a stream:

xd:> stream create --name jmstest --definition "jms | file" --deploy

This receives messages from a queue named jmstest and writes them to the default file sink (/tmp/xd/output/jmstest). It uses the default ActiveMQ broker running on localhost, port 61616.

To destroy the stream, enter the following at the shell prompt:

xd:> stream destroy --name jmstest

To test the above stream, you can use something like the following…​

public class Broker {

	public static void main(String[] args) throws Exception {
		BrokerService broker = new BrokerService();
		broker.setBrokerName("broker");
		String brokerURL = "tcp://localhost:61616";
		broker.addConnector(brokerURL);
		broker.start();
		ConnectionFactory cf = new ActiveMQConnectionFactory(brokerURL);
		JmsTemplate template = new JmsTemplate(cf);
		while (System.in.read() >= 0) {
			template.convertAndSend("jmstest", "testFoo");
		}
	}
}

and tail -f /tmp/xd/output/jmstest

Run this as a Java application; each time you hit <enter> in the console, it will send a message to queue jmstest.

The out of the box configuration is setup to use ActiveMQ. To use another JMS provider you will need to update a few files in the XD distribution. There are sample files for HornetMQ in the distribution as an example for you to follow. You will also need to add the appropriate libraries for your provider in the JMS module lib directory or in the main XD lib directory.

JMS with Options

The jms source has the following options:

acknowledge

the session acknowledge mode (String, default: auto)

clientId

an identifier for the client, to be associated with a durable topic subscription (String, no default)

destination

the destination name from which messages will be received (String, default: <stream name>)

durableSubscription

when true, indicates the subscription to a topic is durable (boolean, default: false)

provider

the JMS provider (String, default: activemq)

pubSub

when true, indicates that the destination is a topic (boolean, default: false)

subscriptionName

a name that will be assigned to the topic subscription (String, no default)

Note
the selected broker requires an infrastructure configuration file jms-<provider>-infrastructure-context.xml in modules/common. This is used to declare any infrastructure beans needed by the provider. See the default (jms-activemq-infrastructure-context.xml) for an example. Typically, all that is required is a ConnectionFactory. The activemq provider uses a properties file jms-activemq.properties which can be found in the config directory. This contains the broker URL.

Kafka

This source module ingests data from a single or comma separated list of Kafka topics. When using single topic configuration, one can also specify explicit partitions list and initial offset to fetch data from. Also note that for the stream with the given name or kafka source with the given groupId, the offsets for the configured topics aren’t deleted when the stream is undeployed/destroyed. This allows the re-deployed stream read from where it left when it was undeployed/destroyed.

The kafka source has the following options:

autoOffsetReset

strategy to reset the offset when there is no initial offset in ZK or if an offset is out of range (AutoOffsetResetStrategy, default: smallest, possible values: smallest,largest)

encoding

string encoder to translate bytes into string (String, default: UTF8)

fetchMaxBytes

max messages to attempt to fetch for each topic-partition in each fetch request (int, default: 1048576)

fetchMaxWait

max wait time before answering the fetch request (int, default: 100)

fetchMinBytes

the minimum amount of data the server should return for a fetch request (int, default: 1)

groupId

kafka consumer configuration group id (String, default: <stream name>)

initialOffsets

comma separated list of <partition>@<offset> pairs indicating where the source should start consuming from (String, default: ``)

kafkaOffsetTopicBatchBytes

maximum batched bytes for writes to offset topic, if Kafka offset strategy is chosen (int, default: 200)

kafkaOffsetTopicBatchTime

maximum time for batching writes to offset topic, if Kafka offset strategy is chosen (int, default: 1000)

kafkaOffsetTopicMaxSize

maximum size of reads from offset topic, if Kafka offset strategy is chosen (int, default: 1048576)

kafkaOffsetTopicName

name of the offset topic, if Kafka offset strategy is chosen (String, default: <stream name>-${xd.module.name}-offsets)

kafkaOffsetTopicRequiredAcks

required acks for writing to the Kafka offset topic, if Kafka offset strategy is chosen (int, default: 1)

kafkaOffsetTopicRetentionTime

retention time for dead records (tombstones), if Kafka offset strategy is chosen (int, default: 60000)

kafkaOffsetTopicSegmentSize

segment size of the offset topic, if Kafka offset strategy is chosen (int, default: 262144000)

offsetStorage

strategy for persisting offset values (OffsetStorageStrategy, default: kafka, possible values: inmemory,redis,kafka,kafkaNative)

offsetUpdateCount

frequency, in number of messages, with which offsets are persisted, per concurrent processor, mutually exclusive with the time-based offset update option (use 0 to disable either) (int, default: 0)

offsetUpdateShutdownTimeout

timeout for ensuring that all offsets have been written, on shutdown (int, default: 2000)

offsetUpdateTimeWindow

frequency (in milliseconds) with which offsets are persisted mutually exclusive with the count-based offset update option (use 0 to disable either) (int, default: 10000)

partitions

comma separated list of partition IDs to listen on (String, default: ``)

queueSize

the maximum number of messages held internally and waiting for processing, per concurrent handler. Value must be a power of 2 (int, default: 8192)

socketBufferBytes

socket receive buffer for network requests (int, default: 2097152)

socketTimeout

sock timeout for network requests in milliseconds (int, default: 30000)

streams

number of streams in the topic (int, default: 1)

topic

single topic name (String, default: ``)

topics

comma separated kafka topic names (String, default: ``)

zkconnect

zookeeper connect string (String, default: localhost:2181)

zkconnectionTimeout

the max time the client waits to connect to ZK in milliseconds (int, default: 6000)

zksessionTimeout

zookeeper session timeout in milliseconds (int, default: 6000)

zksyncTime

how far a ZK follower can be behind a ZK leader in milliseconds (int, default: 2000)

Configure a stream that has kafka source with a single topic:

xd:> stream create myKafkaSource1 --definition "kafka --zkconnect=localhost:2181 --topic=mytopic | log" --deploy

Configure a stream that has kafka source with a multiple topics:

xd:> stream create myKafkaSource2 --definition "kafka --zkconnect=localhost:2181 --topics=mytopic1,mytopic2 | log" --deploy

Mail

Spring XD provides a source module for receiving emails, named mail. Depending on the protocol used, in can work by polling or receive mails as they become available.

Let’s see an example:

xd:> stream create --name mailstream --definition "mail --host=imap.gmail.com --username=your.user%40gmail.com --password=password --port=993 | file" --deploy

One can also specify JavaMail properties as comma separated key=value pairs or as Spring supported resource URL location for the properties file.

xd:> stream create --name mailstream --definition "mail --host=imap.gmail.com --username=your.user%40gmail.com --password=password --port=993
      --properties=mail.debug=true | file" --deploy
xd:> stream create --name mailstream --definition "mail --host=imap.gmail.com --username=your.user%40gmail.com --password=password --port=993
--propertiesFile=file:/<path>/java-mail.properties | file" --deploy

If you are using imaps protocol, the mail source is configured to use these default properties:

mail.imap.socketFactory.class=javax.net.ssl.SSLSocketFactory
mail.imap.socketFactory.fallback=false
mail.store.protocol=imaps

Then send an email to yourself and you should see it appear inside a file at /tmp/xd/output/mailstream

Note: If the username/password have special characters like @, <space> then you need to enter appropriate unicode characters for them. For example the character @ can be specified with its unicode %40 as in the above definition.

The full list of options for the mail source is below:

The mail source has the following options:

charset

the charset used to transform the body of the incoming emails to Strings (String, default: UTF-8)

delete

whether to delete the emails once they’ve been fetched (boolean, default: true)

expression

a SpEL expression which filters which mail messages will be processed (non polling imap only) (String, default: true)

fixedDelay

the polling interval used for looking up messages (s) (int, default: 60)

folder

the folder to take emails from (String, default: INBOX)

host

the hostname of the mail server (String, default: localhost)

markAsRead

whether to mark emails as read once they’ve been fetched (boolean, default: false)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: 1)

password

the password to use to connect to the mail server (String, no default)

port

the port of the mail server (int, default: 25)

properties

comma separated JavaMail property values (String, no default)

propertiesFile

file to load the JavaMail properties (String, no default)

protocol

the protocol to use to retrieve messages (MailProtocol, default: imap, possible values: imap,imaps,pop3,pop3s)

usePolling

whether to use polling or not (no polling works with imap(s) only) (boolean, default: false)

username

the username to use to connect to the mail server (String, no default)

Warning
Of special attention are the markAsRead and delete options, which by default will delete the emails once they are consumed. It is hard to come up with a sensible default option for this (please refer to the Spring Integration documentation section on mail handling for a discussion about this), so just be aware that the default for XD is to delete incoming messages.

MongoDB Source (mongodb)

The MongoDB source allows one to query a MongoDB collection and emit messages for each and every matching result. This source works by regularly polling MongoDB and emitting the result list, as independent objects. If split is set to false, the whole list is emitted as payload.

Here is an example:

xd:> stream create foo --definition "mongodb --collectionName=orders --fixedDelay=1 | log" --deploy

The mongodb source has the following options:

authenticationDatabaseName

the MongoDB authentication database used for connecting (String, default: ``)

collectionName

the MongoDB collection to read from (String, default: <stream name>)

databaseName

the MongoDB database name (String, default: xd)

fixedDelay

the time delay between polls for data, expressed in TimeUnits (seconds by default) (int, default: 1000)

host

the MongoDB host to connect to (String, default: localhost)

initialDelay

an initial delay when using a fixed delay trigger, expressed in TimeUnits (seconds by default) (int, default: 0)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: 1)

password

the MongoDB password used for connecting (String, default: ``)

port

the MongoDB port to connect to (int, default: 27017)

query

the query to make to the mongo db (String, default: {})

split

whether to split the query result as individual messages (boolean, default: true)

timeUnit

the time unit for the fixed and initial delays (String, default: SECONDS)

username

the MongoDB username used for connecting (String, default: ``)

MQTT

The mqtt source connects to an mqtt server and receives telemetry messages.

Configure a stream:

xd:> stream create tcptest --definition "mqtt --url='tcp://localhost:1883' --topics='xd.mqtt.test' | log" --deploy

If you wish to use the MQTT Source defaults you can execute the command as follows:

xd:> stream create tcptest --definition "mqtt | log" --deploy

Options

The mqtt source has the following options:

binary

true to leave the payload as bytes (boolean, default: false)

charset

the charset used to convert bytes to String (when binary is false) (String, default: UTF-8)

cleanSession

whether the client and server should remember state across restarts and reconnects (boolean, default: true)

clientId

identifies the client (String, default: xd.mqtt.client.id.src)

connectionTimeout

the connection timeout in seconds (int, default: 30)

keepAliveInterval

the ping interval in seconds (int, default: 60)

password

the password to use when connecting to the broker (String, default: guest)

persistence

'memory' or 'file' (String, default: memory)

persistenceDirectory

file location when using 'file' persistence (String, default: /tmp/paho)

qos

the qos; a single value for all topics or a comma-delimited list to match the topics (String, default: 0)

topics

the topic(s) (comma-delimited) to which the source will subscribe (String, default: xd.mqtt.test)

url

location of the mqtt broker(s) (comma-delimited list) (String, default: tcp://localhost:1883)

username

the username to use when connecting to the broker (String, default: guest)

Note
The defaults are set up to connect to the RabbitMQ MQTT adapter on localhost.

RabbitMQ

The "rabbit" source enables receiving messages from RabbitMQ.

The following example shows the default settings.

Configure a stream:

xd:> stream create --name rabbittest --definition "rabbit | file --binary=true" --deploy

This receives messages from a queue named rabbittest and writes them to the default file sink (/tmp/xd/output/rabbittest.out). It uses the default RabbitMQ broker running on localhost, port 5672.

The queue(s) must exist before the stream is deployed. We do not create the queue(s) automatically. However, you can easily create a Queue using the RabbitMQ web UI. Then, using that same UI, you can navigate to the "rabbittest" Queue and publish test messages to it.

Notice that the file sink has --binary=true; this is because, by default, the data emitted by the source will be bytes. This can be modified by setting the content_type property on messages to text/plain. In that case, the source will convert the message to a String; you can then omit the --binary=true and the file sink will then append a newline after each message.

To destroy the stream, enter the following at the shell prompt:

xd:> stream destroy --name rabbittest

RabbitMQ with Options

The rabbit source has the following options:

ackMode

the acknowledge mode (AUTO, NONE, MANUAL) (String, default: AUTO)

addresses

a comma separated list of 'host[:port]' addresses (String, default: ${spring.rabbitmq.addresses})

concurrency

the minimum number of consumers (int, default: 1)

converterClass

the class name of the message converter (String, default: org.springframework.amqp.support.converter.SimpleMessageConverter)

enableRetry

enable retry; when retries are exhausted the message will be rejected; message disposition will depend on dead letter configuration (boolean, default: false)

initialRetryInterval

initial interval between retries (int, default: 1000)

keyStore

keyStore location (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.keyStore})

keyStorePassphrase

keyStore passphrase (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.keyStorePassphrase})

mappedRequestHeaders

request message header names to be propagated to/from the adpater/gateway (String, default: STANDARD_REQUEST_HEADERS)

maxAttempts

maximum delivery attempts (int, default: 3)

maxConcurrency

the maximum number of consumers (int, default: 1)

maxRetryInterval

maximum retry interval (int, default: 30000)

password

the password to use to connect to the broker (String, default: ${spring.rabbitmq.password})

prefetch

the prefetch size (int, default: 1)

queues

the queue(s) from which messages will be received (String, default: <stream name>)

requeue

whether rejected messages will be requeued by default (boolean, default: true)

retryMultiplier

retry interval multiplier (double, default: 2.0)

sslPropertiesLocation

resource containing SSL properties (String, default: ${spring.rabbitmq.sslProperties})

transacted

true if the channel is to be transacted (boolean, default: false)

trustStore

trustStore location (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.trustStore})

trustStorePassphrase

trustStore passphrase (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.trustStorePassphrase})

txSize

the number of messages to process before acking (int, default: 1)

useSSL

true if SSL should be used for the connection (String, default: ${spring.rabbitmq.useSSL})

username

the username to use to connect to the broker (String, default: ${spring.rabbitmq.username})

vhost

the RabbitMQ virtual host to use (String, default: ${spring.rabbitmq.virtual_host})

See the RabbitMQ MessageBus Documentation for more information about SSL configuration.

A Note About Retry

Note
With the default ackMode (AUTO) and requeue (true) options, failed message deliveries will be retried indefinitely. Since there is not much processing in the rabbit source, the risk of failure in the source itself is small. However, when using the LocalMessageBus or Direct Binding, exceptions in downstream modules will be thrown back to the source. Setting requeue to false will cause messages to be rejected on the first attempt (and possibly sent to a Dead Letter Exchange/Queue if the broker is so configured). The enableRetry option allows configuration of retry parameters such that a failed message delivery can be retried and eventually discarded (or dead-lettered) when retries are exhausted. The delivery thread is suspended during the retry interval(s). Retry options are enableRetry, maxAttempts, initialRetryInterval, retryMultiplier, and maxRetryInterval. Message deliveries failing with a MessageConversionException (perhaps when using a custom converterClassName) are never retried; the assumption being that if a message could not be converted on the first attempt, subsequent attempts will also fail. Such messages are discarded (or dead-lettered).

Reactor IP (reactor-ip)

The reactor-ip source acts as a server and allows a remote party to connect to XD and submit data over a raw TCP or UDP socket. The reactor-ip source differs from the standard tcp source in that it is based on the Reactor Project and can be configured to use the LMAX Disruptor RingBuffer library allowing for extremely high ingestion rates, e.g. ~ 1M/sec.

To create a stream definition use the following XD shell command

xd:> stream create --name tcpReactor --definition "reactor-ip | file" --deploy

This will create the reactor TCP source and send data read from it to the file named tcpReactor.

The reactor-ip source has the following options:

codec

codec used to transcode data (String, default: string)

dispatcher

type of Reactor Dispatcher to use (String, default: shared)

framing

method of framing the data (String, default: linefeed)

host

host to bind the server to (String, default: 0.0.0.0)

lengthFieldLength

byte precision of the number used in the length field (int, default: 4)

port

port to bind the server to (int, default: 3000)

transport

whether to use TCP or UDP as a transport protocol (String, no default)

SFTP

This source module supports transfer of files using the SFTP protocol. Files are transferred from the remote directory to the local directory where the module is deployed.

Messages emitted by the source are provided as a byte array by default. However, this can be customized using the --mode option:

  • ref Provides a java.io.File reference

  • lines Will split files line-by-line and emit a new message for each line

  • contents The default. Provides the contents of a file as a byte array

When using --mode=lines, you can also provide the additional option --withMarkers=true. If set to true, the underlying FileSplitter will emit additional start-of-file and end-of-file marker messages before and after the actual data. The payload of these 2 additional marker messages is of type FileSplitter.FileMarker. The option withMarkers defaults to false if not explicitly set.

Options

The sftp source has the following options:

allowUnknownKeys

true to allow connecting to a host with an unknown or changed key (boolean, default: false)

autoCreateLocalDir

if local directory must be auto created if it does not exist (boolean, default: true)

deleteRemoteFiles

delete remote files after transfer (boolean, default: false)

fixedDelay

fixed delay in SECONDS to poll the remote directory (int, default: 1)

host

the remote host to connect to (String, default: localhost)

initialDelay

an initial delay when using a fixed delay trigger, expressed in TimeUnits (seconds by default) (int, default: 0)

knownHostsExpression

a SpEL expresssion location of known hosts file; required if 'allowUnknownKeys' is false; examples: systemProperties["user.home"]+"/.ssh/known_hosts", "/foo/bar/known_hosts" (String, no default)

localDir

set the local directory the remote files are transferred to (String, default: /tmp/xd/output)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: -1)

mode

specifies how the file is being read. By default the content of a file is provided as byte array (FileReadingMode, default: contents, possible values: ref,lines,contents)

passPhrase

the passphrase to use (String, default: ``)

password

the password for the provided user (String, default: ``)

pattern

simple filename pattern to apply to the filter (String, no default)

port

the remote port to connect to (int, default: 22)

privateKey

the private key location (a valid Spring Resource URL) (String, default: ``)

regexPattern

filename regex pattern to apply to the filter (String, no default)

remoteDir

the remote directory to transfer the files from (String, no default)

timeUnit

the time unit for the fixed and initial delays (String, default: SECONDS)

tmpFileSuffix

extension to use when downloading files (String, default: .tmp)

user

the username to use (String, no default)

withMarkers

if true emits start of file/end of file marker messages before/after the data. Only valid with FileReadingMode 'lines' (Boolean, no default)

Stdout Capture

There isn’t actually a source named "stdin" but it is easy to capture stdin by redirecting it to a tcp source. For example if you wanted to capture the output of a command, you would first create the tcp stream, as above, using the appropriate sink for your requirements:

xd:> stream create tcpforstdout --definition "tcp --decoder=LF | log" --deploy

You can then capture the output from commands using the netcat command:

$ cat mylog.txt | netcat localhost 1234

Syslog

Three syslog sources are provided: reactor-syslog, syslog-udp, and syslog-tcp. The reactor-syslog adapter uses tcp and builds upon the functionality available in the Reactor project and provides improved throughput over the syslog-tcp adapter.

The reactor-syslog source has the following options:

port

the port on which the system will listen for syslog messages (int, default: 5140)

The syslog-udp source has the following options:

port

the port on which to listen (int, default: 5140)

rfc

the format of the syslog (String, default: 3164)

The syslog-tcp source has the following options:

nio

use nio (recommend false for a small number of senders, true for many) (boolean, default: false)

port

the port on which to listen (int, default: 5140)

rfc

the format of the syslog (String, default: 3164)

To create a stream definition (using shell command)

xd:> stream create --name syslogtest --definition "reactor-syslog --port=5140 | file" --deploy

or

xd:> stream create --name syslogtest --definition "syslog-udp --port=5140 | file" --deploy

or

xd:> stream create --name syslogtest --definition "syslog-tcp --port=5140 | file" --deploy

(--port is not required when using the default 5140)

Send a test message to the syslog

logger -p local3.info -t TESTING "Test Syslog Message"

See if the data ended up in the file

$ cat /tmp/xd/output/syslogtest

Refer to your syslog documentation to configure the syslog daemon to forward syslog messages to the stream; some examples are:

UDP - Mac OSX (syslog.conf) and Ubuntu (rsyslog.conf)

*.*    @localhost:5140

TCP - Ubuntu (rsyslog.conf)

$ModLoad omfwd
*.*    @@localhost:5140

Restart the syslog daemon after reconfiguring.

Tail

Make sure the default input directory exists

$ mkdir -p /tmp/xd/input

Create an empty file to tail (this is not needed on some platforms such as Linux)

$ touch /tmp/xd/input/tailtest

To create a stream definition using the XD shell

xd:> stream create --name tailtest --definition "tail | file" --deploy

Send some text into the file being monitored

$ echo blah >> /tmp/xd/input/tailtest

See if the data ended up in the file

$ cat /tmp/xd/output/tailtest

Tail with options

The tail source has the following options:

delay

how often (ms) to poll for new lines (forces use of the Apache Tailer, requires nativeOptions='') (long, no default)

fileDelay

on platforms that don't wait for a missing file to appear, how often (ms) to look for the file (long, default: 5000)

fromEnd

whether to tail from the end (true) or from the start (false) of the file (forces use of the Apache Tailer, requires nativeOptions='') (boolean, no default)

lines

the number of lines prior to the end of an existing file to tail; does not apply if 'nativeOptions' is provided (int, default: 0)

name

the absolute path of the file to tail (String, default: /tmp/xd/input/<stream name>)

nativeOptions

options for a native tail command; do not set and use 'end', 'delay', and/or 'reOpen' to use the Apache Tailer (String, no default)

reOpen

whether to reopen the file each time it is polled (forces use of the Apache Tailer, requires nativeOptions='') (boolean, no default)

Here is an example

xd:> stream create --name tailtest --definition "tail --name=/tmp/foo | file --name=bar" --deploy
$ echo blah >> /tmp/foo

$ cat /tmp/xd/output/bar

Tail Status Events

Some platforms, such as linux, send status messages to stderr. The tail module sends these events to a logging adapter, at WARN level; for example…​

[message=tail: cannot open `/tmp/xd/input/tailtest' for reading: No such file or directory, file=/tmp/xd/input/tailtest]
[message=tail: `/tmp/xd/input/tailtest' has become accessible, file=/tmp/xd/input/tailtest]

TCP

The tcp source acts as a server and allows a remote party to connect to XD and submit data over a raw tcp socket.

To create a stream definition in the server, use the following XD shell command

xd:> stream create --name tcptest --definition "tcp | file" --deploy

This will create the default TCP source and send data read from it to the tcptest file.

TCP is a streaming protocol and some mechanism is needed to frame messages on the wire. A number of decoders are available, the default being CRLF which is compatible with Telnet.

$ telnet localhost 1234
Trying ::1...
Connected to localhost.
Escape character is '^]'.
foo
^]

telnet> quit
Connection closed.

See if the data ended up in the file

$ cat /tmp/xd/output/tcptest

By default, the TCP module will emit a byte[]; to convert to a String, add --outputType=text/plain to the module definition.

TCP with options

The tcp source has the following options:

bufferSize

the size of the buffer (bytes) to use when encoding/decoding (int, default: 2048)

charset

the charset used when converting from bytes to String (String, default: UTF-8)

decoder

the decoder to use when receiving messages (Encoding, default: CRLF, possible values: CRLF,LF,NULL,STXETX,RAW,L1,L2,L4)

nio

whether or not to use NIO (boolean, default: false)

port

the port on which to listen (int, default: 1234)

reverseLookup

perform a reverse DNS lookup on the remote IP Address (boolean, default: false)

socketTimeout

the timeout (ms) before closing the socket when no data is received (int, default: 120000)

useDirectBuffers

whether or not to use direct buffers (boolean, default: false)

Available Decoders

Text Data
CRLF (default)

text terminated by carriage return (0x0d) followed by line feed (0x0a)

LF

text terminated by line feed (0x0a)

NULL

text terminated by a null byte (0x00)

STXETX

text preceded by an STX (0x02) and terminated by an ETX (0x03)

Text and Binary Data
RAW

no structure - the client indicates a complete message by closing the socket

L1

data preceded by a one byte (unsigned) length field (supports up to 255 bytes)

L2

data preceded by a two byte (unsigned) length field (up to 216-1 bytes)

L4

data preceded by a four byte (signed) length field (up to 231-1 bytes)

Examples

The following examples all use echo to send data to netcat which sends the data to the source.

The echo options -en allows echo to interpret escape sequences and not send a newline.

CRLF Decoder
xd:> stream create --name tcptest --definition "tcp | file" --deploy

This uses the default (CRLF) decoder and port 1234; send some data

$ echo -en 'foobar\r\n' | netcat localhost 1234

See if the data ended up in the file

$ cat /tmp/xd/output/tcptest
LF Decoder
xd:> stream create --name tcptest2 --definition "tcp --decoder=LF --port=1235 | file" --deploy
$ echo -en 'foobar\n' | netcat localhost 1235
$ cat /tmp/xd/output/tcptest2
NULL Decoder
xd:> stream create --name tcptest3 --definition "tcp --decoder=NULL --port=1236 | file" --deploy
$ echo -en 'foobar\x00' | netcat localhost 1236
$ cat /tmp/xd/output/tcptest3
STXETX Decoder
xd:> stream create --name tcptest4 --definition "tcp --decoder=STXETX --port=1237 | file" --deploy
$ echo -en '\x02foobar\x03' | netcat localhost 1237
$ cat /tmp/xd/output/tcptest4
RAW Decoder
xd:> stream create --name tcptest5 --definition "tcp --decoder=RAW --port=1238 | file" --deploy
$ echo -n 'foobar' | netcat localhost 1238
$ cat /tmp/xd/output/tcptest5
L1 Decoder
xd:> stream create --name tcptest6 --definition "tcp --decoder=L1 --port=1239 | file" --deploy
$ echo -en '\x06foobar' | netcat localhost 1239
$ cat /tmp/xd/output/tcptest6
L2 Decoder
xd:> stream create --name tcptest7 --definition "tcp --decoder=L2 --port=1240 | file" --deploy
$ echo -en '\x00\x06foobar' | netcat localhost 1240
$ cat /tmp/xd/output/tcptest7
L4 Decoder
xd:> stream create --name tcptest8 --definition "tcp --decoder=L4 --port=1241 | file" --deploy
$ echo -en '\x00\x00\x00\x06foobar' | netcat localhost 1241
$ cat /tmp/xd/output/tcptest8

Binary Data Example

xd:> stream create --name tcptest9 --definition "tcp --decoder=L1 --port=1242 | file --binary=true" --deploy

Note that we configure the file sink with binary=true so that a newline is not appended.

$ echo -en '\x08foo\x00bar\x0b' | netcat localhost 1242
$ hexdump -C /tmp/xd/output/tcptest9
00000000  66 6f 6f 00 62 61 72 0b                           |foo.bar.|
00000008

TCP Client (tcp-client)

The tcp-client source module uses raw tcp sockets, as does the tcp module but contrary to the tcp module, acts as a client. Whereas the tcp module will open a listening socket and wait for connections from a remote party, the tcp-client will initiate the connection to a remote server and emit as messages what that remote server sends over the wire. As an optional feature, the tcp-client can itself emit messages to the remote server, so that a simple conversation can take place.

TCP Client options

The tcp-client source has the following options:

bufferSize

the size of the buffer (bytes) to use when encoding/decoding (int, default: 2048)

charset

the charset used when converting from bytes to String (String, default: UTF-8)

close

whether to close the socket after each message (boolean, default: false)

decoder

the decoder to use when receiving messages (Encoding, default: CRLF, possible values: CRLF,LF,NULL,STXETX,RAW,L1,L2,L4)

encoder

the encoder to use when sending messages (Encoding, default: CRLF, possible values: CRLF,LF,NULL,STXETX,RAW,L1,L2,L4)

expression

a SpEL expression used to transform messages (String, default: payload.toString())

fixedDelay

the rate at which stimulus messages will be emitted (seconds) (int, default: 5)

host

the remote host to connect to (String, default: localhost)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: 1)

nio

whether or not to use NIO (boolean, default: false)

port

the port on the remote host to connect to (int, default: 1234)

propertiesLocation

the path of a properties file containing custom script variable bindings (String, no default)

refreshDelay

how often to check (in milliseconds) whether the script has changed; -1 for never (long, default: 60000)

reverseLookup

perform a reverse DNS lookup on the remote IP Address (boolean, default: false)

script

reference to a script used to process messages (String, no default)

socketTimeout

the timeout (ms) before closing the socket when no data is received (int, default: 120000)

useDirectBuffers

whether or not to use direct buffers (boolean, default: false)

variables

variable bindings as a comma delimited string of name-value pairs, e.g., 'foo=bar,baz=car' (String, no default)

Implementing a simple conversation

That "stimulus" counter concept bears some explanation. By default, the module will emit (at interval set by fixedDelay) an incrementing number, starting at 1. Given that the default is to use an expression of payload.toString(), this results in the module sending 1, 2, 3, ... to the remote server.

By using another expression, or more certainly a script, one can implement a simple conversation, assuming it is time based. As an example, let’s assume we want to join some kind of chat server where one first needs to authenticate, then specify which rooms to join. Lastly, all clients are supposed to send some keepalive commands to make sure that the connection is open.

The following groovy script could be used to that effect:

def commands = ['', // index 0 is not used
'LOGIN user=johndoe', // first command sent
'JOIN weather',
'JOIN news',
'JOIN gossip'
]


// payload will contain an incrementing counter, starting at 1
if (commands.size > payload)
  return commands[payload] + "\n"
else
  return "PING\n"  // send keep alive after 4th 'real' command
Tip
If the script option is set, the script file’s modified timestamp is checked for changes every 60 seconds by default; this can be changed with the refreshDelay deployment property: --refreshDelay=30000 (every 30 seconds or 30,000ms), --refreshDelay=-1 to disable refresh.

Time

The time source will simply emit a String with the current time every so often.

The time source has the following options:

fixedDelay

time delay between messages, expressed in TimeUnits (seconds by default) (int, default: 1)

format

how to render the current time, using SimpleDateFormat (String, default: yyyy-MM-dd HH:mm:ss)

initialDelay

an initial delay when using a fixed delay trigger, expressed in TimeUnits (seconds by default) (int, default: 0)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: 1)

timeUnit

the time unit for the fixed and initial delays (String, default: SECONDS)

Trigger Source (trigger)

The trigger source emits a message or messages according to the provided trigger configuration. The message payload is a simple literal value, provided in the payload property.

The trigger source has the following options:

cron

cron expression specifying when the trigger should fire (String, no default)

date

a one-time date when the trigger should fire; only applies if 'fixedDelay' and 'cron' are not provided (String, default: The current time)

dateFormat

the format specifying how the 'date' should be parsed (String, default: MM/dd/yy HH:mm:ss)

fixedDelay

time delay between executions, expressed in TimeUnits (seconds by default) (Integer, no default)

initialDelay

an initial delay when using a fixed delay trigger, expressed in TimeUnits (seconds by default) (int, default: 0)

maxMessages

the maximum messages per poll; -1 for unlimited (long, default: 1)

payload

the message that will be sent when the trigger fires (String, default: ``)

timeUnit

the time unit for the fixed and initial delays (String, default: SECONDS)

The twittersearch source runs a continuous query against Twitter.

The twittersearch source has the following options:

connectTimeout

the connection timeout for making a connection to Twitter (ms) (int, default: 5000)

consumerKey

a consumer key issued by twitter (String, no default)

consumerSecret

consumer secret corresponding to the consumer key (String, no default)

geocode

geo-location given as latitude,longitude,radius. e.g., '37.781157,-122.398720,1mi' (String, default: ``)

includeEntities

whether to include entities such as urls, media and hashtags (boolean, default: true)

language

language code e.g. 'en' (String, default: ``)

query

the query string (String, default: ``)

readTimeout

the read timeout for the underlying URLConnection to the twitter stream (ms) (int, default: 9000)

resultType

result type: recent, popular, or mixed (ResultType, default: mixed, possible values: mixed,recent,popular)

For information on how to construct a query, see the Search API v1.1.

To get a consumerKey and consumerSecret you need to register a twitter application. If you don’t already have one set up, you can create an app at the Twitter Developers site to get these credentials.

Tip
For both twittersearch and twitterstream you can put these keys in a module properties file instead of supplying them in the stream definition. If both sources share the same credentials, it is easiest to configure the required credentials in config/modules/modules.yml. Alternately, each module has its own properties file. For twittersearch, the file would be config/modules/source/twittersearch/twittersearch.properties.

To create and deploy a stream definition in the server using the XD shell:

xd:> stream create --name springone2gx --definition "twittersearch --query='#springone2gx' | file" --deploy

Let the twittersearch run for a little while and then check to see if some data ended up in the file

$ cat /tmp/xd/output/springone2gx
Note
Both twittersearch and twitterstream emit JSON in the native Twitter format.

Twitter Stream (twitterstream)

This source ingests data from Twitter’s streaming API v1.1. It uses the sample and filter stream endpoints rather than the full "firehose" which needs special access. The endpoint used will depend on the parameters you supply in the stream definition (some are specific to the filter endpoint).

You need to supply all keys and secrets (both consumer and accessToken) to authenticate for this source, so it is easiest if you just add these to XD_HOME/config/modules/modules.yml or XD_HOME/config/modules/source/twitterstream/twitterstream.properties file.

Stream creation is then straightforward:

xd:> stream create --name tweets --definition "twitterstream | file" --deploy

The twitterstream source has the following options:

accessToken

a valid OAuth access token (String, no default)

accessTokenSecret

an OAuth secret corresponding to the access token (String, no default)

connectTimeout

the connection timeout for making a connection to Twitter (ms) (int, default: 5000)

consumerKey

a consumer key issued by twitter (String, no default)

consumerSecret

consumer secret corresponding to the consumer key (String, no default)

delimited

set to true to get length delimiters in the stream data (boolean, default: false)

discardDeletes

set to discard 'delete' events (boolean, default: true)

filterLevel

controls which tweets make it through to the stream: none,low,or medium (FilterLevel, default: none, possible values: none,low,medium)

follow

comma delimited set of user ids whose tweets should be included in the stream (String, default: ``)

language

language code e.g. 'en' (String, default: ``)

locations

comma delimited set of latitude/longitude pairs to include in the stream (String, default: ``)

readTimeout

the read timeout for the underlying URLConnection to the twitter stream (ms) (int, default: 9000)

stallWarnings

set to true to enable stall warnings (boolean, default: false)

track

comma delimited set of terms to include in the stream (String, default: ``)

Note: The options available are pretty much the same as those listed in the Twitter API docs and unless otherwise stated, the accepted formats are the same.

Note
Both twittersearch and twitterstream emit JSON in the native Twitter format.

Processors

This section describes the processor modules included with Spring XD. A processor implements a processing task within a stream. A stream may chain multiple processors sequentially as needed. To run the examples shown here, start the XD Container as instructed in the Getting Started page.

The available Processors are

See the section Creating a Processor Module for information on how to create custom processor modules.

Aggregator

The aggregator module does the opposite of the splitter, and builds upon the concept of the same name found in Spring Integration. By default, it will consider all incoming messages from a stream to belong to the same group:

xd:> stream create --name aggregates --definition "http | aggregator --count=3 --aggregation=T(org.springframework.util.StringUtils).collectionToDelimitedString(#this.![payload],' ') | log" --deploy

This uses a SpEL expression that will basically concatenate all payloads together, inserting a space character in between. As such,

xd:> http post --data Hello
xd:> http post --data World
xd:> http post --data !

would emit a single message whose contents is "Hello World !". This is because we set the aggregator release strategy to accumulate 3 messages.

The aggregator processor has the following options:

aggregation

how to construct the aggregated message (SpEL expression against a collection of messages) (String, default: #this.![payload])

correlation

how to correlate messages (SpEL expression against each message) (String, default: '<stream name>')

count

the number of messages to group together before emitting a group (int, default: 50)

dbkind

which flavor of init scripts to use for the jdbc store (blank to attempt autodetection) (String, no default)

driverClassName

the jdbc driver to use when using the jdbc store (String, no default)

hostname

hostname of the redis instance to use as a store (String, default: localhost)

initializeDatabase

whether to auto-create the database tables for the jdbc store (boolean, default: false)

password

the password to use when using the jdbc or redis store (String, default: ``)

port

port of the redis instance to use as a store (int, default: 6379)

release

when to release messages (SpEL expression against a collection of messages accumulated so far) (String, no default)

store

the kind of store to use to retain messages (StoreKind, default: memory, possible values: memory,jdbc,redis)

timeout

the delay (ms) after which messages should be released, even if the completion criteria is not met (int, default: 50000)

url

the jdbc url to connect to when using the jdbc store (String, no default)

username

the username to use when using the jdbc store (String, no default)

Note
  • Some of the options are only relevant when using a particular store

  • The default correlation of '<stream name>' actually considers all messages to be correlated, since they all belong to the same stream.

  • Using the release option overrides the count option (which is a simpler approach)

  • The default for aggregation creates a new collection made of the payloads of the accumulated messages

  • About the timeout option: due to the way it is implemented (see MessageGroupStoreReaper in the Spring Integration documentation), the actual observed delay may vary between timeout and 2xtimeout.

Filter

Use the filter module in a stream to determine whether a Message should be passed to the output channel.

The filter processor has the following options:

expression

a SpEL expression used to transform messages (String, default: payload.toString())

propertiesLocation

the path of a properties file containing custom script variable bindings (String, no default)

refreshDelay

how often to check (in milliseconds) whether the script has changed; -1 for never (long, default: 60000)

script

reference to a script used to process messages (String, no default)

variables

variable bindings as a comma delimited string of name-value pairs, e.g., 'foo=bar,baz=car' (String, no default)

Filter with SpEL expression

The simplest way to use the filter processor is to pass a SpEL expression when creating the stream. The expression should evaluate the message and return true or false. For example:

xd:> stream create --name filtertest --definition "http | filter --expression=payload=='good' | log" --deploy

This filter will only pass Messages to the log sink if the payload is the word "good". Try sending "good" to the HTTP endpoint and you should see it in the XD log:

xd:> http post --target http://localhost:9000 --data "good"

Alternatively, if you send the word "bad" (or anything else), you shouldn’t see the log entry.

Filter using jsonPath evaluation

As part of the SpEL expression you can make use of the pre-registered JSON Path function.

This filter example shows to pass messages to the output channel if they contain a specific JSON field matching a specific value.

xd:> stream create --name jsonfiltertest --definition "http --port=9002 | filter --expression=#jsonPath(payload,'$.firstName').contains('John') | log" --deploy

Note: There is no space between payload JSON and the jsonPath in the expression

This filter will only pass Messages to the log sink if the JSON payload contains the firstName "John". Try sending this payload to the HTTP endpoint and you should see it in the XD log:

xd:> http post --target http://localhost:9002 --data "{\"firstName\":\"John\", \"lastName\":\"Smith\"}"

Alternatively, if you send a different firstName, you shouldn’t see the log entry.

Here is another example usage of filter

filter --expression=#jsonPath(payload,'$.entities.hashtags[*].text').contains('obama')

This is an example that is operating on a JSON payload of tweets as consumed from the twitter search module.

Filter with Groovy Script

For more complex filtering, you can pass the location of a Groovy script using the script option. If you want to pass variable values to your script, you can statically bind values using the variables option or optionally pass the path to a properties file containing the bindings using the propertiesLocation option.All properties in the file will be made available to the script as variables. Note that payload and headers are implicitly bound to give you access to the data contained in a message.

Example:

Note
These features are common to all modules backed by Groovy scripts.
//custom-filter.groovy
return payload.size()> 4 || shortstrings=='true'
#custom-filter.properties
shortstrings=false

By default, Spring XD will search the classpath for custom-filter.groovy and custom-filter.properties. You can place the script in ${xd.home}/modules/processor/scripts and the properties file in ${xd.home}/config to make them available on the classpath. Alternatively, you can prefix the script and properties-location values with file: to load from the file system.

In the following stream definitions, the filter will pass only the first message:

xd>: stream create --name groovyfiltertest1 --definition "http --port=9001 | filter --script=file:<absolute-path-to>/custom-filter.groovy --variables='shortstrings=false' | log" --deploy
Created and deployed new stream 'groovyfiltertest1'
xd:>http post --target http://localhost:9001 --data hello
xd:http post --target http://localhost:9001 --data hi
xd>: stream create --name groovyfiltertest2 --definition "http --port=9002 | filter --script=file:<absolute-path-to>/custom-filter.groovy --propertiesLocation=file:<absolute-path-to>/custom-filter.properties | log" --deploy
Created and deployed new stream 'groovyfiltertest2'
xd:>http post --target http://localhost:9002 --data hello
xd:http post --target http://localhost:9002 --data hi

In the following stream definitions, the filter will pass all messages (provided the payload type supports a size() method):

xd>: stream create --name groovyfiltertest1 --definition "http --port=9001 | filter --script=file:<absolute-path-to>/custom-filter.groovy --variables='shortstrings=false' | log" --deploy
Created and deployed new stream 'groovyfiltertest1'
xd>: stream create --name groovyfiltertest2 --definition "http --port=9002 | filter --script=file:<absolute-path-to>/custom-filter.groovy --variables='shortstring=false' --propertiesLocation=file:<absolute-path-to>/custom-filter.properties | log" --deploy
Created and deployed new stream 'groovyfiltertest2'

Note the last example demonstrates that values specified in variables override values from propertiesLocation

Tip
The script file’s modified timestamp is checked for changes every 60 seconds by default; this can be changed with the refreshDelay deployment property: --refreshDelay=30000 (every 30 seconds or 30,000ms), --refreshDelay=-1 to disable refresh.

Header Enricher (header-enricher)

The header-enricher processor provides a basic header enricher to allow a stream to add runtime state in one or more message headers. Message headers are preserved across the entire stream flow and may be referenced by down stream modules using SpEL, e.g., expression=headers['foo'].

Header expressions are provided using the headers module option which expects a JSON string.

Some Examples:

A Simple SpEL expression

xd:>stream create t1 --definition "http | header-enricher --headers={\"foo\":\"payload.toUpperCase()\"} | log --expression=headers" --deploy
xd:>http post --data hello --target http://localhost:9000

You should see the message headers, including foo=HELLO in the container console log:

2015-05-24T11:40:40-0400 1.2.0.SNAP INFO pool-12-thread-4 sink.t1 - {requestMethod=POST, foo=HELLO, User-Agent=Java/1.8.0_25, Host=localhost:9000, id=f6b2c420-cd12-2d5e-e0cb-675f60fb9a63, Content-Length=5, contentType=text/plain;Charset=UTF-8, requestPath=/, timestamp=1432482040939}

Multiple Headers: This will add 2 headers, foo and bar

xd:> stream create t2 --definition "http | header-enricher --headers={\"foo\":\"payload.toUpperCase()\",\"bar\":\"payload+',world'\"} | log --expression=headers" --deploy
xd:>http post --data hello --target http://localhost:9000

The result:

2015-05-24T11:49:19-0400 1.2.0.SNAP INFO pool-27-thread-4 sink.t2 - {bar=hello,world, requestMethod=POST, foo=HELLO, User-Agent=Java/1.8.0_25, Host=localhost:9000, id=c9220992-f71c-db3f-0f22-b5ab262d4aee, Content-Length=5, contentType=text/plain;Charset=UTF-8, requestPath=/, timestamp=1432482559740}

JSON Path expressions: This example uses a JSON Path expression to extract a field value in a JSON Payload:

xd:>stream create t3 --definition "http | header-enricher --headers={\"foo\":\"#jsonPath(payload,'$.duration')\"} | log --expression=headers" --deploy
xd:>http post --data {"duration":123.45} --target http://localhost:9000

foo=123.45 !!

2015-05-24T11:45:16-0400 1.2.0.SNAP INFO pool-23-thread-4 sink.t3 - {requestMethod=POST, foo=123.45, User-Agent=Java/1.8.0_25, Host=localhost:9000, id=87d229b9-a434-31c2-eb35-b42ca3f8352a, Content-Length=19, contentType=text/plain;Charset=UTF-8, requestPath=/, timestamp=1432482316080}

Literal Strings with Embedded Spaces: SpEL expects literals to be enclosed in single quotes. This is straight forward when the literal does not contain embedded spaces as in the Multple Headers example above. Using embedded spaces requires either wrapping the headers value in single quotes and escaping the single quote for the literal string

xd:>stream create t4 --definition "http | header-enricher --headers='{\"foo\":\"''this is a literal string''\"}' | log --expression=headers" --deploy

or you can encode embedded spaces using the Unicode escape sequence (the leading \ must itself be escaped):

xd:>stream create t4 --definition "http | header-enricher --headers={\"foo\":\"'this\\u0020is\\u0020a\\u0020literal\\u0020string'\"} | log --expression=headers" --deploy

the result:

2015-05-24T12:26:32-0400 1.2.0.SNAP INFO pool-31-thread-4 sink.t4 - {requestMethod=POST, foo=this is a literal string, User-Agent=Java/1.8.0_25, Host=localhost:9000, id=b8892a1b-57b1-de6e-71de-e30f84cd199a, Content-Length=5, contentType=text/plain;Charset=UTF-8, requestPath=/, timestamp=1432484792949}

The header-enricher processor has the following options:

headers

a JSON document representing headers in which values are SpEL expressions, e.g {"h1":"exp1","h2":"exp2"} (String, no default)

overwrite

set to true to overwrite any existing message headers (Boolean, default: false)

HTTP Client (http-client)

The http-client processor acts as a client that issues HTTP requests to a remote server, submitting the message payload it receices to that server and in turn emitting the response it receives to the next module down the line.

For example, the following command will result in an immediate fetching of earthquake data and it being logged in the container:

xd:>stream create earthquakes --definition "trigger | http-client --url='''http://earthquake.usgs.gov/earthquakes/feed/geojson/all/day''' --httpMethod=GET | log" --deploy
Note

Please be aware that the url option above is actually a SpEL expression, hence the triple quotes. If you’d like to learn more about quotes, please read the relevant documentation.

The http-client processor has the following options:

charset

the charset to use when in the Content-Type header when emitting Strings (String, default: UTF-8)

httpMethod

the http method to use when performing the request (HttpMethod, default: POST, possible values: OPTIONS,GET,HEAD,POST,PUT,PATCH,DELETE,TRACE,CONNECT)

mappedRequestHeaders

request message header names to be propagated to/from the adpater/gateway (String, default: HTTP_REQUEST_HEADERS)

mappedResponseHeaders

response message header names to be propagated from the adpater/gateway (String, default: HTTP_RESPONSE_HEADERS)

replyTimeout

the amount of time to wait (ms) for a response from the remote server (int, default: 0)

url

the url to perform an http request on (String, no default)

JSON to Tuple (json-to-tuple)

The json-to-tuple processor is able to transform a String representation of some JSON map into a Tuple.

Here is a simple example:

xd:>stream create tuples --definition "http | json-to-tuple | transform --expression='payload.firstName + payload.lastName' | log" --deploy

xd:>http post --data '{"firstName": "Spring", "lastName": "XD"}'
Note

Transformation to Tuple can be used as an alternative to Type Conversion.

The json-to-tuple processor has no particular option (in addition to options shared by all modules)

Object to JSON (object-to-json)

The object-to-json processor can be used to convert any java Object to a JSON String.

In the following example, notice how the collection of three elements is transformed to JSON (in particular, the three Strings are surrounded by quotes):

xd:>stream create json --deploy --definition "http | aggregator --count | object-to-json | log"

xd:>http post --data hello
xd:>http post --data world
xd:>http post --data !

results in ["hello", "world", "!"] appearing in the log.

Note

Transformation to JSON can be used as an alternative to Type Conversion.

The object-to-json processor has no particular option (in addition to options shared by all modules)

Script

The script processor contains a Service Activator that invokes a specified Groovy script. This is a slightly more generic way to accomplish processing logic, as the provided script may simply terminate the stream as well as transform or filter Messages.

The script processor has the following options:

propertiesLocation

the path of a properties file containing custom script variable bindings (String, no default)

refreshDelay

how often to check (in milliseconds) whether the script has changed; -1 for never (long, default: 60000)

script

reference to a script used to process messages (String, no default)

variables

variable bindings as a comma delimited string of name-value pairs, e.g., 'foo=bar,baz=car' (String, no default)

To use the module, pass the location of a Groovy script using the script attribute. If you want to pass variable values to your script, you can statically bind values using the variables option or optionally pass the path to a properties file containing the bindings using the propertiesLocation option. All properties in the file will be made available to the script as variables. Note that payload and headers are implicitly bound to give you access to the data contained in a message. See the Filter example for a more detailed discussion of script variables.

xd:> stream create --name groovyprocessortest --definition "http --port=9006 | script --script=custom-processor.groovy --variables='x=foo' | log" --deploy
xd:> stream create --name groovyprocessortest --definition "http --port=9006 | script --script=custom-processor.groovy --propertiesLocation=custom-processor.properties | log" --deploy

By default, Spring XD will search the classpath for custom-processor.groovy and custom-processor.properties. You can place the script in ${xd.home}/modules/processor/scripts and the properties file in ${xd.home}/config to make them available on the classpath. Alternatively, you can prefix the location and properties-location values with file: to load from the file system.

Tip
The script file’s modified timestamp is checked for changes every 60 seconds by default; this can be changed with the refreshDelay deployment property: --refreshDelay=30000 (every 30 seconds or 30,000ms), --refreshDelay=-1 to disable refresh.

Shell

The shell processor forks an external process by running a shell command to launch a process written in any language. The process should implement a continual loop that waits for input from stdin and writes a result to stdout in a request-response manner. The process will be destroyed when the stream is undeployed. For example, it is possible to invoke a Python script within a stream in this manner. Since the shell processor relies on low-level stream processing there are some additional requirements:

  • Input and output data are expected to be Strings, the charset is configurable.

  • The shell process must not write out of band data to stdout, such as a start up message or prompt.

  • Anything written to stderr will be logged as an ERROR in Spring XD but will not terminate the stream.

  • Responses written to stdout must be terminated using the configured encoder (CRLF or "\r\n" is the default) for the module and must not exceed the configured bufferSize

  • Any external software required to run the script must be installed on the container node to which the module is deployed.

Here is a simple Python example that echos the input:

#echo.py
import sys

#=====================
# Write data to stdout
#=====================
def send(data):
  sys.stdout.write(data)
  sys.stdout.flush()

#===========================================
# Terminate a message using the default CRLF
#===========================================
def eod():
  send("\r\n")

#===========================
# Main - Echo the input
#===========================

while True:
  try:
    data = raw_input()
    if data:
      send(data)
      eod()
  except EOFError:
      eod()
      break
Note

Spring XD provides additional Python programming support for handling basic stream processing, as shown above, see xref:creating a Python module.

To try this example, copy the above script and save it to echo.py. Start Spring XD and create a stream:

xd:>stream create pytest --definition "time | shell --command='python <absolute-path-to>/echo.py' | log" --deploy
Created and deployed new stream 'pytest'

you should see the time echoed in the log:

09:49:14,856  INFO task-scheduler-5 sink.pytest - 2014-10-10 09:49:14
09:49:15,860  INFO task-scheduler-1 sink.pytest - 2014-10-10 09:49:15
09:49:16,862  INFO task-scheduler-1 sink.pytest - 2014-10-10 09:49:16
09:49:17,864  INFO task-scheduler-1 sink.pytest - 2014-10-10 09:49:17

This script can be easily modified to do some actual work by providing a function that takes the input as an argument and returns a string. Then insert the function call:

while True:
  try:
    data = raw_input()
    if data:
      result = myfunc(data)
      send(result)
      eod()
  except EOFError:
      eod()
      break

The shell processor has the following options:

bufferSize

the size of the buffer (bytes) to use when encoding/decoding (int, default: 2048)

charset

the charset used when converting from String to bytes (String, default: UTF-8)

command

the shell command (String, no default)

encoder

the encoder to use when sending messages (Encoding, default: CRLF, possible values: CRLF,LF,NULL,STXETX,RAW,L1,L2,L4)

environment

additional process environment variables as comma delimited name-value pairs (String, no default)

redirectErrorStream

redirects stderr to stdout (boolean, default: false)

workingDir

the process working directory (String, no default)

Splitter

The splitter module builds upon the concept of the same name in Spring Integration and allows the splitting of a single message into several distinct messages.

The splitter processor has the following options:

expression

a SpEL expression which would typically evaluate to an array or collection (String, default: payload)

Note
The default value for expression is payload, which actually does not split, unless the message is already a collection.

As part of the SpEL expression you can make use of the pre-registered JSON Path function. The syntax is #jsonPath(payload,<json path expression>)

Extract the value of a specific field

This splitter converts a JSON message payload to the value of a specific JSON field.

xd:> stream create --name jsontransformtest --definition "http --port=9005 | splitter --expression=#jsonPath(payload,'$.firstName') | log" --deploy

Try sending this payload to the HTTP endpoint and you should see just the value "John" in the XD log:

xd:> http post --target http://localhost:9005 --data '{"firstName":"John", "lastName":"Smith"}'

Transform

Use the transform module in a stream to convert a Message’s content or structure.

The transform processor has the following options:

expression

a SpEL expression used to transform messages (String, default: payload.toString())

propertiesLocation

the path of a properties file containing custom script variable bindings (String, no default)

refreshDelay

how often to check (in milliseconds) whether the script has changed; -1 for never (long, default: 60000)

script

reference to a script used to process messages (String, no default)

variables

variable bindings as a comma delimited string of name-value pairs, e.g., 'foo=bar,baz=car' (String, no default)

Transform with SpEL expression

The simplest way to use the transform processor is to pass a SpEL expression when creating the stream. The expression should return the modified message or payload. For example:

xd:> stream create --name transformtest --definition "http --port=9003 | transform --expression=payload.toUpperCase() | log" --deploy

This transform will convert all message payloads to upper case. If sending the word "foo" to the HTTP endpoint and you should see "FOO" in the XD log:

xd:> http post --target http://localhost:9003 --data "foo"

As part of the SpEL expression you can make use of the pre-registered JSON Path function. The syntax is #jsonPath(payload,<json path expression>)

Transform with Groovy Script

For more complex transformations, you can pass the location of a Groovy script using the script option. If you want to pass variable values to your script, you can statically bind values using the variables option or optionally pass the path to a properties file containing the bindings using the propertiesLocation option. All properties in the file will be made available to the script as variables. Note that payload and headers are implicitly bound to give you access to the data contained in a message. See the Filter example for a more detailed discussion of script variables.

xd:> stream create --name groovytransformtest1 --definition "http --port=9004 | transform --script=custom-transform.groovy --variables="x=foo" | log" --deploy
xd:> stream create --name groovytransformtest2 --definition "http --port=9004 | transform --script=custom-transform.groovy --propertiesLocation=custom-transform.properties | log" --deploy

By default, Spring XD will search the classpath for custom-transform.groovy and custom-transform.properties. You can place the script in ${xd.home}/modules/processor/scripts and the properties file in ${xd.home}/config to make them available on the classpath. Alternatively, you can prefix the script and properties-location values with file: to load from the file system.

Tip
The script file’s modified timestamp is checked for changes every 60 seconds by default; this can be changed with the refreshDelay deployment property: --refreshDelay=30000 (every 30 seconds or 30,000ms), --refreshDelay=-1 to disable refresh.

Sinks

This section describes the sink modules included with Spring XD. A sink terminates a stream to persist data or push it to an external consumer. To run the examples shown here, start the XD Container as instructed in the Getting Started page.

The available sinks are

Additionally, Spring XD provides a number of counters and gauges, which are specialized sinks useful for real time analytics.

See the section Creating a Sink Module for information on how to create sink modules using other Spring Integration Adapters.

Dynamic Router (router)

The Dynamic Router support allows for routing Spring XD messages to named channels based on the evaluation of SpEL expressions or Groovy Scripts.

SpEL-based Routing

In the following example, 2 streams are created that listen for message on the foo and the bar channel. Furthermore, we create a stream that receives messages via HTTP and then delegates the received messages to a router:

xd:>stream create f --definition "queue:foo > transform --expression=payload+'-foo' | log" --deploy
Created new stream 'f'

xd:>stream create b --definition "queue:bar > transform --expression=payload+'-bar' | log" --deploy
Created new stream 'b'

xd:>stream create r --definition "http | router --expression=payload.contains('a')?'queue:foo':'queue:bar'" --deploy
Created new stream 'r'

Now we make 2 requests to the HTTP source:

xd:>http post --data "a"
> POST (text/plain;Charset=UTF-8) http://localhost:9000 a
> 200 OK

xd:>http post --data "b"
> POST (text/plain;Charset=UTF-8) http://localhost:9000 b
> 200 OK

In the server log you should see the following output:

11:54:19,868  WARN ThreadPoolTaskScheduler-1 sink.f:145 - a-foo
11:54:25,669  WARN ThreadPoolTaskScheduler-1 sink.b:145 - b-bar

For more information, please also consult the Spring Integration Reference manual: http://static.springsource.org/spring-integration/reference/html/messaging-routing-chapter.html#router-namespace particularly the section "Routers and the Spring Expression Language (SpEL)".

Groovy-based Routing

Instead of SpEL expressions, Groovy scripts can also be used. Let’s create a Groovy script in the file system at "/my/path/router.groovy"

println("Groovy processing payload '" + payload +"'");
if (payload.contains('a')) {
	return ":foo"
}
else {
	return ":bar"
}

Now we create the following streams:

xd:>stream create f --definition ":foo > transform --expression=payload+'-foo' | log" --deploy
Created new stream 'f'

xd:>stream create b --definition ":bar > transform --expression=payload+'-bar' | log" --deploy
Created new stream 'b'

xd:>stream create g --definition "http | router --script='file:/my/path/router.groovy'" --deploy

Now post some data to the HTTP source:

xd:>http post --data "a"
> POST (text/plain;Charset=UTF-8) http://localhost:9000 a
> 200 OK

xd:>http post --data "b"
> POST (text/plain;Charset=UTF-8) http://localhost:9000 b
> 200 OK

In the server log you should see the following output:

Groovy processing payload 'a'
11:29:27,274  WARN ThreadPoolTaskScheduler-1 sink.f:145 - a-foo
Groovy processing payload 'b'
11:34:09,797  WARN ThreadPoolTaskScheduler-1 sink.b:145 - b-bar
Note

You can also use Groovy scripts located on your classpath by specifying:

--script='org/my/package/router.groovy'

If you want to pass variable values to your script, you can statically bind values using the variables option or optionally pass the path to a properties file containing the bindings using the propertiesLocation option. All properties in the file will be made available to the script as variables. You may specify both variables and propertiesLocation, in which case any duplicate values provided as variables override values provided in propertiesLocation. Note that payload and headers are implicitly bound to give you access to the data contained in a message.

For more information, see the Spring Integration Reference manual: "Groovy support" http://static.springsource.org/spring-integration/reference/html/messaging-endpoints-chapter.html#groovy

Options

The router sink has the following options:

expression

a SpEL expression used to transform messages (String, default: payload.toString())

propertiesLocation

the path of a properties file containing custom script variable bindings (String, no default)

refreshDelay

how often to check (in milliseconds) whether the script has changed; -1 for never (long, default: 60000)

script

reference to a script used to process messages (String, no default)

variables

variable bindings as a comma delimited string of name-value pairs, e.g., 'foo=bar,baz=car' (String, no default)

Tip
If the script option is set, the script file’s modified timestamp is checked for changes every 60 seconds by default; this can be changed with the refreshDelay deployment property: --refreshDelay=30000 (every 30 seconds or 30,000ms), --refreshDelay=-1 to disable refresh.

File Sink (file)

Another simple option is to stream data to a file on the host OS. This can be done using the file sink module to create a stream.

xd:> stream create --name myfilestream --definition "http --port=8000 | file" --deploy

We’ve used the http source again, so run the following command to send a message

xd:> http post --target http://localhost:8000 --data "hello"

The file sink uses the stream name as the default name for the file it creates, and places the file in the /tmp/xd/output/ directory.

$ less /tmp/xd/output/myfilestream
hello

You can cutomize the behavior and specify the name and dir options of the output file. For example

xd:> stream create --name otherfilestream --definition "http --port=8000 | file --name=myfile --dir=/some/custom/directory" --deploy

To set the filename from a SpEL expression (e.g. headers[file_name]), you can use the nameExpression option.

xd:> stream create --name myfilestream --definition "http --port=8000 | file --nameExpression=payload.trim()" --deploy

If you run this command :

xd:> http post --target http://localhost:8000 --data "hello.txt"

It will take the payload of the message ("hello.txt"), as it’s defined previously with nameExpression, and use it as the filename. In this example, the filename is equal to the content of the file.

You can use dirExpression to specify the name of the directory that will contain the new file.

xd:> stream create --name myfilestream --definition "http --port=8000 | file --nameExpression=payload.trim() --dirExpression='''/tmp/test/dir-'' + payload.trim()'" --deploy

If you run this command :

xd:> http post --target http://localhost:8000 --data "hello.txt"

For the filename, it will do the same thing as explained previously. For the directory name it will use the content of the file (trimmed) concatenated with dir- (in that case : "/tmp/test/dir-hello.txt"). If the destination directory does not exists, the respective destination directory and any non-existing parent directories are being created automatically.

When you use the nameExpression option you have to use the dirExpression option (not the dir option) to specify the destination directory name, even if it’s a simple string (e.g. 'mydir').

File with Options

The file sink has the following options:

binary

if false, will append a newline character at the end of each line (boolean, default: false)

charset

the charset to use when writing a String payload (String, default: UTF-8)

dir

the directory in which files will be created (String, default: /tmp/xd/output/)

dirExpression

spring expression used to define directory name (String, no default)

mode

what to do if the file already exists (Mode, default: APPEND, possible values: APPEND,REPLACE,FAIL,IGNORE)

name

filename pattern to use (String, default: <stream name>)

nameExpression

spring expression used to define filename (String, no default)

suffix

filename extension to use (String, no default)

FTP Sink (ftp)

FTP sink is a simple option to push files to an FTP server from incoming messages.

It uses an ftp-outbound-adapter, therefore incoming messages could be either a java.io.File object, a String (content of the file) or an array of bytes (file content as well).

To use this sink, you need a username and a password to login. Once you have this you can stream data from, for instance, a file source to the ftp sink:

xd:> stream create --name mystream --definition "file | ftp --username=me --password=mypwd" --deploy

We use the file source, so create a file:

$ echo hello > /tmp/xd/input/mystream/test.txt

On the ftp server, you should see the file test.txt with the content hello.

To pass the filename to the module you can use the header file_name with the filename you wish to be used.

NOTE: By default Spring Integration will use o.s.i.file.DefaultFileNameGenerator if none is specified. DefaultFileNameGenerator will determine the file name based on the value of the file_name header (if it exists) in the MessageHeaders, or if the payload of the Message is already a java.io.File, then it will use the original name of that file.

FTP with Options

The ftp sink has the following options:

autoCreateDir

remote directory must be auto created if it does not exist (boolean, default: true)

clientMode

client mode to use: 2 for passive mode and 0 for active mode (int, default: 0)

host

the host name for the FTP server (String, default: localhost)

mode

what to do if the file already exists (Mode, default: REPLACE, possible values: APPEND,REPLACE,FAIL,IGNORE)

password

the password for the FTP connection (Password, no default)

port

the port for the FTP server (int, default: 21)

remoteDir

the remote directory to transfer the files to (String, default: /)

remoteFileSeparator

file separator to use on the remote side (String, default: /)

temporaryRemoteDir

temporary remote directory that should be used (String, default: /)

tmpFileSuffix

extension to use on server side when uploading files (String, default: .tmp)

useTemporaryFilename

use a temporary filename while transferring the file and rename it to its final name once it's fully transferred (boolean, default: true)

username

the username for the FTP connection (String, no default)

GemFire Server

Currently XD supports GemFire’s client-server topology. A sink that writes data to a GemFire cache requires at least one cache server to be running in a separate process and may also be configured to use a Locator. While Gemfire configuration is outside of the scope of this document, details are covered in the GemFire Product documentation. The XD distribution includes a standalone GemFire server executable suitable for development and test purposes and bootstrapped using a Spring configuration file provided as a command line argument. The GemFire jar is distributed freely under GemFire’s development license and is subject to the license’s terms and conditions. Sink modules provided with the XD distrubution that write data to GemFire create a client cache and client region. No data is cached on the client.

Tip
If native gemfire properties are required to configure the client cache, e.g., for security, place a gemfire.properties file in $XD_HOME/config.

Launching the XD GemFire Server

To start the GemFire cache server GemFire Server included in the Spring XD distribution, go to the XD install directory:

$cd gemfire/bin
$./gemfire-server ../config/cq-demo.xml

The command line argument is the path of a Spring Data Gemfire configuration file with including a configured cache server and one or more regions. A sample cache configuration is provided cq-demo.xml located in the config directory. Note that Spring interprets the path as a relative path unless it is explicitly preceded by file:. The sample configuration starts a server on port 40404 and creates a region named Stocks.

Gemfire sinks

There are 2 implementations of the gemfire sink: gemfire-server and gemfire-json-server. They are identical except the latter converts JSON string payloads to a JSON document format proprietary to GemFire and provides JSON field access and query capabilities. If you are not using JSON, the gemfire-server module will write the payload using java serialization to the configured region. Both modules accept the same options.

The gemfire-server sink has the following options:

host

host name of the cache server or locator (if useLocator=true). May be a comma delimited list (String, no default)

keyExpression

a SpEL expression which is evaluated to create a cache key (String, default: '<stream name>')

port

port of the cache server or locator (if useLocator=true). May be a comma delimited list (String, no default)

regionName

name of the region to use when storing data (String, default: <stream name>)

useLocator

indicates whether a locator is used to access the cache server (boolean, default: false)

Tip
The keyExpression, as its name suggests, is a SpEL. Typically, the key value is derived from the payload. The default of '<streamname>' (mind the quotes), will overwrite the same entry for every message received on the stream.
Note
The useLocator option is intended for integration with an existing GemFire installation in which the cache servers are configured to use locators in accordance with best practice. GemFire supports configuration of multiple locators (or direct server connections) and this is specified by supplying comma-delimited values for the host and port options. You may specify a single value for either of these options otherwise each value must contain the same size list. The following are examples are valid for multiple connection addresses:
gemfire-server --host=myhost --port=10334,10335
gemfire server --host=myhost1,myhost2 --port=10334
gemfire server --host=myhost1,myhost2,myhost3 --port=10334,10335,10336

The last example creates connections to myhost1:10334, myhost2:10335, myhost3:10336

Note
You may also configure default Gemfire connection settings for all gemfire modules in config\modules.yml:
gemfire:
   useLocator: true
   host: myhost1,myhost2
   port: 10334

Example

Suppose we have a JSON document containing a stock price:

{"symbol":"FAKE", "price":73}

We want this to be cached using the stock symbol as the key. The stream definition is:

http | gemfire-json-server --regionName=Stocks --keyExpression=payload.getField('symbol')

The keyExpression is a SpEL expression that depends on the payload type. In this case, com.gemstone.org.json.JSONObject. JSONObject which provides the getField method. To run this example:

xd:> stream create --name stocks --definition "http --port=9090 | gemfire-json-server --regionName=Stocks --keyExpression=payload.getField('symbol')" --deploy
xd:> http post --target http://localhost:9090 --data {"symbol":"FAKE","price":73}

This will write an entry to the GemFire Stocks region with the key FAKE. Please do not put spaces when separating the JSON key-value pairs, only a comma.

You should see a message on STDOUT for the process running the GemFire server like:

INFO [LoggingCacheListener] - updated entry FAKE
Tip
If you are deploying on Java 7 or earlier and need to deploy more than 4 Gemfire modules, be sure to increase the permsize of the singlenode or container. i.e. JAVA_OPTS="-XX:PermSize=256m".

GPFDIST

The gpfdist sink allows you to stream data in parallel to either Pivotal Greenplum DB or Pivotal HAWQ. Internally, this sink creates a custom http listener that supports the gpfdist protcol and schedules a task that orchestrates a gploadd session in the same way it is done natively in Greenplum.

No data is written into temporary files and all data is kept in stream buffers waiting to get inserted into Greenplum DB or HAWQ. If there are no existing load sessions from Greenplum, the sink will block until such sessions are established.

Example usage

The load-generator-gpfdist source can be used to send dummy test data to the gpfdist sink.

Using psql, create the following table with a simple schema that matches the data produced by the load-generator-string source, two integer values, a producer ID and a timestamp separated by a tab.

create table xdsink (date integer, time integer) distributed randomly;

Now create the stream definition and deploy. You should ensure that your pg_hba.conf (e.g. /data/master/gpsne-1/pg_hba.conf) is configured to allow a connection from your host where you are running the gpfdist sink. (an entry such as host all gpadmin 192.168.70.128/32 trust)

xd:>stream create --name gpfdiststream --definition "load-generator-gpfdist --messageCount=10000000 --producers=1 --recordType=counter  | gpfdist --dbHost=192.168.70.138 --table=xdsink --batchTimeout=5 --batchCount=1000 --batchPeriod=0 --flushCount=200 --flushTime=2 --rateInterval=1000000" --deploy
Created and deployed new stream 'gpfdiststream'

In this XD stream we send 10M messages from the load-generator-string source to the gpfdist sink. We roughly keep load session alive for 5 seconds while flushing data after 2s or 200 entries which ever comes first and sleep 0s in between load sessions.

You will see log output (you will probably need to set the log level of the package log4j.logger.org.springframework.xd.greenplum to INFO.)

2015-05-14 22:48:23,669 1.2.0.SNAP  INFO pool-14-thread-1 gpfdist.GPFDistMessageHandler - METER: 1 minute rate = 200000.0 mean rate = 269618.7284878825
2015-05-14 22:48:25,495 1.2.0.SNAP  INFO sqlTaskScheduler-1 support.CleanableJdbcOperations - DROP EXTERNAL TABLE xdsink_ext_fabcf3bb_c514_49ca_bfd6_cacb009463dc
2015-05-14 22:48:25,498 1.2.0.SNAP  INFO sqlTaskScheduler-1 support.CleanableJdbcOperations - CREATE READABLE EXTERNAL TABLE xdsink_ext_ae89e85d_eb65_4e11_ad72_4b8302086ebd ( LIKE xdsink ) LOCAT

gpfdist sink currently contains a throughput meter for this POC to get perf numbers. In this case it is showing about 270K/sec messages per second to be transferred from XD into Greenplum.

In previous example we did a simple inserts into a table. Let’s see how we can update data in a table. Create a simple table xdsink2 with three text columns and insert some data.

create table xdsink2 (col1 text, col2 text, col3 text);
insert into xdsink2 values ('DATA1', 'DATA', 'DATA');
insert into xdsink2 values ('DATA2', 'DATA', 'DATA');
insert into xdsink2 values ('DATA3', 'DATA', 'DATA');

We’d have different data in col1 which is used to match data to be updated.

# select * from xdsink2;
 col1  | col2 | col3
-------+------+------
 DATA3 | DATA | DATA
 DATA2 | DATA | DATA
 DATA1 | DATA | DATA

Let’s create a stream which will update table xdsink2 by matching a column col1 and updates columns col2 and col3.

xd:>stream create --name gpfdiststream2 --definition "http|gpfdist --mode=UPDATE --table=xdsink2 --dbHost=mdw --columnDelimiter=',' --matchColumns=col1 --updateColumns=col2,col3" --deploy

Post some data into a stream which will be passed into a gpfdist sink via http source.

xd:>http post --data "DATA1,DATA1,DATA1"

If you query table again, you’ll see that row for DATA1 has been updated.

# select * from xdsink2;
 col1  | col2  | col3
-------+-------+-------
 DATA3 | DATA  | DATA
 DATA2 | DATA  | DATA
 DATA1 | DATA1 | DATA1

Performance Notes

On a Lenovo W540, Spring XD singlenode, load-generator-string | gpfdist inserted data at ~ 540K/sec. The underlying message handler in the gpfdist sink is able to achieve ~1.2M/sec, which is comprable to the use of the native gpload client. Additional performance optimizations when used within an XD stream are on the roadmap.

Implementation Notes

Within a gpfdist sink we have a Reactor based stream where data is published from the incoming SI channel. This channel receives data from the Message Bus. The Reactor stream is then connected to Netty based http channel adapters so that when a new http connection is established, the Reactor stream is flushed and balanced among existing http clients. When Greenplum does a load from an external table, each segment will initiate a http connection and start loading data. The net effect is that incoming data is automatically spread among the Greenplum segments.

GPFDIST with Options

The options flushCount and flushTime are used to determine when to flush data that is buffered in an internal Reactor stream to the http connection. Data is flushed based on if the count value has been reached or the time specified has elapsed. Note that with too high a value, memory consumption will go up. Too small a value combined with a low ingestion rate will result in data being inserted into the database less frequently.

batchCount defines the maximum count of aggregated windows the client takes before the internal Reactor stream and http channel is closed.

batchTimeout defines how many seconds each http connection should be kept alive if no data is streamed to a client. Use this together with batchCount to estimate how long each loading session should last.

batchPeriod defines how many seconds a task running load operation should sleep in between a loads.

mode defines a database load logic which is either INSERT or UPDATE. INSERT is a default mode. Similar to control file GPLOAD.OUTPUT.MODE property. MERGE is not currently supported.

columnDelimiter defines a data delimiter character within a line of data. Defaults to tabulator character. Similar to control file GPLOAD.SOURCE.DELIMITER property.

updateColumns defines updated column names and required with mode UPDATE. Similar to control file GPLOAD.OUTPUT.UPDATE_COLUMNS property.

matchColumns defines matched column names and required with mode UPDATE. Similar to control file GPLOAD.OUTPUT.MATCH_COLUMNS property.

sqlBefore defines a simple sql clause to be run before of every load operation. Similar to control file GPLOAD.SQL.BEFORE property.

sqlAfter defines a simple sql clause to be run after of every load operation. Similar to control file GPLOAD.SQL.AFTER property.

delimiter is used to postfix incoming data with a line termination because Greenplum expects line terminated data.

controlFile can be used to introduce more parameters for a load operation. For simple use cases, the table property can be used.

rateInterval if set, enables rate logging passing through sink.

The gpfdist sink has the following options:

batchCount

batch count (int, default: 100)

batchPeriod

batch period (int, default: 10)

batchTimeout

batch timeout (int, default: 4)

columnDelimiter

column delimiter (Character, no default)

controlFile

path to yaml control file (String, no default)

dbHost

database host (String, default: localhost)

dbName

database name (String, default: gpadmin)

dbPassword

database password (String, default: gpadmin)

dbPort

database port (int, default: 5432)

dbUser

database user (String, default: gpadmin)

delimiter

data line delimiter (String, default: ` `)

flushCount

flush item count (int, default: 100)

flushTime

flush item time (int, default: 2)

matchColumns

match columns with update (String, no default)

mode

mode, either insert or update (String, no default)

port

gpfdist listen port (int, default: 0)

rateInterval

enable transfer rate interval (int, default: 0)

sqlAfter

sql to run after load (String, no default)

sqlBefore

sql to run before load (String, no default)

table

target database table (String, no default)

updateColumns

update columns with update (String, no default)

Cassandra

The Cassandra sink write into a Cassandra table. Here is a simple example

xd:>stream create cassandraTest --definition "http | cassandra --initScript=file:<absolut-path-to>/int-db.cql --ingestQuery='insert into book (isbn, title, author) values (uuid(), ?, ?)'" --deploy

Where int-db.cql looks like:

DROP TABLE IF EXISTS book;

CREATE TABLE book  (
    isbn        uuid PRIMARY KEY,
    author      text,
    instock     boolean,
    pages       int,
    saledate    timestamp,
    title       text
);

You can then send data to this stream via

xd:>http post --data "{\"title\": \"The Art of War\", \"author\": \"Sun Tzu\"}" --target http://localhost:9000

and see the table contents using the CQL

SELECT * FROM cassandratest.book;

Another Insert example using a domain Book object is

xd:>stream create cassandraTest2 --definition "http | transform --expression='new org.springframework.xd.test.domain.Book(T(java.util.UUID).randomUUID(), #jsonPath(payload, \"$.title\"), #jsonPath(payload, \"$.author\"))' | cassandra --keyspace=cassandraTest" --deploy
http post --data "{\"title\": \"The Art of War\", \"author\": \"Sun Tzu\"}" --target http://localhost:9000

You must place your domain object into the xd/lib directory, since some Spring Data comons classes are in the xd/lib directory and loaded into root classpath, not the module’s classpath.

The cassandra sink has the following options:

compressionType

the compression to use for the transport (CompressionType, default: NONE, possible values: NONE,SNAPPY)

consistencyLevel

the consistencyLevel option of WriteOptions (ConsistencyLevel, no default, possible values: ANY,ONE,TWO,THREE,QUOROM,LOCAL_QUOROM,EACH_QUOROM,ALL,LOCAL_ONE,SERIAL,LOCAL_SERIAL)

contactPoints

the comma-delimited string of the hosts to connect to Cassandra (String, default: localhost)

entityBasePackages

the base packages to scan for entities annotated with Table annotations (String[], default: [])

ingestQuery

the ingest Cassandra query (String, no default)

initScript

the path to file with CQL scripts (delimited by ';') to initialize keyspace schema (String, no default)

keyspace

the keyspace name to connect to (String, default: <stream name>)

metricsEnabled

enable/disable metrics collection for the created cluster (boolean, default: true)

password

the password for connection (String, no default)

port

the port to use to connect to the Cassandra host (int, default: 9042)

queryType

the queryType for Cassandra Sink (Type, default: INSERT, possible values: INSERT,UPDATE,DELETE,STATEMENT)

retryPolicy

the retryPolicy option of WriteOptions (RetryPolicy, no default, possible values: DEFAULT,DOWNGRADING_CONSISTENCY,FALLTHROUGH,LOGGING)

statementExpression

the expression in Cassandra query DSL style (String, no default)

ttl

the time-to-live option of WriteOptions (int, default: 0)

username

the username for connection (String, no default)

Hadoop (HDFS) (hdfs)

If you do not have Hadoop installed, you can install Hadoop as described in our separate guide. Spring XD supports 4 Hadoop distributions, see using Hadoop for more information on how to start Spring XD to target a specific distribution.

Once Hadoop is up and running, you can then use the hdfs sink when creating a stream

xd:> stream create --name myhdfsstream1 --definition "time | hdfs" --deploy

In the above example, we’ve scheduled time source to automatically send ticks to hdfs once in every second. If you wait a little while for data to accumuluate you can then list can then list the files in the hadoop filesystem using the shell’s built in hadoop fs commands. Before making any access to HDFS in the shell you first need to configure the shell to point to your name node. This is done using the hadoop config command.

xd:>hadoop config fs --namenode hdfs://localhost:8020

In this example the hdfs protocol is used but you may also use the webhdfs protocol. Listing the contents in the output directory (named by default after the stream name) is done by issuing the following command.

xd:>hadoop fs ls /xd/myhdfsstream1
Found 1 items
-rw-r--r--   3 jvalkealahti supergroup          0 2013-12-18 18:10 /xd/myhdfsstream1/myhdfsstream1-0.txt.tmp

While the file is being written to it will have the tmp suffix. When the data written exceeds the rollover size (default 1GB) it will be renamed to remove the tmp suffix. There are several options to control the in use file file naming options. These are --inUsePrefix and --inUseSuffix set the file name prefix and suffix respectfully.

When you destroy a stream

xd:>stream destroy --name myhdfsstream1

and list the stream directory again, in use file suffix doesn’t exist anymore.

xd:>hadoop fs ls /xd/myhdfsstream1
Found 1 items
-rw-r--r--   3 jvalkealahti supergroup        380 2013-12-18 18:10 /xd/myhdfsstream1/myhdfsstream1-0.txt

To list the list the contents of a file directly from a shell execute the hadoop cat command.

xd:> hadoop fs cat /xd/myhdfsstream1/myhdfsstream1-0.txt
2013-12-18 18:10:07
2013-12-18 18:10:08
2013-12-18 18:10:09
...

In the above examples we didn’t yet go through why the file was written in a specific directory and why it was named in this specific way. Default location of a file is defined as /xd/<stream name>/<stream name>-<rolling part>.txt. These can be changed using options --directory and --fileName respectively. Example is shown below.

xd:>stream create --name myhdfsstream2 --definition "time | hdfs --directory=/xd/tmp --fileName=data" --deploy
xd:>stream destroy --name myhdfsstream2
xd:>hadoop fs ls /xd/tmp
Found 1 items
-rw-r--r--   3 jvalkealahti supergroup        120 2013-12-18 18:31 /xd/tmp/data-0.txt

It is also possible to control the size of a files written into HDFS. The --rollover option can be used to control when file currently being written is rolled over and a new file opened by providing the rollover size in bytes, kilobytes, megatypes, gigabytes, and terabytes.

xd:>stream create --name myhdfsstream3 --definition "time | hdfs --rollover=100" --deploy
xd:>stream destroy --name myhdfsstream3
xd:>hadoop fs ls /xd/myhdfsstream3
Found 3 items
-rw-r--r--   3 jvalkealahti supergroup        100 2013-12-18 18:41 /xd/myhdfsstream3/myhdfsstream3-0.txt
-rw-r--r--   3 jvalkealahti supergroup        100 2013-12-18 18:41 /xd/myhdfsstream3/myhdfsstream3-1.txt
-rw-r--r--   3 jvalkealahti supergroup        100 2013-12-18 18:41 /xd/myhdfsstream3/myhdfsstream3-2.txt

Shortcuts to specify sizes other than bytes are written as --rollover=64M, --rollover=512G or --rollover=1T.

The stream can also be compressed during the write operation. Example of this is shown below.

xd:>stream create --name myhdfsstream4 --definition "time | hdfs --codec=gzip" --deploy
xd:>stream destroy --name myhdfsstream4
xd:>hadoop fs ls /xd/myhdfsstream4
Found 1 items
-rw-r--r--   3 jvalkealahti supergroup         80 2013-12-18 18:48 /xd/myhdfsstream4/myhdfsstream4-0.txt.gzip

From a native os shell we can use hadoop’s fs commands and pipe data into gunzip.

# bin/hadoop fs -cat /xd/myhdfsstream4/myhdfsstream4-0.txt.gzip | gunzip
2013-12-18 18:48:10
2013-12-18 18:48:11
...

Often a stream of data may not have a high enough rate to roll over files frequently, leaving the file in an opened state. This prevents users from reading a consistent set of data when running mapreduce jobs. While one can alleviate this problem by using a small rollover value, a better way is to use the idleTimeout option that will automatically close the file if there was no writes during the specified period of time. This feature is also useful in cases where burst of data is written into a stream and you’d like that data to become visible in HDFS.

Note
The idleTimeout value should not exceed the timeout values set on the Hadoop cluster. These are typically configured using the dfs.socket.timeout and/or dfs.datanode.socket.write.timeout properties in the hdfs-site.xml configuration file.
xd:> stream create --name myhdfsstream5 --definition "http --port=8000 | hdfs --rollover=20 --idleTimeout=10000" --deploy

In the above example we changed a source to http order to control what we write into a hdfs sink. We defined a small rollover size and a timeout of 10 seconds. Now we can simply post data into this stream via source end point using a below command.

xd:> http post --target http://localhost:8000 --data "hello"

If we repeat the command very quickly and then wait for the timeout we should be able to see that some files are closed before rollover size was met and some were simply rolled because of a rollover size.

xd:>hadoop fs ls /xd/myhdfsstream5
Found 4 items
-rw-r--r--   3 jvalkealahti supergroup         12 2013-12-18 19:02 /xd/myhdfsstream5/myhdfsstream5-0.txt
-rw-r--r--   3 jvalkealahti supergroup         24 2013-12-18 19:03 /xd/myhdfsstream5/myhdfsstream5-1.txt
-rw-r--r--   3 jvalkealahti supergroup         24 2013-12-18 19:03 /xd/myhdfsstream5/myhdfsstream5-2.txt
-rw-r--r--   3 jvalkealahti supergroup         18 2013-12-18 19:03 /xd/myhdfsstream5/myhdfsstream5-3.txt

Files can be automatically partitioned using a partitionPath expression. If we create a stream with idleTimeout and partitionPath with simple format yyyy/MM/dd/HH/mm we should see writes ending into its own files within every minute boundary.

xd:>stream create --name myhdfsstream6 --definition "time|hdfs --idleTimeout=10000 --partitionPath=dateFormat('yyyy/MM/dd/HH/mm')" --deploy

Let a stream run for a short period of time and list files.

xd:>hadoop fs ls --recursive true --dir /xd/myhdfsstream6
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:42 /xd/myhdfsstream6/2014
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:42 /xd/myhdfsstream6/2014/05
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:42 /xd/myhdfsstream6/2014/05/28
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:45 /xd/myhdfsstream6/2014/05/28/09
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:43 /xd/myhdfsstream6/2014/05/28/09/42
-rw-r--r--   3 jvalkealahti supergroup        140 2014-05-28 09:43 /xd/myhdfsstream6/2014/05/28/09/42/myhdfsstream6-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:44 /xd/myhdfsstream6/2014/05/28/09/43
-rw-r--r--   3 jvalkealahti supergroup       1200 2014-05-28 09:44 /xd/myhdfsstream6/2014/05/28/09/43/myhdfsstream6-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 09:45 /xd/myhdfsstream6/2014/05/28/09/44
-rw-r--r--   3 jvalkealahti supergroup       1200 2014-05-28 09:45 /xd/myhdfsstream6/2014/05/28/09/44/myhdfsstream6-0.txt

Partitioning can also be based on defined lists. In a below example we simulate feeding data by using a time and a transform elements. Data passed to hdfs sink has a content APP0:foobar, APP1:foobar, APP2:foobar or APP3:foobar.

xd:>stream create --name myhdfsstream7 --definition "time | transform --expression=\"'APP'+T(Math).round(T(Math).random()*3)+':foobar'\" | hdfs --idleTimeout=10000 --partitionPath=path(dateFormat('yyyy/MM/dd/HH'),list(payload.split(':')[0],{{'0TO1','APP0','APP1'},{'2TO3','APP2','APP3'}}))" --deploy

Let the stream run few seconds, destroy it and check what got written in those partitioned files.

xd:>stream destroy --name myhdfsstream7
Destroyed stream 'myhdfsstream7'
xd:>hadoop fs ls --recursive true --dir /xd
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7/2014
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7/2014/05
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7/2014/05/28
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7/2014/05/28/19
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7/2014/05/28/19/0TO1_list
-rw-r--r--   3 jvalkealahti supergroup        108 2014-05-28 19:24 /xd/myhdfsstream7/2014/05/28/19/0TO1_list/myhdfsstream7-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:24 /xd/myhdfsstream7/2014/05/28/19/2TO3_list
-rw-r--r--   3 jvalkealahti supergroup        180 2014-05-28 19:24 /xd/myhdfsstream7/2014/05/28/19/2TO3_list/myhdfsstream7-0.txt
xd:>hadoop fs cat /xd/myhdfsstream7/2014/05/28/19/0TO1_list/myhdfsstream7-0.txt
APP1:foobar
APP1:foobar
APP0:foobar
APP0:foobar
APP1:foobar

Partitioning can also be based on defined ranges. In a below example we simulate feeding data by using a time and a transform elements. Data passed to hdfs sink has a content ranging from APP0 to APP15. We simple parse the number part and use it to do a partition with ranges {3,5,10}.

xd:>stream create --name myhdfsstream8 --definition "time | transform --expression=\"'APP'+T(Math).round(T(Math).random()*15)\" | hdfs --idleTimeout=10000 --partitionPath=path(dateFormat('yyyy/MM/dd/HH'),range(T(Integer).parseInt(payload.substring(3)),{3,5,10}))" --deploy

Let the stream run few seconds, destroy it and check what got written in those partitioned files.

xd:>stream destroy --name myhdfsstream8
Destroyed stream 'myhdfsstream8'
xd:>hadoop fs ls --recursive true --dir /xd
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014/05
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19/10_range
-rw-r--r--   3 jvalkealahti supergroup         16 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19/10_range/myhdfsstream8-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19/3_range
-rw-r--r--   3 jvalkealahti supergroup         35 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19/3_range/myhdfsstream8-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19/5_range
-rw-r--r--   3 jvalkealahti supergroup          5 2014-05-28 19:34 /xd/myhdfsstream8/2014/05/28/19/5_range/myhdfsstream8-0.txt
xd:>hadoop fs cat /xd/myhdfsstream8/2014/05/28/19/3_range/myhdfsstream8-0.txt
APP3
APP3
APP1
APP0
APP1
xd:>hadoop fs cat /xd/myhdfsstream8/2014/05/28/19/5_range/myhdfsstream8-0.txt
APP4
xd:>hadoop fs cat /xd/myhdfsstream8/2014/05/28/19/10_range/myhdfsstream8-0.txt
APP6
APP15
APP7

Partition using a dateFormat can be based on content itself. This is a good use case if old log files needs to be processed where partitioning should happen based on timestamp of a log entry. We create a fake log data with a simple date string ranging from 1970-01-10 to 1970-01-13.

xd:>stream create --name myhdfsstream9 --definition "time | transform --expression=\"'1970-01-'+1+T(Math).round(T(Math).random()*3)\" | hdfs --idleTimeout=10000 --partitionPath=path(dateFormat('yyyy/MM/dd/HH',payload,'yyyy-MM-DD'))" --deploy

Let the stream run few seconds, destroy it and check what got written in those partitioned files. If you see the partition paths, those are based on year 1970, not present year.

xd:>stream destroy --name myhdfsstream9
Destroyed stream 'myhdfsstream9'
xd:>hadoop fs ls --recursive true --dir /xd
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9/1970
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9/1970/01
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9/1970/01/10
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/10/00
-rw-r--r--   3 jvalkealahti supergroup         44 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/10/00/myhdfsstream9-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9/1970/01/11
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/11/00
-rw-r--r--   3 jvalkealahti supergroup         99 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/11/00/myhdfsstream9-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9/1970/01/12
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/12/00
-rw-r--r--   3 jvalkealahti supergroup         44 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/12/00/myhdfsstream9-0.txt
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:56 /xd/myhdfsstream9/1970/01/13
drwxr-xr-x   - jvalkealahti supergroup          0 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/13/00
-rw-r--r--   3 jvalkealahti supergroup         55 2014-05-28 19:57 /xd/myhdfsstream9/1970/01/13/00/myhdfsstream9-0.txt
xd:>hadoop fs cat /xd/myhdfsstream9/1970/01/10/00/myhdfsstream9-0.txt
1970-01-10
1970-01-10
1970-01-10
1970-01-10

HDFS with Options

The hdfs sink has the following options:

closeTimeout

timeout in ms, regardless of activity, after which file will be automatically closed (long, default: 0)

codec

compression codec alias name (gzip, snappy, bzip2, lzo, or slzo) (String, default: ``)

directory

where to output the files in the Hadoop FileSystem (String, default: /xd/<stream name>)

enableSync

whether writer will sync to datanode when flush is called, setting this to 'true' could impact throughput (boolean, default: false)

fileExtension

the base filename extension to use for the created files (String, default: txt)

fileName

the base filename to use for the created files (String, default: <stream name>)

fileOpenAttempts

maximum number of file open attempts to find a path (int, default: 10)

fileUuid

whether file name should contain uuid (boolean, default: false)

flushTimeout

timeout in ms, regardless of activity, after which data written to file will be flushed (long, default: 0)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

idleTimeout

inactivity timeout in ms after which file will be automatically closed (long, default: 0)

inUsePrefix

prefix for files currently being written (String, default: ``)

inUseSuffix

suffix for files currently being written (String, default: .tmp)

overwrite

whether writer is allowed to overwrite files in Hadoop FileSystem (boolean, default: false)

partitionPath

a SpEL expression defining the partition path (String, default: ``)

rollover

threshold in bytes when file will be automatically rolled over (String, default: 1G)

Note
In the context of the fileOpenAttempts option, attempt is either one rollover request or failed stream open request for a path (if another writer came up with a same path and already opened it).

Partition Path Expression

SpEL expression is evaluated against a Spring Messaging Message passed internally into a HDFS writer. This allows expression to use headers and payload from that message. While you could do a custom processing within a stream and add custom headers, timestamp is always going to be there. Data to be written is then available in a payload.

Accessing Properties

Using a payload simply returns whatever is currently being written. Access to headers is via headers property. Any other property is automatically resolved from headers if found. For example headers.timestamp is equivalent to timestamp.

Custom Methods

Addition to a normal SpEL functionality, few custom methods has been added to make it easier to build partition paths. These custom methods can be used to work with a normal partition concepts like date formatting, lists, ranges and hashes.

path
path(String... paths)

Concatenates paths together with a delimiter /. This method can be used to make the expression less verbose than using a native SpEL functionality to combine path parts together. To create a path part1/part2, expression 'part1' + '/' + 'part2' is equivalent to path('part1','part2').

Parameters
paths

Any number of path parts

Return Value

Concatenated value of paths delimited with /.

dateFormat
dateFormat(String pattern)
dateFormat(String pattern, Long epoch)
dateFormat(String pattern, Date date)
dateFormat(String pattern, String datestring)
dateFormat(String pattern, String datestring, String dateformat)

Creates a path using date formatting. Internally this method delegates into SimpleDateFormat and needs a Date and a pattern. On default if no parameter used for conversion is given, timestamp is expected. Effectively dateFormat('yyyy') equals to dateFormat('yyyy', timestamp) or dateFormat('yyyy', headers.timestamp).

Method signature with three parameters can be used to create a custom Date object which is then passed to SimpleDateFormat conversion using a dateformat pattern. This is useful in use cases where partition should be based on a date or time string found from a payload content itself. Default dateformat pattern if omitted is yyyy-MM-dd.

Parameters
pattern

Pattern compatible with SimpleDateFormat to produce a final output.

epoch

Timestamp as Long which is converted into a Date.

date

A Date to be formatted.

dateformat

Secondary pattern to convert datestring into a Date.

datestring

Date as a String

Return Value

A path part representation which can be a simple file or directory name or a directory structure.

list
list(Object source, List<List<Object>> lists)

Creates a partition path part by matching a source against a lists denoted by lists.

Lets assume that data is being written and it’s possible to extrace an appid either from headers or payload. We can automatically do a list based partition by using a partition method list(headers.appid,{{'1TO3','APP1','APP2','APP3'},{'4TO6','APP4','APP5','APP6'}}). This method would create three partitions, 1TO3_list, 4TO6_list and list. Latter is used if no match is found from partition lists passed to lists.

Parameters
source

An Object to be matched against lists.

lists

A definition of list of lists.

Return Value

A path part prefixed with a matched key i.e. XXX_list or list if no match.

range
range(Object source, List<Object> list)

Creates a partition path part by matching a source against a list denoted by list using a simple binary search.

The partition method takes a source as first argument and list as a second argument. Behind the scenes this is using jvm’s binarySearch which works on an Object level so we can pass in anything. Remember that meaningful range match only works if passed in Object and types in list are of same type like Integer. Range is defined by a binarySearch itself so mostly it is to match against an upper bound except the last range in a list. Having a list of {1000,3000,5000} means that everything above 3000 will be matched with 5000. If that is an issue then simply adding Integer.MAX_VALUE as last range would overflow everything above 5000 into a new partition. Created partitions would then be 1000_range, 3000_range and 5000_range.

Parameters
source

An Object to be matched against list.

list

A definition of list.

Return Value

A path part prefixed with a matched key i.e. XXX_range.

hash
hash(Object source, int bucketcount)

Creates a partition path part by calculating hashkey using source`s hashCode and bucketcount. Using a partition method hash(timestamp,2) would then create partitions named 0_hash, 1_hash and 2_hash. Number suffixed with _hash is simply calculated using Object.hashCode() % bucketcount.

Parameters
source

An Object which hashCode will be used.

bucketcount

A number of buckets

Return Value

A path part prefixed with a hash key i.e. XXX_hash.

HDFS Dataset (Avro/Parquet) (hdfs-dataset)

The HDFS Dataset sink is used to store Java classes that are sent as the payload on the stream. It uses the Kite SDK Data Module's Dataset implementation to store the payload data serialized in either Avro or Parquet format. The Avro schema is generated from the Java class that is persisted. For Parquet the Java object must follow JavaBean conventions with properties for any fields to be persisted. The fields can only be simple scalar values like Strings and numbers.

The HDFS Dataset sink requires that you have a Hadoop installation that is based on Hadoop v2 (Hadoop 2.2.0, Pivotal HD 1.0, Cloudera CDH4 or Hortonworks HDP 2.0), see using Hadoop for more information on how to start Spring XD to target a specific distribution.

Once Hadoop is up and running, you can then use the hdfs-dataset sink when creating a stream

xd:>stream create --name mydataset --definition "time | hdfs-dataset --batchSize=20" --deploy

In the above example, we’ve scheduled time source to automatically send ticks to the hdfs-dataset sink once every second. The data will be stored in a directory named /xd/<streamname> by default, so in this example it will be /xd/mydataset. You can change this by supplying a --basePath parameter and/or --namespace parameter. The --basePath defaults to /xd and the --namespace defaults to <streamname>. The Avro format is used by default and the data files are stored in a sub-directory named after the payload Java class. In this example the stream payload is a String so the name of the data sub-directory is string. If you have multiple Java classes as payloads, each class will get its own sub-directory.

Let the stream run for a minute or so. You can then list the contents of the hadoop filesystem using the shell’s built in hadoop fs commands. You will first need to configure the shell to point to your name node using the hadoop config command. We use the hdfs protocol is to access the hadoop name node.

xd:>hadoop config fs --namenode hdfs://localhost:8020

Then list the contents of the stream’s data directory.

xd:>hadoop fs ls /xd/mydataset/string
Found 3 items
drwxr-xr-x   - trisberg supergroup          0 2013-12-19 12:23 /xd/mydataset/string/.metadata
-rw-r--r--   3 trisberg supergroup        202 2013-12-19 12:23 /xd/mydataset/string/1387473825754-63.avro
-rw-r--r--   3 trisberg supergroup        216 2013-12-19 12:24 /xd/mydataset/string/1387473846708-80.avro

You can see that the sink has created two files containing the first two batches of 20 stream payloads each. There is also a .metadata directory created that contains the metadata that the Kite SDK Dataset implementation uses as well as the generated Avro schema for the persisted type.

xd:>hadoop fs ls /xd/mydataset/string/.metadata
Found 2 items
-rw-r--r--   3 trisberg supergroup        136 2013-12-19 12:23 /xd/mydataset/string/.metadata/descriptor.properties
-rw-r--r--   3 trisberg supergroup          8 2013-12-19 12:23 /xd/mydataset/string/.metadata/schema.avsc

Now destroy the stream.

xd:>stream destroy --name mydataset

HDFS Dataset with Options

The hdfs-dataset sink has the following options:

allowNullValues

whether null property values are allowed, if set to true then schema will use UNION for each field (boolean, default: false)

basePath

the base directory path where the files will be written in the Hadoop FileSystem (String, default: /xd)

batchSize

threshold in number of messages when file will be automatically flushed and rolled over (long, default: 10000)

compressionType

compression type name (snappy, deflate, bzip2 (avro only) or uncompressed) (String, default: snappy)

format

the format to use, valid options are avro and parquet (String, default: avro)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

idleTimeout

idle timeout in milliseconds when Hadoop file resource is automatically closed (long, default: -1)

namespace

the sub-directory under the basePath where files will be written (String, default: <stream name>)

partitionPath

the partition path strategy to use, a list of KiteSDK partition expressions separated by a '/' symbol (String, default: ``)

writerCacheSize

the size of the cache to be used for partition writers (10 if omitted) (int, default: -1)

About null values

If allowNullValues is set to true then each field in the generated schema will use a union of null and the data type of the field. You can also set allowNullValues to false and instead annotate fields in a POJO using Avro’s org.apache.avro.reflect.Nullable annotation to create a schema using a union with null for that annotated field.

About partitionPath

The partitionPath option lets you specify one or more paths that will be used to partition the files that the data is written to based on the content of the data. You can use any of the FieldPartitioners that are available for the Kite SDK project. We simply pass in what is specified to create the corresponding partition strategy. You can separate multiple paths with a / character. The following partitioning functions are available:

  • year, month, day, hour, minute creates partitions based on the value of a timestamp and creates directories named like "YEAR=2014" (works well with fields of datatype long)

    • specify function plus field name like: year('timestamp')

  • dateformat creates partitions based on a timestamp and a dateformat expression provided - creates directories based on the name provided (works well with fields of datatype long)

    • specify function plus field name, a name for the partition and the date format like: dateFormat('timestamp', 'Y-M', 'yyyyMM')

  • range creates partitions based on a field value and the upper bounds for each bucket that is specified (works well with fields of datatype int and string)

    • specify function plus field name and the upper bounds for each partition bucket like: range('age',20,50,80,T(Integer).MAX_VALUE) (Note that you can use SpEL expressions like we just did for the Integer.MAX_VALUE)

  • identity creates partitions based on the exact value of a field (works well with fields of datatype string, long and int)

    • specify function plus field name, a name for the partition, the type of the field (String or Integer) and the number of values/buckets for the partition like: identity('region','R',T(String),10)

  • hash creates partitions based on the hash calculated from the value of a field divided into a number of buckets that is specified (works well with all data types)

    • specify function plus field name and number of buckets like: hash('lastname',10)

Multiple expressions can be specified by separating them with a / like: identity('region','R',T(String),10)/year('timestamp')/month('timestamp')

JDBC

The JDBC sink can be used to insert message payload data into a relational database table. By default it inserts the entire payload into a table named after the stream name in the HSQLDB database that XD uses to store metadata for batch jobs. To alter this behavior, the jdbc sink accepts several options that you can pass using the --foo=bar notation in the stream, or change globally. There is also a config/init_db.sql file that contains the SQL statements used to initialize the database table. You can modify this file if you’d like to create a table with your specific layout when the sink starts. You should also change the initializeDatabase property to true to have this script execute when the sink starts up.

The payload data will be inserted as-is if the names option is set to payload. This is the default behavior. If you specify any other column names the payload data will be assumed to be a JSON document that will be converted to a hash map. This hash map will be used to populate the data values for the SQL insert statement. A matching of column names with underscores like user_name will match onto camel case style keys like userName in the hash map. There will be one insert statement executed for each message.

To create a stream using a jdbc sink relying on all defaults you would use a command like

xd:> stream create --name mydata --definition "time | jdbc --initializeDatabase=true" --deploy

This will insert the time messages into a payload column in a table named mydata. Since the default is using the XD batch metadata HSQLDB database we can connect to this database instance from an external tool. After we let the stream run for a little while, we can connect to the database and look at the data stored in the database.

You can query the database with your favorite SQL tool using the following database URL: jdbc:hsqldb:hsql://localhost:9101/xdjob with sa as the user name and a blank password. You can also use the HSQL provided SQL Tool (download from HSQLDB) to run a quick query from the command line:

Note
If you access any database other than HSQLDB or Postgres in a stream module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.
$ java -cp ~/Downloads/hsqldb-2.3.0/hsqldb/lib/sqltool.jar org.hsqldb.cmdline.SqlTool --inlineRc url=jdbc:hsqldb:hsql://localhost:9101/xdjob,user=sa,password= --sql "select payload from mydata;"

This should result in something similar to the following output:

2014-01-06 09:33:25
2014-01-06 09:33:26
2014-01-06 09:33:27
2014-01-06 09:33:28
2014-01-06 09:33:29
2014-01-06 09:33:30
2014-01-06 09:33:31
2014-01-06 09:33:32
2014-01-06 09:33:33
2014-01-06 09:33:34
2014-01-06 09:33:35
2014-01-06 09:33:36
2014-01-06 09:33:37

Now we can destroy the stream using:

xd:> stream destroy --name mydata

JDBC with Options

The jdbc sink has the following options:

abandonWhenPercentageFull

connections that have timed out wont get closed and reported up unless the number of connections in use are above the percentage (int, default: 0)

alternateUsernameAllowed

uses an alternate user name if connection fails (boolean, default: false)

columns

the database columns to map the data to (String, default: payload)

connectionProperties

connection properties that will be sent to our JDBC driver when establishing new connections (String, no default)

driverClassName

the JDBC driver to use (String, no default)

fairQueue

set to true if you wish that calls to getConnection should be treated fairly in a true FIFO fashion (boolean, default: true)

initSQL

custom query to be run when a connection is first created (String, no default)

initialSize

initial number of connections that are created when the pool is started (int, default: 0)

initializeDatabase

whether the database initialization script should be run (boolean, default: false)

initializerScript

the name of the SQL script (in /config) to run if 'initializeDatabase' is set (String, default: init_db.sql)

jdbcInterceptors

semicolon separated list of classnames extending org.apache.tomcat.jdbc.pool.JdbcInterceptor (String, no default)

jmxEnabled

register the pool with JMX or not (boolean, default: true)

logAbandoned

flag to log stack traces for application code which abandoned a Connection (boolean, default: false)

maxActive

maximum number of active connections that can be allocated from this pool at the same time (int, default: 100)

maxAge

time in milliseconds to keep this connection (int, default: 0)

maxIdle

maximum number of connections that should be kept in the pool at all times (int, default: 100)

maxWait

maximum number of milliseconds that the pool will wait for a connection (int, default: 30000)

minEvictableIdleTimeMillis

minimum amount of time an object may sit idle in the pool before it is eligible for eviction (int, default: 60000)

minIdle

minimum number of established connections that should be kept in the pool at all times (int, default: 10)

password

the JDBC password (Password, no default)

removeAbandoned

flag to remove abandoned connections if they exceed the removeAbandonedTimout (boolean, default: false)

removeAbandonedTimeout

timeout in seconds before an abandoned connection can be removed (int, default: 60)

suspectTimeout

this simply logs the warning after timeout, connection remains (int, default: 0)

tableName

the database table to which the data will be written (String, default: <stream name>)

testOnBorrow

indication of whether objects will be validated before being borrowed from the pool (boolean, default: false)

testOnReturn

indication of whether objects will be validated before being returned to the pool (boolean, default: false)

testWhileIdle

indication of whether objects will be validated by the idle object evictor (boolean, default: false)

timeBetweenEvictionRunsMillis

number of milliseconds to sleep between runs of the idle connection validation/cleaner thread (int, default: 5000)

url

the JDBC URL for the database (String, no default)

useEquals

true if you wish the ProxyConnection class to use String.equals (boolean, default: true)

username

the JDBC username (String, no default)

validationInterval

avoid excess validation, only run validation at most at this frequency - time in milliseconds (long, default: 30000)

validationQuery

sql query that will be used to validate connections from this pool (String, no default)

validatorClassName

name of a class which implements the org.apache.tomcat.jdbc.pool.Validator (String, no default)

Note
To include the whole message into a single column, use payload (the default) for the columns option
Tip
The connection pool settings for xd are located in servers.yml (i.e. spring.datasource.* )

Kafka Sink (kafka)

Kafka sink can be used to ingest data into a specific Kafka topic configuration.

For example,

xd:>stream create push-to-kafka --definition "http | kafka --topic=myTopic" --deploy
xd:>Created and deployed new stream 'push-to-kafka'
xd:>http post --data "push-messages"
> POST (text/plain;Charset=UTF-8) http://localhost:9000 push-messages
> 200 OK

Now, the posted messages will be available on kafka topic myTopic.

The kafka sink has the following options:

ackTimeoutOnServer

the maximum amount of time the server will wait for acknowledgments from followers to meet the acknowledgment requirements the producer has specified with the acks configuration (int, default: 30000)

batchBytes

batch size in bytes, per partition (int, default: 16384)

blockOnBufferFull

whether to block or not when the memory buffer is full (boolean, default: true)

brokerList

comma separated broker list (String, default: localhost:9092)

bufferMemory

the total bytes of memory the producer can use to buffer records waiting to be sent to the server (int, default: 33554432)

compressionCodec

compression codec to use (String, default: none)

maxBufferTime

the amount of time, in ms that the producer will wait before sending a batch to the server (int, default: 0)

maxRequestSize

the maximum size of a request (int, default: 1048576)

maxSendRetries

number of attempts to automatically retry a failed send request (int, default: 3)

receiveBufferBytes

the size of the TCP receive buffer to use when reading data (int, default: 32768)

reconnectBackoff

the amount of time to wait before attempting to reconnect to a given host when a connection fails (long, default: 10)

requestRequiredAck

producer request acknowledgement mode (int, default: 0)

retryBackoff

the amount of time to wait before attempting to retry a failed produce request to a given topic partition (long, default: 100)

sendBufferBytes

the size of the TCP send buffer to use when sending data (int, default: 131072)

topic

kafka topic name (String, default: <stream name>)

topicMetadataFetchTimeout

the maximum amount of time to block waiting for the metadata fetch to succeed (int, default: 60000)

topicMetadataRefreshInterval

the period of time in milliseconds after which a refresh of metadata is forced (int, default: 300000)

Log

Probably the simplest option for a sink is just to log the data. The log sink uses the application logger to output the data for inspection. The log level is set to WARN and the logger name is created from the stream name. To create a stream using a log sink you would use a command like

xd:> stream create --name mylogstream --definition "http --port=8000 | log" --deploy

You can then try adding some data. We’ve used the http source on port 8000 here, so run the following command to send a message

xd:> http post --target http://localhost:8000 --data "hello"

and you should see the following output in the XD container console.

13/06/07 16:12:18 INFO sink.mylogstream: hello

The log sink has the following options:

expression

the expression to be evaluated for the log content; use '#root' to log the full message (String, default: payload)

level

the log level (String, default: INFO)

name

the name of the log category to log to (will be prefixed by 'xd.sink.') (String, default: <stream name>)

Here are some examples explaining the above options:

The logger name is the sink name prefixed with the string xd.sink.. The sink name is the same as the stream name by default, but you can set it by passing the --name parameter

xd:> stream create --name myotherlogstream --definition "http --port=8001 | log --name=mylogger" --deploy

The log level is INFO by default; this can be changed with the --level property (FATAL, ERROR, WARN, INFO, DEBUG, or TRACE)

xd:> stream create --name mylogstream --definition "http --port=8001 | log --level=WARN" --deploy

By default, the message payload is logged; this can be changed with the --expression property (e.g. payload.foo to log some property foo of the payload, or #root to log the entire message)

xd:> stream create --name mylogstream --definition "http --port=8001 | log --expression=#root" --deploy

Mail

The "mail" sink allows sending of messages as emails, leveraging Spring Integration mail-sending channel adapter. Please refer to Spring Integration documentation for the details, but in a nutshell, the sink is able to handle String, byte[] and MimeMessage messages out of the box.

Here is a simple example of how the mail module is used:

xd:> stream create mystream --definition "http | mail --to='\"[email protected]\"' --host=your.imap.server --subject=payload+' world'" --deploy

Then,

xd:> http post --data Hello

You would then receive an email whose body contains "Hello" and whose subject is "Hellow world". Of special attention here is the way you need to escape strings for most of the parameters, because they’re actually SpEL expressions (so here for example, we used a String literal for the to parameter).

The mail sink has the following options:

bcc

the recipient(s) that should receive a blind carbon copy (SpEL) (String, default: null)

cc

the recipient(s) that should receive a carbon copy (SpEL) (String, default: null)

contentType

the content type to use when sending the email (SpEL) (String, default: null)

from

the primary recipient(s) of the email (SpEL) (String, default: null)

host

the hostname of the mail server (String, default: localhost)

password

the password to use to connect to the mail server (String, no default)

port

the port of the mail server (int, default: 25)

replyTo

the address that will become the recipient if the original recipient decides to "reply to" the email (SpEL) (String, default: null)

subject

the email subject (SpEL) (String, default: null)

to

the primary recipient(s) of the email (SpEL) (String, default: null)

username

the username to use to connect to the mail server (String, no default)

Mongo

The Mongo sink writes into a Mongo collection. Here is a simple example

xd:>stream create --name attendees --definition "http | mongodb --databaseName=test --collectionName=names" --deploy

Then,

xd:>http post --data {"firstName":"mark"}

In the mongo console you will see the document stored

> use test
switched to db test
> show collections
names
system.indexes
> db.names.find()
{ "_id" : ObjectId("53c93bc324ac76925a77b9df"), "firstName" : "mark" }

The mongodb sink has the following options:

authenticationDatabaseName

the MongoDB authentication database used for connecting (String, default: ``)

collectionName

the MongoDB collection to store (String, default: <stream name>)

databaseName

the MongoDB database name (String, default: xd)

host

the MongoDB host to connect to (String, default: localhost)

password

the MongoDB password used for connecting (String, default: ``)

port

the MongoDB port to connect to (int, default: 27017)

username

the MongoDB username used for connecting (String, default: ``)

writeConcern

the default MongoDB write concern to use (WriteConcern, default: SAFE, possible values: NONE,NORMAL,SAFE,FSYNC_SAFE,REPLICAS_SAFE,JOURNAL_SAFE,MAJORITY)

MQTT Sink (mqtt)

The mqtt sink connects to an mqtt server and publishes telemetry messages.

Options

The mqtt sink has the following options:

async

whether or not to use async sends (boolean, default: false)

charset

the charset used to convert a String payload to byte[] (String, default: UTF-8)

cleanSession

whether the client and server should remember state across restarts and reconnects (boolean, default: true)

clientId

identifies the client (String, default: xd.mqtt.client.id.snk)

connectionTimeout

the connection timeout in seconds (int, default: 30)

keepAliveInterval

the ping interval in seconds (int, default: 60)

password

the password to use when connecting to the broker (String, default: guest)

persistence

'memory' or 'file' (String, default: memory)

persistenceDirectory

file location when using 'file' persistence (String, default: /tmp/paho)

qos

the quality of service to use (int, default: 1)

retained

whether to set the 'retained' flag (boolean, default: false)

topic

the topic to which the sink will publish (String, default: xd.mqtt.test)

url

location of the mqtt broker(s) (comma-delimited list) (String, default: tcp://localhost:1883)

username

the username to use when connecting to the broker (String, default: guest)

Note
The defaults are set up to connect to the RabbitMQ MQTT adapter on localhost.

Null Sink (null)

Null sink can be useful when the main stream isn’t focused on a specific destination but drives taps used for analytics etc. It is also useful to iteratively add in steps to a stream without worrying about where the data will end up.

For example,

xd:>stream create nullStream --definition "http | null" --deploy
Created and deployed new stream 'nullStream'
xd:>stream create tap1 --definition "tap:stream:nullStream > counter" --deploy
Created and deployed new stream 'tap1'

In the above, the null sink can be useful as we can create as many number of tap streams off the main stream while we set the main stream sink as null.

RabbitMQ

The "rabbit" sink enables outbound messaging over RabbitMQ.

The following example shows the default settings.

Configure a stream:

xd:> stream create --name rabbittest --definition "time --interval=3 | rabbit" --deploy

This sends the time, every 3 seconds to the default (no-name) Exchange for a RabbitMQ broker running on localhost, port 5672.

The routing key will be the name of the stream by default; in this case: "rabbittest". Since the default Exchange is a direct-exchange to which all Queues are bound with the Queue name as the binding key, all messages sent via this sink will be passed to a Queue named "rabbittest", if one exists. We do not create that Queue automatically. However, you can easily create a Queue using the RabbitMQ web UI. Then, using that same UI, you can navigate to the "rabbittest" Queue and click the "Get Message(s)" button to pop messages off of that Queue (you can choose whether to requeue those messages).

To destroy the stream, enter the following at the shell prompt:

xd:> stream destroy --name rabbittest

RabbitMQ with Options

The rabbit sink has the following options:

addresses

a comma separated list of 'host[:port]' addresses (String, default: ${spring.rabbitmq.addresses})

channelCacheSize

the channel cache size (String, default: ${spring.rabbitmq.channelCacheSize})

converterClass

the class name of the message converter (String, default: org.springframework.amqp.support.converter.SimpleMessageConverter)

deliveryMode

the delivery mode (PERSISTENT, NON_PERSISTENT) (String, default: PERSISTENT)

exchange

the Exchange on the RabbitMQ broker to which messages should be sent (String, default: ``)

keyStore

keyStore location (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.keyStore})

keyStorePassphrase

keyStore passphrase (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.keyStorePassphrase})

mappedRequestHeaders

request message header names to be propagated to/from the adpater/gateway (String, default: STANDARD_REQUEST_HEADERS)

password

the password to use to connect to the broker (String, default: ${spring.rabbitmq.password})

routingKey

the routing key to be passed with the message, as a SpEL expression (String, default: '<stream name>')

sslPropertiesLocation

resource containing SSL properties (String, default: ${spring.rabbitmq.sslProperties})

trustStore

trustStore location (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.trustStore})

trustStorePassphrase

trustStore passphrase (if not using SSL properties) (String, default: ${spring.rabbitmq.ssl.trustStorePassphrase})

useSSL

true if SSL should be used for the connection (String, default: ${spring.rabbitmq.useSSL})

username

the username to use to connect to the broker (String, default: ${spring.rabbitmq.username})

vhost

the RabbitMQ virtual host to use (String, default: ${spring.rabbitmq.virtual_host})

Note

Please be aware that the routingKey option is actually a SpEL expression. Hence if a simple, constant, string literal is to be used, make sure to use something like this:

xd:> stream create rabbitSinkStream --definition "http | rabbit --routingKey='\"myqueue\"'" --deploy

See the RabbitMQ MessageBus Documentation for more information about SSL configuration.

Redis

Redis sink can be used to ingest data into redis store. You can choose queue, topic or key with selcted collection type to point to a specific data store.

For example,

xd:>stream create store-into-redis --definition "http | redis --queue=myList" --deploy
xd:>Created and deployed new stream 'store-into-redis'

Options

The redis sink has the following options:

collectionType

the collection type to use for the given key (CollectionType, default: LIST, possible values: LIST,SET,ZSET,MAP,PROPERTIES)

database

database index used by the connection factory (int, default: 0)

hostname

redis host name (String, default: localhost)

key

name for the key (String, no default)

keyExpression

a SpEL expression to use for keyExpression (String, no default)

maxActive

max number of connections that can be allocated by the pool at a given time; negative value for no limit (int, default: 8)

maxIdle

max number of idle connections in the pool; a negative value indicates an unlimited number of idle connections (int, default: 8)

maxWait

max amount of time (in milliseconds) a connection allocation should block before throwing an exception when the pool is exhausted; negative value to block indefinitely (int, default: -1)

minIdle

target for the minimum number of idle connections to maintain in the pool; only has an effect if it is positive (int, default: 0)

password

redis password (String, default: ``)

port

redis port (int, default: 6379)

queue

name for the queue (String, no default)

queueExpression

a SpEL expression to use for queue (String, no default)

sentinelMaster

name of Redis master server (String, default: ``)

sentinelNodes

comma-separated list of host:port pairs (String, default: ``)

topic

name for the topic (String, no default)

topicExpression

a SpEL expression to use for topic (String, no default)

Shell Sink (shell)

The shell sink forks an external process by running a shell command to launch a process written in any language. The process should implement a continual loop that waits for and consumes input from stdin. The process will be destroyed when the stream is undeployed. For example, it is possible to invoke a Python script within a stream in this manner. Since the shell sink relies on low-level stream processing there are some additional requirements:

  • Input data is expected to be a String, the charset is configurable.

  • Anything written to stderr will be logged as an ERROR in Spring XD but will not terminate the stream.

  • All messages must be terminated using the configured encoder (CRLF or "\r\n" is the default) for the module and must not exceed the configured bufferSize (see the detailed description of encoders in the TCP section).

  • Any external software required to run the script must be installed on the container node to which the module is deployed.

Here is a simple template for a Python script that consumes input:

#sink.py
import sys

while True:
  try:
    data = raw_input()
    if data:
       #insert a function call here, data is a string.
  except EOFError:
      break
Note

Spring XD provides additional Python programming support for handling basic stream processing, as shown above, see creating a Python module.

The shell sink has the following options:

bufferSize

the size of the buffer (bytes) to use when encoding/decoding (int, default: 2048)

charset

the charset used when converting from String to bytes (String, default: UTF-8)

command

the shell command (String, no default)

encoder

the encoder to use when sending messages (Encoding, default: CRLF, possible values: CRLF,LF,NULL,STXETX,RAW,L1,L2,L4)

environment

additional process environment variables as comma delimited name-value pairs (String, no default)

redirectErrorStream

redirects stderr to stdout (boolean, default: false)

workingDir

the process working directory (String, no default)

Splunk Server (splunk)

A Splunk sink that writes data to a TCP Data Input type for Splunk.

Splunk sinks

The Splunk sink converts an object payload to a string using the object’s toString method and then converts this to a SplunkEvent that is sent via TCP to Splunk.

The splunk sink has the following options:

host

the host name or IP address of the Splunk server (String, default: localhost)

owner

the owner of the tcpPort (String, default: admin)

password

the password associated with the username (String, default: password)

port

the TCP port number of the Splunk server (int, default: 8089)

tcpPort

the TCP port number to where XD will send the data (int, default: 9500)

username

the login name that has rights to send data to the tcpPort (String, default: admin)

How To Setup Splunk for TCP Input

  1. From the Manager page select Manage Inputs link

  2. Click the Add data Button

  3. Click the From a TCP port link

  4. TCP Port enter the port you want Splunk to monitor

  5. Set Source Type select Manual

  6. Source Type enter tcp-raw

  7. Click Save

Example

An example stream would be to take data from a twitter search and push it through to a splunk instance.

xd:> stream create --name springone2gx --definition "twittersearch --consumerKey= --consumerSecret= --query='#LOTR' | splunk" --deploy

TCP Sink (tcp)

The TCP Sink provides for outbound messaging over TCP.

The following examples use netcat (linux) to receive the data; the equivalent on Mac OSX is nc.

First, start a netcat to receive the data, and background it

$ netcat -l 1234 &

Now, configure a stream

xd:> stream create --name tcptest --definition "time --fixedDelay=3 | tcp" --deploy

This sends the time, every 3 seconds to the default tcp Sink, which connects to port 1234 on localhost.

$ Thu May 30 10:28:21 EDT 2013
Thu May 30 10:28:24 EDT 2013
Thu May 30 10:28:27 EDT 2013
Thu May 30 10:28:30 EDT 2013
Thu May 30 10:28:33 EDT 2013

TCP is a streaming protocol and some mechanism is needed to frame messages on the wire. A number of encoders are available, the default being CRLF.

Destroy the stream; netcat will terminate when the TCP Sink disconnects.

http://localhost:8080> stream destroy --name tcptest

TCP with Options

The tcp sink has the following options:

bufferSize

the size of the buffer (bytes) to use when encoding/decoding (int, default: 2048)

charset

the charset used when converting from String to bytes (String, default: UTF-8)

close

whether to close the socket after each message (boolean, default: false)

encoder

the encoder to use when sending messages (Encoding, default: CRLF, possible values: CRLF,LF,NULL,STXETX,RAW,L1,L2,L4)

host

the remote host to connect to (String, default: localhost)

nio

whether or not to use NIO (boolean, default: false)

port

the port on the remote host to connect to (int, default: 1234)

reverseLookup

perform a reverse DNS lookup on the remote IP Address (boolean, default: false)

socketTimeout

the timeout (ms) before closing the socket when no data is received (int, default: 120000)

useDirectBuffers

whether or not to use direct buffers (boolean, default: false)

Note
With the default retry configuration, the attempts will be made after 0, 2, 4, 8, and 16 seconds.

Available Encoders

Text Data
CRLF (default)

text terminated by carriage return (0x0d) followed by line feed (0x0a)

LF

text terminated by line feed (0x0a)

NULL

text terminated by a null byte (0x00)

STXETX

text preceded by an STX (0x02) and terminated by an ETX (0x03)

Text and Binary Data
RAW

no structure - the client indicates a complete message by closing the socket

L1

data preceded by a one byte (unsigned) length field (supports up to 255 bytes)

L2

data preceded by a two byte (unsigned) length field (up to 216-1 bytes)

L4

data preceded by a four byte (signed) length field (up to 231-1 bytes)

An Additional Example

Start netcat in the background and redirect the output to a file foo

$ netcat -l 1235 > foo &

Create the stream, using the L4 encoder

xd:> stream create --name tcptest --definition "time --interval=3 | tcp --encoder=L4 --port=1235" --deploy

Destroy the stream

http://localhost:8080> stream destroy --name tcptest

Check the output

$ hexdump -C foo
00000000  00 00 00 1c 54 68 75 20  4d 61 79 20 33 30 20 31  |....Thu May 30 1|
00000010  30 3a 34 37 3a 30 33 20  45 44 54 20 32 30 31 33  |0:47:03 EDT 2013|
00000020  00 00 00 1c 54 68 75 20  4d 61 79 20 33 30 20 31  |....Thu May 30 1|
00000030  30 3a 34 37 3a 30 36 20  45 44 54 20 32 30 31 33  |0:47:06 EDT 2013|
00000040  00 00 00 1c 54 68 75 20  4d 61 79 20 33 30 20 31  |....Thu May 30 1|
00000050  30 3a 34 37 3a 30 39 20  45 44 54 20 32 30 31 33  |0:47:09 EDT 2013|

Note the 4 byte length field preceding the data generated by the L4 encoder.

Counters and Gauges

Counter and Gauges are analytical data structures collectively referred to as metrics. Metrics can be used directly in place of a sink just as if you were creating any other stream, but you can also analyze data from an existing stream using a tap. We’ll look at some examples of using metrics with taps in the following sections. As a prerequisite start the XD Container as instructed in the Getting Started page.

The following types of metrics are available

Spring XD supports these metrics and analytical data structures as a general purpose class library that works with several backend storage technologies. The 1.0 release provides in memory and Redis implementations.

Tip

As of Spring XD 1.2 you can now create data visualizations for the various counters and gauges using the Admin UI. Please see the Admin UI Analytics Chapter for more details.

Counter

A counter is a Metric that associates a unique name with a long value. It is primarily used for counting events triggered by incoming messages on a target stream. You create a counter with a unique name and optionally an initial value then set its value in response to incoming messages. The most straightforward use for counter is simply to count messages coming into the target stream. That is, its value is incremented on every message. This is exactly what the counter module provided by Spring XD does.

Here’s an example:

Start by creating a data ingestion stream. Something like:

xd:> stream create --name springtweets --definition "twittersearch --consumerKey=<your_key> --consumerSecret=<your_secret> --query=spring | file --dir=/tweets/" --deploy

Next, create a tap on the springtweets stream that sets a message counter named tweetcount

xd:> stream create --name tweettap --definition "tap:stream:springtweets > counter --name=tweetcount" --deploy

To retrieve the count:

xd:>counter display tweetcount

The counter sink has the following options:

name

the name of the metric to contribute to (will be created if necessary) (String, default: <stream name>)

nameExpression

a SpEL expression to compute the name of the metric to contribute to (String, no default)

Field Value Counter (field-value-counter)

A field value counter is a Metric used for counting occurrences of unique values for a named field in a message payload. XD Supports the following payload types out of the box:

  • POJO (Java bean)

  • Tuple

  • JSON String

For example suppose a message source produces a payload with a field named user :

class Foo {
   String user;
   public Foo(String user) {
       this.user = user;
   }
}

If the stream source produces messages with the following objects:

   new Foo("fred")
   new Foo("sue")
   new Foo("dave")
   new Foo("sue")

The field value counter on the field user will contain:

fred:1, sue:2, dave:1

Multi-value fields are also supported. For example, if a field contains a list, each value will be counted once:

users:["dave","fred","sue"]
users:["sue","jon"]

The field value counter on the field users will contain:

dave:1, fred:1, sue:2, jon:1

The field-value-counter sink has the following options:

fieldName

the name of the field for which values are counted (String, no default)

name

the name of the metric to contribute to (will be created if necessary) (String, default: <stream name>)

nameExpression

a SpEL expression to compute the name of the metric to contribute to (String, no default)

To try this out, create a stream to ingest twitter feeds containing the word spring and output to a file:

xd:> stream create --name springtweets --definition "twittersearch --consumerKey=<your_key> --consumerSecret=<your_secret> --query=spring | file" --deploy

Now create a tap for a field value counter:

xd:> stream create --name fromUserCount --definition "tap:stream:springtweets > field-value-counter --fieldName=user.screen_name" --deploy

The twittersearch source produces JSON strings which contain the user id of the tweeter in the fromUser field. The field_value_counter sink parses the tweet and updates a field value counter named fromUserCount in Redis. To view the counts:

From xd-shell,
  xd:> field-value-counter display fromUserCount

Aggregate Counter (aggregate-counter)

The aggregate counter differs from a simple counter in that it not only keeps a total value for the count, but also retains the total count values for each minute, hour day and month of the period for which it is run. The data can then be queried by supplying a start and end date and the resolution at which the data should be returned.

Creating an aggregate counter is very similar to a simple counter. For example, to obtain an aggregate count for our spring tweets stream:

xd:> stream create --name springtweets --definition "twittersearch --query=spring | file" --deploy

you’d simply create a tap which pipes the input to aggregate-counter:

xd:> stream create --name tweettap --definition "tap:stream:springtweets > aggregate-counter --name=tweetcount" --deploy

From the XD shell:

xd:> aggregate-counter display tweettap

Note: you can also use some criteria to filter out aggregate counter display values. Please refer to Shell documentation for aggregate counter for more details.

The aggregate-counter sink has the following options:

dateFormat

a pattern (as in SimpleDateFormat) for parsing/formatting dates and timestamps (String, default: yyyy-MM-dd'T'HH:mm:ss.SSS'Z')

incrementExpression

how much to increment each bucket, as a SpEL against the message (String, default: 1)

name

the name of the metric to contribute to (will be created if necessary) (String, default: <stream name>)

nameExpression

a SpEL expression to compute the name of the metric to contribute to (String, no default)

timeField

name of a field in the message that contains the timestamp to contribute to (String, default: null)

Gauge

A gauge is a Metric, similar to a counter in that it holds a single long value associated with a unique name. In this case the value can represent any numeric value defined by the application.

The gauge sink provided with XD stores expects a numeric value as a payload, typically this would be a decimal formatted string.

The gauge sink has the following options:

name

the name of the metric to contribute to (will be created if necessary) (String, default: <stream name>)

nameExpression

a SpEL expression to compute the name of the metric to contribute to (String, no default)

Here is an example of creating a tap for a gauge:

Simple Tap Example

Create an ingest stream

xd:> stream create --name test --definition "http --port=9090 | file" --deploy

Next create the tap:

xd:> stream create --name simplegauge --definition "tap:stream:test > gauge" --deploy

Now Post a message to the ingest stream:

xd:> http post --target http://localhost:9090 --data "10"

Check the gauge:

xd:>gauge display --name simplegauge

Rich Gauge (rich-gauge)

A rich gauge is a Metric that holds a double value associated with a unique name. In addition to the value, the rich gauge keeps a running average, along with the minimum and maximum values and the sample count.

The rich-gauge sink provided with XD expects a numeric value as a payload, typically this would be a decimal formatted string, and keeps its value in a store.

The rich-gauge sink has the following options:

alpha

smoothing constant, or -1 to use arithmetic mean (double, default: -1.0)

name

the name of the metric to contribute to (will be created if necessary) (String, default: <stream name>)

nameExpression

a SpEL expression to compute the name of the metric to contribute to (String, no default)

Note
The smoothing factor behaves as an exponential moving average. The default value does no smoothing.

Here are some examples of creating a tap for a rich gauge:

Simple Tap Example

Create an ingest stream

xd:> stream create --name test --definition "http --port=9090 | file" --deploy

Next create the tap:

xd:> stream create --name testgauge --definition "tap:stream:test > rich-gauge" --deploy

Now Post some messages to the ingest stream:

xd:> http post --target http://localhost:9090 --data "10"
xd:> http post --target http://localhost:9090 --data "13"
xd:> http post --target http://localhost:9090 --data "16"

Check the gauge:

xd:>rich-gauge display testgauge

Stock Price Example

In this example, we will track stock prices, which is a more practical example. The data is ingested as JSON strings like

{"symbol":"VMW","price":72.04}

Create an ingest stream

xd:> stream create --name stocks --definition "http --port=9090 | file"

Next create the tap, using the transform module to extract the stock price from the payload:

xd:> stream create --name stockprice --definition "tap:stream:stocks > transform --expression=#jsonPath(payload,'$.price') | rich-gauge"

Now Post some messages to the ingest stream:

xd:> http post --target http://localhost:9090 --data {"symbol":"VMW","price":72.04}
xd:> http post --target http://localhost:9090 --data {"symbol":"VMW","price":72.06}
xd:> http post --target http://localhost:9090 --data {"symbol":"VMW","price":72.08}

Note: JSON fields should be separated by a comma without any spaces. Alternatively, enclose the whole argument to --data with quotes and escape inner quotes with a backslash.

Check the gauge:

xd:>rich-gauge display stockprice

Improved Stock Price Example

In this example, we will track stock prices for selected stocks. The data is ingested as JSON strings like

{"symbol":"VMW","price":72.04}
{"symbol":"EMC","price":24.92}

The previous example would feed these prices to a single gauge. What we really want is to create a separate tap for each ticker symbol in which we are interested:

Create an ingest stream

xd:> stream create --name stocks --definition "http --port=9090 | file"

Next create the tap, using the transform module to extract the stock price from the payload:

xd:> stream create --name vmwprice --definition "tap:stream:stocks > filter --expression=#jsonPath(payload,'$.symbol')==VMW | transform --expression=#jsonPath(payload,'$.price') | rich-gauge" --deploy
xd:> stream create --name emcprice --definition "tap:stream:stocks > filter --expression=#jsonPath(payload,'$.symbol')==EMC | transform --expression=#jsonPath(payload,'$.price') | rich-gauge" --deploy

Now Post some messages to the ingest stream:

xd:> http post --target http://localhost:9090 --data {"symbol":"VMW","price":72.04}
xd:> http post --target http://localhost:9090 --data {"symbol":"VMW","price":72.06}
xd:> http post --target http://localhost:9090 --data {"symbol":"VMW","price":72.08}
xd:> http post --target http://localhost:9090 --data {"symbol":"EMC","price":24.92}
xd:> http post --target http://localhost:9090 --data {"symbol":"EMC","price":24.90}
xd:> http post --target http://localhost:9090 --data {"symbol":"EMC","price":24.96}

Check the gauge:

xd:>rich-gauge display emcprice
xd:>rich-gauge display vmwprice

Accessing Analytics Data over the RESTful API

Spring XD has a discoverable RESTful API based on the Spring HATEAOS library. You can discover the resources available by making a GET request on the root resource of the Admin server. Here is an example where navigate down to find the data for a counter named httptap that was created by these commands

xd:>stream create --name httpStream --definition "http | file" --deploy
xd:>stream create --name httptap --definition "tap:stream:httpStream > counter" --deploy
xd:>http post --target http://localhost:9000 --data "helloworld"

The root resource returns

xd:>! wget  -q -S -O - http://localhost:9393/
{
  "links":[
    {},
    {
      "rel":"jobs",
      "href":"http://localhost:9393/jobs"
    },
    {
      "rel":"modules",
      "href":"http://localhost:9393/modules"
    },
    {
      "rel":"runtime/modules",
      "href":"http://localhost:9393/runtime/modules"
    },
    {
      "rel":"runtime/containers",
      "href":"http://localhost:9393/runtime/containers"
    },
    {
      "rel":"counters",
      "href":"http://localhost:9393/metrics/counters"
    },
    {
      "rel":"field-value-counters",
      "href":"http://localhost:9393/metrics/field-value-counters"
    },
    {
      "rel":"aggregate-counters",
      "href":"http://localhost:9393/metrics/aggregate-counters"
    },
    {
      "rel":"gauges",
      "href":"http://localhost:9393/metrics/gauges"
    },
    {
      "rel":"rich-gauges",
      "href":"http://localhost:9393/metrics/rich-gauges"
    }
  ]
}

Following the resource location for the counter

xd:>! wget  -q -S -O - http://localhost:9393/metrics/counters
{
  "links":[

  ],
  "content":[
    {
      "links":[
        {
          "rel":"self",
          "href":"http://localhost:9393/metrics/counters/httptap"
        }
      ],
      "name":"httptap"
    }
  ],
  "page":{
    "size":0,
    "totalElements":1,
    "totalPages":1,
    "number":0
  }
}

And then the data for the counter itself

xd:>! wget  -q -S -O - http://localhost:9393/metrics/counters/httptap
{
  "links":[
    {
      "rel":"self",
      "href":"http://localhost:9393/metrics/counters/httptap"
    }
  ],
  "name":"httptap",
  "value":2
}

Jobs

This section describes the job modules included with Spring XD. For a general overview of creating, deploying, and launching batch jobs in Spring XD, see the Batch Jobs section. To run the examples shown here, start the XD Container as instructed in the Getting Started page.

The available jobs are

Note
Regarding HDFS Configuration

To use the hdfs based jobs below, XD needs to have append enabled for hdfs. Update the hdfs-site.xml with the following settings:

    <property>
        <name>dfs.support.append</name>
        <value>true</value>
    </property>

Import CSV Files to HDFS (filepollhdfs)

This module is designed to be driven by a stream polling a directory. It imports data from CSV files and requires that you supply a list of named columns for the data using the names parameter. For example:

xd:> job create myjob --definition "filepollhdfs --names=forename,surname,address" --deploy

You would then use a stream with a file source to scan a directory for files and drive the job. A separate job will be started for each file found:

xd:> stream create csvStream --definition "file --mode=ref --dir=/mycsvdir --pattern=*.csv > queue:job:myjob" --deploy

The filepollhdfs job has the following options:

commitInterval

the commit interval to be used for the step (int, default: 1000)

deleteFiles

whether to delete files after successful import (boolean, default: false)

directory

the directory to write the file(s) to in HDFS (String, default: /xd/<job name>)

fileExtension

the file extension to use (String, default: csv)

fileName

the filename to use in HDFS (String, default: <job name>)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

names

the field names in the CSV file (String, no default)

restartable

whether the job should be restartable or not in case of failure (boolean, default: false)

rollover

the number of bytes to write before creating a new file in HDFS (int, default: 1000000)

Import CSV Files to JDBC (filejdbc)

A module which loads CSV files into a JDBC table using a single batch job. By default it uses the internal HSQL DB which is used by Spring Batch. Refer to how module options are resolved for further details on how to change defaults (one can of course always use --foo=bar notation in the job definition to achieve the same effect).

Note
If you access any database other than HSQLDB or Postgres in a job module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.

The filejdbc job has the following options:

abandonWhenPercentageFull

connections that have timed out wont get closed and reported up unless the number of connections in use are above the percentage (int, default: 0)

alternateUsernameAllowed

uses an alternate user name if connection fails (boolean, default: false)

commitInterval

the commit interval to be used for the step (int, default: 1000)

connectionProperties

connection properties that will be sent to our JDBC driver when establishing new connections (String, no default)

deleteFiles

whether to delete files after successful import (boolean, default: false)

delimiter

the delimiter for the delimited file (String, default: ,)

driverClassName

the JDBC driver to use (String, no default)

fairQueue

set to true if you wish that calls to getConnection should be treated fairly in a true FIFO fashion (boolean, default: true)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

initSQL

custom query to be run when a connection is first created (String, no default)

initialSize

initial number of connections that are created when the pool is started (int, default: 0)

initializeDatabase

whether the database initialization script should be run (boolean, default: false)

initializerScript

the name of the SQL script (in /config) to run if 'initializeDatabase' is set (String, default: init_batch_import.sql)

jdbcInterceptors

semicolon separated list of classnames extending org.apache.tomcat.jdbc.pool.JdbcInterceptor (String, no default)

jmxEnabled

register the pool with JMX or not (boolean, default: true)

logAbandoned

flag to log stack traces for application code which abandoned a Connection (boolean, default: false)

maxActive

maximum number of active connections that can be allocated from this pool at the same time (int, default: 100)

maxAge

time in milliseconds to keep this connection (int, default: 0)

maxIdle

maximum number of connections that should be kept in the pool at all times (int, default: 100)

maxWait

maximum number of milliseconds that the pool will wait for a connection (int, default: 30000)

minEvictableIdleTimeMillis

minimum amount of time an object may sit idle in the pool before it is eligible for eviction (int, default: 60000)

minIdle

minimum number of established connections that should be kept in the pool at all times (int, default: 10)

names

the field names in the CSV file (String, no default)

partitionResultsTimeout

time (ms) that the partition handler will wait for results (long, default: 3600000)

password

the JDBC password (Password, no default)

removeAbandoned

flag to remove abandoned connections if they exceed the removeAbandonedTimout (boolean, default: false)

removeAbandonedTimeout

timeout in seconds before an abandoned connection can be removed (int, default: 60)

resources

the list of paths to import (Spring resources) (String, no default)

restartable

whether the job should be restartable or not in case of failure (boolean, default: false)

suspectTimeout

this simply logs the warning after timeout, connection remains (int, default: 0)

tableName

the database table to which the data will be written (String, default: <job name>)

testOnBorrow

indication of whether objects will be validated before being borrowed from the pool (boolean, default: false)

testOnReturn

indication of whether objects will be validated before being returned to the pool (boolean, default: false)

testWhileIdle

indication of whether objects will be validated by the idle object evictor (boolean, default: false)

timeBetweenEvictionRunsMillis

number of milliseconds to sleep between runs of the idle connection validation/cleaner thread (int, default: 5000)

url

the JDBC URL for the database (String, no default)

useEquals

true if you wish the ProxyConnection class to use String.equals (boolean, default: true)

username

the JDBC username (String, no default)

validationInterval

avoid excess validation, only run validation at most at this frequency - time in milliseconds (long, default: 30000)

validationQuery

sql query that will be used to validate connections from this pool (String, no default)

validatorClassName

name of a class which implements the org.apache.tomcat.jdbc.pool.Validator (String, no default)

The job should be defined with the resources parameter defining the files which should be loaded. It also requires a names parameter (for the CSV field names) and these should match the database column names into which the data should be stored. You can either pre-create the database table or the module will create it for you if you use --initializeDatabase=true when the job is created. The table initialization is configured in a similar way to the JDBC sink and uses the same parameters. The default table name is the job name and can be customized by setting the tableName parameter. As an example, if you run the command

xd:> job create myjob --definition "filejdbc --resources=file:///mycsvdir/*.csv --names=forename,surname,address --tableName=people --initializeDatabase=true" --deploy

it will create the table "people" in the database with three varchar columns called "forename", "surname" and "address". When you launch the job it will load the files matching the resources pattern and write the data to this table. As with the filepollhdfs job, this module also supports the deleteFiles parameter which will remove the files defined by the resources parameter on successful completion of the job.

Launch the job using:

xd:> job launch myjob
Tip
The connection pool settings for xd are located in servers.yml (i.e. spring.datasource.* )

Import FTP to HDFS (ftphdfs)

Copies files from FTP directory into HDFS. Job is partitioned in a way that each separate file copy is executed on its own partitioned step.

An example which copies files:

job create --name ftphdfsjob --definition "ftphdfs --host=ftp.example.com --port=21" --deploy
job launch --name ftphdfsjob --params {"remoteDirectory":"/pub/files","hdfsDirectory":"/ftp"}

Full path is preserved so that above command would result files in HDFS shown below:

/ftp/pub/files
/ftp/pub/files/file1.txt
/ftp/pub/files/file2.txt

The ftphdfs job has the following options:

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

host

the host name for the FTP server (String, default: localhost)

partitionResultsTimeout

time (ms) that the partition handler will wait for results (long, default: 3600000)

password

the password for the FTP connection (Password, no default)

port

the port for the FTP server (int, default: 21)

restartable

whether the job should be restartable or not in case of failure (boolean, default: false)

username

the username for the FTP connection (String, no default)

Running gpload as a batch job (gpload)

The gpload utility can be deployed and launched from Spring XD as a batch job. The gpload job uses a GploadTasklet that submits a gpload job as an external process. The Spring XD gpload batch job aims to support most of the gpload functionality.

We need to provide the following required options:

  • gploadHome - this must be the path to where gpload utility is installed. This is usually /usr/local/greenplum-loaders-<version>.

  • controlFile - this file defines the gpload options in effect for this load job and is documented in the Greenplum Load Tools Reference documentation.

  • password or passswordFile - you can either speciy the passord or provide a password file that must follow the general format for a PostgreSQL password file.

Here is an example of a basic load job definition. Please note that some options like host, port, database and username could have been specified in the control file as well.

The content of the control file:

VERSION: 1.0.0.1
GPLOAD:
   INPUT:
    - SOURCE:
        FILE: [/home/demo/data/test_file.csv]
    - FORMAT: CSV
    - DELIMITER: ','
    - NULL_AS: '\N'
    - QUOTE: '"'
    - HEADER: FALSE
    - ENCODING: 'UTF8'
    - ERROR_LIMIT: 1000
    - ERROR_TABLE: public.err_table
   OUTPUT:
    - TABLE: demo.test
    - MODE: INSERT
   PRELOAD:
    - TRUNCATE: FALSE
    - REUSE_TABLES: FALSE

This is the command used to create and launch the job:

xd:>job create myload --definition "gpload --gploadHome=/usr/local/greenplum-loaders-4.3.4.1-build-2 --controlFile=/home/demo/basic.yml --host=pivhdsne --port=5432 --database=pivotal --username=gpadmin --passwordFile=/home/demo/.pgpass" --deploy
xd:>job launch --name myload

Once the job is launched, go to Spring XD admin-ui to verify the job results. Jobs → Executions → Select the job to verify that step execution context holds the log for gpload execution results.

We can override the file name for the source file by providing it as a job parameter like this:

job launch --name myload --params {"input.source.file":"/home/demo/data/inputfile2.csv"}

This allows us to define a stream to capture new files created in a specific directory:

xd>stream create loadFiles --definition "file --ref=true --dir=/home/demo/input --pattern='*.csv' | transform --expression='{\"input.source.file\":\"'+#{'payload.getAbsolutePath()'}+'\"}' > queue:job:myload" --deploy

Now, any new file created in that directory will launch a gpload job for that new file.

The gpload job has the following options:

controlFile

path to the gpload control file (String, no default)

database

the name of the database to load into (String, no default)

gploadHome

the gpload home location (String, no default)

host

the host name for the Greenplum master database server (String, no default)

options

the gpload options to use (String, no default)

password

the password to use when connecting (String, no default)

passwordFile

the location of the password file (String, no default)

port

the port for the Greenplum master database server (Integer, no default)

username

the username to connect as (String, no default)

Export HDFS to JDBC (hdfsjdbc)

This module functions very similarly to the filejdbc one except that the resources you specify should actually be in HDFS, rather than the OS filesystem.

xd:> job create myjob --definition "hdfsjdbc --resources=/xd/data/*.csv --names=forename,surname,address --tableName=people --initializeDatabase=true" --deploy

Launch the job using:

xd:> job launch myjob
Note
If you access any database other than HSQLDB or Postgres in a job module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.

The hdfsjdbc job has the following options:

abandonWhenPercentageFull

connections that have timed out wont get closed and reported up unless the number of connections in use are above the percentage (int, default: 0)

alternateUsernameAllowed

uses an alternate user name if connection fails (boolean, default: false)

commitInterval

the commit interval to be used for the step (int, default: 1000)

connectionProperties

connection properties that will be sent to our JDBC driver when establishing new connections (String, no default)

delimiter

the delimiter for the delimited file (String, default: ,)

driverClassName

the JDBC driver to use (String, no default)

fairQueue

set to true if you wish that calls to getConnection should be treated fairly in a true FIFO fashion (boolean, default: true)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

initSQL

custom query to be run when a connection is first created (String, no default)

initialSize

initial number of connections that are created when the pool is started (int, default: 0)

initializeDatabase

whether the database initialization script should be run (boolean, default: false)

initializerScript

the name of the SQL script (in /config) to run if 'initializeDatabase' is set (String, default: init_batch_import.sql)

jdbcInterceptors

semicolon separated list of classnames extending org.apache.tomcat.jdbc.pool.JdbcInterceptor (String, no default)

jmxEnabled

register the pool with JMX or not (boolean, default: true)

logAbandoned

flag to log stack traces for application code which abandoned a Connection (boolean, default: false)

maxActive

maximum number of active connections that can be allocated from this pool at the same time (int, default: 100)

maxAge

time in milliseconds to keep this connection (int, default: 0)

maxIdle

maximum number of connections that should be kept in the pool at all times (int, default: 100)

maxWait

maximum number of milliseconds that the pool will wait for a connection (int, default: 30000)

minEvictableIdleTimeMillis

minimum amount of time an object may sit idle in the pool before it is eligible for eviction (int, default: 60000)

minIdle

minimum number of established connections that should be kept in the pool at all times (int, default: 10)

names

the field names in the CSV file (String, no default)

password

the JDBC password (Password, no default)

removeAbandoned

flag to remove abandoned connections if they exceed the removeAbandonedTimout (boolean, default: false)

removeAbandonedTimeout

timeout in seconds before an abandoned connection can be removed (int, default: 60)

resources

the list of paths to import (Spring resources) (String, no default)

restartable

whether the job should be restartable or not in case of failure (boolean, default: false)

suspectTimeout

this simply logs the warning after timeout, connection remains (int, default: 0)

tableName

the database table to which the data will be written (String, default: <job name>)

testOnBorrow

indication of whether objects will be validated before being borrowed from the pool (boolean, default: false)

testOnReturn

indication of whether objects will be validated before being returned to the pool (boolean, default: false)

testWhileIdle

indication of whether objects will be validated by the idle object evictor (boolean, default: false)

timeBetweenEvictionRunsMillis

number of milliseconds to sleep between runs of the idle connection validation/cleaner thread (int, default: 5000)

url

the JDBC URL for the database (String, no default)

useEquals

true if you wish the ProxyConnection class to use String.equals (boolean, default: true)

username

the JDBC username (String, no default)

validationInterval

avoid excess validation, only run validation at most at this frequency - time in milliseconds (long, default: 30000)

validationQuery

sql query that will be used to validate connections from this pool (String, no default)

validatorClassName

name of a class which implements the org.apache.tomcat.jdbc.pool.Validator (String, no default)

Tip
The connection pool settings for xd are located in servers.yml (i.e. spring.datasource.* )

Export HDFS to MongoDB (hdfsmongodb)

Exports CSV data from HDFS and stores it in a MongoDB collection which defaults to the job name. This can be overridden with the collectionName parameter. Once again, the field names should be defined by supplying the names parameter. The data is converted internally to a Spring XD Tuple and the collection items will have an id matching the tuple’s UUID. You can override this by setting the idField parameter to one of the field names if desired.

An example:

xd:> job create myjob --definition "hdfsmongodb --resources=/data/*.log --names=employeeId,forename,surname,address --idField=employeeId --collectionName=people" --deploy

The hdfsmongodb job has the following options:

authenticationDatabaseName

the MongoDB authentication database used for connecting (String, default: ``)

collectionName

the MongoDB collection to store (String, default: <job name>)

commitInterval

the commit interval to be used for the step (int, default: 1000)

databaseName

the MongoDB database name (String, default: xd)

delimiter

the delimiter for the delimited file (String, default: ,)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

host

the MongoDB host to connect to (String, default: localhost)

idField

the name of the field to use as the identity in MongoDB (String, no default)

names

the field names in the CSV file (String, no default)

password

the MongoDB password used for connecting (String, default: ``)

port

the MongoDB port to connect to (int, default: 27017)

resources

the list of paths to import (Spring resources) (String, no default)

restartable

whether the job should be restartable or not in case of failure (boolean, default: false)

username

the MongoDB username used for connecting (String, default: ``)

writeConcern

the default MongoDB write concern to use (WriteConcern, default: SAFE, possible values: NONE,NORMAL,SAFE,FSYNC_SAFE,REPLICAS_SAFE,JOURNAL_SAFE,MAJORITY)

Import JDBC to HDFS (jdbchdfs)

Performs the reverse of the previous module. The database configuration is the same as for filejdbc but without the initialization options since you need to already have the data to import into HDFS. When creating the job, you must either supply the select statement by setting the sql parameter, or you can supply both tableName and columns options (which will be used to build the SQL statement).

To import data from the database table some_table, you could use

xd:> job create myjob --definition "jdbchdfs --sql='select col1,col2,col3 from some_table'" --deploy

You can customize how the data is written to HDFS by supplying the options directory (defaults to /xd/(job name)), fileName (defaults to job name), rollover (in bytes, default 1000000) and fileExtension (defaults to csv).

Launch the job using:

xd:> job launch myjob
Note
If you access any database other than HSQLDB or Postgres in a job module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.

If you want to partition your job across multiple XD containers you can provide the partitionColumn and partitions option. When the job is launched the partitioner will query the database for the range of values and evenly divide the load between the partitions. This assumes that there is an even distribution of column values in the table. When using the partitioning support you must also use the tableName and columns options instead of the sql option. This is so the partitioner can construct the queries with the appropriate where clauses for the different partitions.

An example of a partitioned job could look like this:

xd:> job create partitionedJob --definition "jdbchdfs --columns='id,col1,col2' --tableName=some_table --partitionColumn=id --partitions=4" --deploy
Note
When using the partitioning support you can not use the sql option. Use tableName and columns instead.

You can perform incremental imports using this job by defining a column to check against. Currently the column must be numeric (similar to how the partitionColumn works). An example of launching a job that performs incremental imports would look like the following:

xd:> job create incrementalImportJob --definition "jdbchdfs --columns='id,col1,col2' --tableName=some_table --checkColumn=sequence --restartable=true" --deploy

If you want to specify the value for the checkColumn, you can pass the override value in as a JobParameter named overrideCheckColumnValue as shown below:

xd:> job launch incrementalImportJob --params {"overrideCheckColumnValue" : 2}

There are two things to keep in mind when using incremental imports with this job:

  • When using incremental imports, the sql option is not available. Use tableName and columns instead.

  • If an import fails, it must be rerun to completion before running the next import. Without this, inconsistent data may result. Since HDFS is a non-transactional store, failed records may not be rolled back. An administrator may need to check HDFS for completeness and the last imported value.

The jdbchdfs job has the following options:

abandonWhenPercentageFull

connections that have timed out wont get closed and reported up unless the number of connections in use are above the percentage (int, default: 0)

alternateUsernameAllowed

uses an alternate user name if connection fails (boolean, default: false)

checkColumn

the column to be examined when determining which rows to import (String, default: ``)

columns

the column names to read from the supplied table (String, default: ``)

commitInterval

the commit interval to be used for the step (int, default: 1000)

connectionProperties

connection properties that will be sent to our JDBC driver when establishing new connections (String, no default)

delimiter

the delimiter for the delimited file (String, default: ,)

directory

the directory to write the file(s) to in HDFS (String, default: /xd/<job name>)

driverClassName

the JDBC driver to use (String, no default)

fairQueue

set to true if you wish that calls to getConnection should be treated fairly in a true FIFO fashion (boolean, default: true)

fileExtension

the file extension to use (String, default: csv)

fileName

the filename to use in HDFS (String, default: <job name>)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

initSQL

custom query to be run when a connection is first created (String, no default)

initialSize

initial number of connections that are created when the pool is started (int, default: 0)

jdbcInterceptors

semicolon separated list of classnames extending org.apache.tomcat.jdbc.pool.JdbcInterceptor (String, no default)

jmxEnabled

register the pool with JMX or not (boolean, default: true)

logAbandoned

flag to log stack traces for application code which abandoned a Connection (boolean, default: false)

maxActive

maximum number of active connections that can be allocated from this pool at the same time (int, default: 100)

maxAge

time in milliseconds to keep this connection (int, default: 0)

maxIdle

maximum number of connections that should be kept in the pool at all times (int, default: 100)

maxWait

maximum number of milliseconds that the pool will wait for a connection (int, default: 30000)

minEvictableIdleTimeMillis

minimum amount of time an object may sit idle in the pool before it is eligible for eviction (int, default: 60000)

minIdle

minimum number of established connections that should be kept in the pool at all times (int, default: 10)

partitionColumn

the column to use for partitioning, should be numeric and uniformly distributed (String, default: ``)

partitionResultsTimeout

time (ms) that the partition handler will wait for results (long, default: 3600000)

partitions

the number of partitions (int, default: 1)

password

the JDBC password (Password, no default)

removeAbandoned

flag to remove abandoned connections if they exceed the removeAbandonedTimout (boolean, default: false)

removeAbandonedTimeout

timeout in seconds before an abandoned connection can be removed (int, default: 60)

restartable

whether the job should be restartable or not in case of failure (boolean, default: false)

rollover

the number of bytes to write before creating a new file in HDFS (int, default: 1000000)

sql

the SQL to use to extract data (String, default: ``)

suspectTimeout

this simply logs the warning after timeout, connection remains (int, default: 0)

tableName

the table to read data from (String, default: ``)

testOnBorrow

indication of whether objects will be validated before being borrowed from the pool (boolean, default: false)

testOnReturn

indication of whether objects will be validated before being returned to the pool (boolean, default: false)

testWhileIdle

indication of whether objects will be validated by the idle object evictor (boolean, default: false)

timeBetweenEvictionRunsMillis

number of milliseconds to sleep between runs of the idle connection validation/cleaner thread (int, default: 5000)

url

the JDBC URL for the database (String, no default)

useEquals

true if you wish the ProxyConnection class to use String.equals (boolean, default: true)

username

the JDBC username (String, no default)

validationInterval

avoid excess validation, only run validation at most at this frequency - time in milliseconds (long, default: 30000)

validationQuery

sql query that will be used to validate connections from this pool (String, no default)

validatorClassName

name of a class which implements the org.apache.tomcat.jdbc.pool.Validator (String, no default)

Tip
The connection pool settings for xd are located in servers.yml (i.e. spring.datasource.* )

Running Spark application as a batch job (sparkapp)

A Spark Application can be deployed and launched from Spring XD as a batch job. SparkTasklet submits the Spark application into Spark cluster manager using org.apache.spark.deploy.SparkSubmit. Through this approach, you can also launch a Spark application with specific criteria via Spring XD stream (for instance: A real time scoring algorithm through MLlib spark job can be triggered based on the streaming data events). To get started, please refer to Spark examples here: https://spark.apache.org/examples.html.

Note
The current Spark release that is supported is Spark 1.2.1

Lets run some Spark examples as Spring XD batch jobs:

xd:>job create SparkPiExample --definition "sparkapp --appJar=<the location of spark-examples-1.2.1 jar> --name=MyApp --master=<spark master url or local> --mainClass=org.apache.spark.examples.SparkPi" --deploy
xd:>job launch SparkPiExample
xd:>job create JavaWordCountExample --definition "sparkapp --appJar=<the location of spark-examples-1.2.1 jar> --name=MyApp --master=<spark master url or local> --mainClass=org.apache.spark.examples.JavaWordCount --programArgs=<location of the file to count the words>" --deploy
xd>job launch JavaWordCountExample

Once the job is launched, go to Spring XD admin-ui to verify the job results. Jobs → Executions → Select the job to verify that execution context holds the log for Spark application results. If you launch the Spark application through Spark Master, then the results and application status can be verified from SparkUI as well.

The sparkapp job has the following options:

appJar

path to a bundled jar that includes your application and its dependencies - excluding spark (String, no default)

conf

comma seperated list of key value pairs as config properties (String, default: ``)

files

comma separated list of files to be placed in the working directory of each executor (String, default: ``)

mainClass

the main class for Spark application (String, no default)

master

the master URL for Spark (String, default: local)

name

the name of the Spark application (String, default: ``)

programArgs

program arguments for the application main class (String, default: ``)

Running Sqoop as a batch job (sqoop)

A Sqoop job can be deployed and launched from Spring XD as a batch job. The Sqoop job uses a SqoopTasklet and a SqoopRunner that submits a Sqoop job using org.apache.sqoop.Sqoop.runTool. The Spring XD Sqoop batch job aims to support most of the Sqoop functionality, but at this point we have only tested a subset:

  • import

  • export

  • codegen

  • merge

  • job

  • list-tables

Note
The current release supports Sqoop 1.4.5

The intention is to eventually support all features of the Sqoop tool. See Sqoop User Guide for full documentation of the Sqoop features.

Note
If you access any database other than HSQLDB or Postgres in a job module then the JDBC driver jar for that database needs to be present in the $XD_HOME/lib directory.

We can test the Sqoop job by just listing the tables in the database:

xd:>job create sqoopListTables --definition "sqoop --command=list-tables" --deploy
xd:>job launch --name sqoopListTables

The definition contains the name of the provided job as sqoop and the --command option names the Sqoop command we want to run, which in this case is "list-tables".

Once the job is launched, go to Spring XD admin-ui to verify the job results. Jobs → Executions → Select the job to verify that step execution context holds the log for Sqoop Tool execution results. You should see some tables listed there. Since we didn’t provide any connection arguments Spring XD will by default use the batch respoitory database for the Sqoop Tool execution. We could provide options specifying a different database using the --url, --username and --password options for the job:

xd:>job create sqoopListTables2 --definition "sqoop --command=list-tables --url=jdbc:mysql://localhost:3306/test --username=myuser --password=mypasswd" --deploy
xd:>job launch --name sqoopListTables2

Here we connect to a local MySQL database. It’s important to note that you need to provide the MySQL JDBC driver jar in the Spring XD lib directory for this to work.

There also is an option to specify connection arguments using the --args option. This allows you to use the same arguments that you are used to provide on the command line when running the Sqoop Tool directly. To connect to the same MySQL database as above using --args we would use:

xd:>job create sqoopListTables3 --definition "sqoop --command=list-tables --args='--connect=jdbc:mysql://localhost:3306/test --username=myuser --password=mypasswd'" --deploy
xd:>job launch --name sqoopListTables3

When importing data, you simply use "import" as the command to run. Here is an example:

xd:>job create sqoopImport1 --definition "sqoop --command=import --args='--table=MYTABLE' --url=jdbc:mysql://localhost:3306/test --username=myuser --password=mypasswd" --deploy
xd:>job launch --name sqoopImport1

In this example we provided the connection arguments using the -args option. We could also have used --url, --username and --password options like we did above for the "list-tables" example. The "import" command will use the spring.hadoop.fsUri that is specified when Spring XD starts up. You can override this by providing the --fsUri option when defining the job. The same is true for spring.hadoop.resourceManagerHost and spring.hadoop.resourceManagerPort. You can override the Spring XD configured values with --resourceManagerHost and --resourceManagerPort options.

For exports we use the "export" command. Here is an example:

xd:>job create sqoopExport1 --definition "sqoop --command=export --args='--table=NEWTABLE --export-dir=/user/xduser/MYTABLE'" --deploy
xd:>job launch --name sqoopExport1

Here we rely on the connection options to default to the same database used for the batch repository. Note that Sqoop requires that the table to export data into must already exist.

Note
If your Sqoop args are more complex, as is the case when you provide a query expression or a where clause, then you will need to use escaping for double quotes used within the --args option. A quick example of using a where clause:
job create sqoopComplexArgs1 --definition "sqoop --command=import --args='--table MYFILES --where \"ID < 390000\" --target-dir /user/xduser/TEST --split-by ID'"

(For this example we have omitted the equal sign for the individual Sqoop arguments within the --args option. Either style works fine.)

Note
If your Sqoop args use escape sequences (common when working with Hive data) then you should provide double back-slash characters when working with the XD Shell (this effectively escapes the escape character and only one back-slash will be passed on). Here is a brief example:
job create sqoopHiveArgs1 --definition "sqoop --command=import --args='--table MYFILES --target-dir /user/xduser/TEST --split-by ID --null-string \\\\N --fields-terminated-by \\0001'"

For more detailed coverage of using quotes and escaping please see Single quotes, Double quotes, Escaping.

Note
Some JDBC drivers and some compression codecs require additional jars to work properly. If this is the case, then you can use the --libjars option to provide a comma separated list of jars to be added to the job execution. You should only specify the name of the jar and not the full path. The jars will be looked up from the classpath and included in the job submitted to the Hadoop cluster.
Note
When using Sqoop’s --as-avrodatafile argument we will automatically include the Avro jars in the Sqoop job submission. No need to specify them as part of the --libjars option.
Note
Advanced Hadoop configuration options can be provided in one of several configuration files. The hadoop-site.xml file is only used by the Sqoop job while the other configuration files are used by all Hadoop related jobs and streams:
  • $XD_HOME/config/hadoop.properties — just add the property you would like to set:

    dfs.client.socket-timeout=20000
  • $XD_HOME/config/hadoop-site.xml — add a property entry:

        <property>
          <name>dfs.client.socket-timeout</name>
          <value>20000</value>
        </property>
  • $XD_HOME/config/servers.yml — add a spring.hadoop.config entry:

    spring:
      hadoop:
        config:
          dfs.client.socket-timeout: 20000

Using Sqoop’s metastore

It is possible to use Sqoop’s metastore with some restrictions.

Warning
Sqoop ships with HSQLDB version 1.8 and Spring XD ships with HSQLDB version 2.3. Since these two versions are not compatible you can not use a Sqoop metastore that uses HSQLDB. This is unfortunate since HSQLDB version 1.8 is the only database that is fully supported for the metastore by Sqoop. We can however use another database for the metastore as long as we use some workarounds.
Note
You can use PostgreSQL for the Sqoop metastore. We recommend that you run the commands listed below to create and initialize the tables to be used by the Sqoop metastore.

Create and initialize the Sqoop metastore tables:

CREATE TABLE
    SQOOP_ROOT
    (
        version INTEGER,
        propname CHARACTER VARYING(128) NOT NULL,
        propval CHARACTER VARYING(256),
        UNIQUE (version, propname)
    );
CREATE TABLE
    SQOOP_SESSIONS
    (
        job_name CHARACTER VARYING(64) NOT NULL,
        propname CHARACTER VARYING(128) NOT NULL,
        propval CHARACTER VARYING(1024),
        propclass CHARACTER VARYING(32) NOT NULL,
        UNIQUE (job_name, propname, propclass)
    );
INSERT INTO sqoop_root (version, propname, propval) VALUES (null, 'sqoop.hsqldb.job.storage.version', '0');
INSERT INTO sqoop_root (version, propname, propval) VALUES (0, 'sqoop.hsqldb.job.info.table', 'SQOOP_SESSIONS');

You can now modify the scoop-site.xml file in the Spring XD config directory. Add the JDBC URL, username and password to use for connection to the PostgreSQL database that hosts the Sqoop metastore tables. You need to provide the following properties:

  • sqoop.metastore.client.autoconnect.url

  • sqoop.metastore.client.autoconnect.username

  • sqoop.metastore.client.autoconnect.password

Note
In addition to the above configurations you need to use a --password-file option when creating the Sqoop job definitions. If you don’t then Sqoop will prompt for a password as Spring XD runs the job. This will cause the job to hang.

Here is an example of defining a Sqoop job using Spring XD’s sqoop job:

xd>job create job1create --definition "sqoop --command=job --args='--create job1 -- import --table PETS --incremental append --check-column ID --last-value 0 --connect jdbc:hsqldb:hsql://localhost:9001/test --username sa --password-file /xd/hsql.password --target-dir /xd/job1 --num-mappers 1'" --deploy
xd>job launch job1create

Here is an example of executing the predefined Sqoop job using Spring XD’s sqoop job:

xd>job create job1exec --definition "sqoop --command=job --args='--exec job1'" --deploy
xd>job launch job1exec

Options for Sqoop job

The sqoop job has the following options:

args

the arguments for the Sqoop command (String, default: ``)

command

the Sqoop command to run (String, default: ``)

driverClassName

the JDBC driver to use (String, no default)

fsUri

the URI to use to access the Hadoop FileSystem (String, default: ${spring.hadoop.fsUri})

libjars

extra jars from the classpath to add to the job (String, default: ``)

password

the JDBC password (Password, no default)

resourceManagerHost

the Host for Hadoop's ResourceManager (String, default: ${spring.hadoop.resourceManagerHost})

resourceManagerPort

the Port for Hadoop's ResourceManager (String, default: ${spring.hadoop.resourceManagerPort})

url

the JDBC URL for the database (String, no default)

username

the JDBC username (String, no default)

Taps

Introduction

A Tap allows you to "listen" to data while it is processed in an existing stream and process the data in a separate stream. The original stream is unaffected by the tap and isn’t aware of its presence, similar to a phone wiretap. (WireTap is included in the standard catalog of EAI patterns and implemented in the Spring Integration EAI framework used by Spring XD).

Simply put, a Tap is a stream that uses a point in another stream as a source.

Example

The following XD shell commands create a stream foo1 and a tap named foo1tap:

xd:> stream create --name foo1 --definition "time | log" --deploy
xd:> stream create --name foo1tap --definition "tap:stream:foo1 > log" --deploy

Since a tap is a type of stream, use the stream create command to create the tap. The tap source is specified using the named channel syntax and always begins with tap:. In this case, we are tapping the stream named foo1 specified by :stream:foo1

Note
stream: is required in this case as it is possible to tap alternate XD targets such as jobs. This tap consumes data at the source of the target stream.

A tap can consume data from any point along the target stream’s processing pipeline. XD provides a few ways to tap a stream after a given processor has been applied:

Example - tap after a processor has been applied

If the module name is unique in the target stream, use tap:stream:<stream_name>.<module_name>

If you have a stream called mystream, defined as

http | filter --expression=payload.startsWith('A') | transform --expression=payload.toLowerCase() | file

Create a tap after the filter is applied using

tap:stream:mystream.filter > ....

Example - using a label

You may also use labels to create an alias for a module and reference the label in the tap

If you have a stream called mystream, defined as

http | transform --expression=payload.toLowerCase() | flibble: transform --expression=payload.substring(3) | file

Create a tap after the second transformer is applied using

tap:stream:mystream.flibble > ....

A primary use case for a Tap is to perform realtime analytics at the same time as data is being ingested via its primary stream. For example, consider a Stream of data that is consuming Twitter search results and writing them to HDFS. A tap can be created before the data is written to HDFS, and the data piped from the tap to a counter that correspond to the number of times specific hashtags were mentioned in the tweets.

Creating a tap on a named channel, a stream whose source is a named channel, or a label is not yet supported. This is planned for a future release.

You’ll find specific examples of creating taps on existing streams in the Analytics section.

Note
In cases where a multiple modules with the same module name, a label must be specified on the module to be tapped. For example if you want to tap the 2nd transform: http | transform --expression=payload.toLowerCase() | tapMe: transform --expression=payload.substring(3) | file

Tap Lifecycle

A side effect of a stream being unaware of any taps on its pipeline is that deleting the stream will not automatically delete the taps. The taps have to be deleted separately. However if the tapped stream is re-created, the existing tap will continue to function.

Analytics

Introduction

Spring XD provides support for the real-time evaluation of various machine learning scoring algorithms as well simple real-time data analytics using various types of counters and gauges. The analytics functionality is provided via modules that can be added to a stream. In that sense, real-time analytics is accomplished via the same exact model as data-ingestion. It’s possible that the primary role of a stream is to perform real-time analytics, but it’s quite common to add a tap to initiate a secondary stream where analytics, e.g. a field-value-counter, is applied to the same data being ingested through a primary stream. You will see both approaches in the examples below.

Predictive analytics

Spring XD’s support for implementing predictive analytics by scoring analytical models that leverage machine learning algorithms begins with an extensible class library foundation upon which implementations can be built, such as the PMML Module that we describe here.

That module integrates with the JPMML-Evaluator library that provides support for a wide range of model types and is interoperable with models exported from R, Rattle, KNIME, and RapidMiner. For counter and gauge analytics, in-memory and Redis implementations are provided.

Incorporating the evaluation of machine learning algorithms into stream processing is as easy as using any other processing module. Here is a simple example

http --outputType=application/x-xd-tuple | analytic-pmml
     --location=/models/iris-flower-naive-bayes.pmml.xml
     --inputFieldMapping=
       'sepalLength:Sepal.Length,
        sepalWidth:Sepal.Width,
        petalLength:Petal.Length,
        petalWidth:Petal.Width'
     --outputFieldMapping='Predicted_Species:predictedSpecies' | log"

The http source converts posted data to a Tuple. The analytic-pmml processor loads the model from the specifed file and creates two mappings so that fields from the Tuple can be mapped into the input and output model names. The log sink writes the payload of the event message to the log file of the XD container.

Posting the following JSON data to the http source

{
  "sepalLength": "6.4",
  "sepalWidth":  "3.2",
  "petalLength": "4.5",
  "petalWidth":  "1.5"
}

will produce output in the log file as shown below.

{
   "id":"1722ec00-baad-11e3-b988-005056c00008",
   "timestamp":1396473833152,
   "sepalLength":"6.4",
   "sepalWidth":"3.2",
   "petalLength":"4.5",
   "petalWidth":"1.5",
   "predictedSpecies":"versicolor"
}

The next section on analytical models goes into more detail on the general infrastructure

Analytical Models

We provide some core abstractions for implementing analytical models in stream processing applications. The main interface for integrating analytical models is Analytic. Some analytical models need to adjust the domain input and the model output in some way, therefore we provide a special base class MappedAnalytic which has core abstractions for implementing that mapping via InputMapper and OutputMapper.

Since Spring XD 1.0.0.M6 we support the integration of analytical models, also called statistical models or mining models, that are defined via PMML. PMML is the abbreviation for Predictive Model Markup Language and is a standard XML representation that allows specifications of different mining models, their ensembles, and associated preprocessing.

Note

PMML is maintained by the Data Mining Group (DMG) and supported by several state-of-the-art statistics and data mining software tools such as InfoSphere Warehouse, R / Rattle, SAS Enterprise Miner, SPSS®, and Weka. The current version of the PMML specification is 4.2 at the time of this writing. Applications can produce and consume PMML models, thus allowing an analytical model created in one application to be implemented and used for scoring or prediction in another.

PMML is just one of many other technologies that one can integrate to implement analytics with, more will follow in upcoming releases.

Modeling and Evaluation

Analytical models are usually defined by a statistician aka data scientist or quant by using some statistical tool to analyze the data and build an appropriate model. In order to implement those models in a business application they are usually transformed and exported in some way (e.g. in the form of a PMML definition). This model is then loaded into the application which then evaluates it against a given input (event, tuple, example).

Modeling

Analytical models can be defined in various ways. For the sake of brevity we use R from the r-project to demonstrate how easy it is to export an analytical model to PMML and use it later in stream processing.

For our example we use the iris example dataset in R to generate a classifier for iris flower species by applying the Naive Bayes algorithm.

library(e1071) # Load library with the naive bayes algorithm support.

library(pmml) # Load library with PMML export support.

data(iris) # Load the IRIS example dataset

#Helper function to split the given dataset into a dataset used for training (trainset) and (testset) used for evaulation.
splitDataFrame <- function(dataframe, seed = NULL, n = trainSize) {

   if (!is.null(seed)){
      set.seed(seed)
   }

   index <- 1:nrow(dataframe)
   trainindex <- sample(index, n)
   trainset <- dataframe[trainindex, ]
   testset <- dataframe[-trainindex, ]

   list(trainset = trainset, testset = testset)
}

#We want to use 95% of the IRIS data as training data and 5% as test data for evaluation.
datasets <- splitDataFrame(iris, seed = 1337, n= round(0.95 * nrow(iris)))

#Create a naive Bayes classifier to predict iris flower species (iris[,5]) from [,1:4] = Sepal.Length Sepal.Width Petal.Length Petal.Width
model <- naiveBayes(datasets$trainset[,1:4], datasets$trainset[,5])

#The name of the model and it's externalId could be used to uniquely identify this version of the model.
modelName = "iris-flower-classifier"
externalId = 42

#Convert the given model into a PMML model definition
pmmlDefinition = pmml.naiveBayes(model,model.name=paste(modelName,externalId,sep = ";"), predictedField='Species')

#Print the PMML definition to stdout
cat(toString(pmmlDefinition))

The r script above should produce the following PMML document that contains the abstract definition of the naive bayes classifier that we derived from the training dataset of the IRIS dataset.

<PMML version="4.1" xmlns="http://www.dmg.org/PMML-4_1" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.dmg.org/PMML-4_1 http://www.dmg.org/v4-1/pmml-4-1.xsd">
<Header copyright="Copyright (c) 2014 tom" description="NaiveBayes Model">
 <Extension name="user" value="tom" extender="Rattle/PMML"/>
 <Application name="Rattle/PMML" version="1.4"/>
 <Timestamp>2014-04-02 13:22:15</Timestamp>
</Header>
<DataDictionary numberOfFields="6">
 <DataField name="Species" optype="categorical" dataType="string">
  <Value value="setosa"/>
  <Value value="versicolor"/>
  <Value value="virginica"/>
 </DataField>
 <DataField name="Sepal.Length" optype="continuous" dataType="double"/>
 <DataField name="Sepal.Width"  optype="continuous" dataType="double"/>
 <DataField name="Petal.Length" optype="continuous" dataType="double"/>
 <DataField name="Petal.Width"  optype="continuous" dataType="double"/>
 <DataField name="DiscretePlaceHolder" optype="categorical" dataType="string">
  <Value value="pseudoValue"/>
 </DataField>
</DataDictionary>
<NaiveBayesModel modelName="iris-flower-classifier;42"
                 functionName="classification" threshold="0.001">
 <MiningSchema>
  <MiningField name="Species"      usageType="predicted"/>
  <MiningField name="Sepal.Length" usageType="active"/>
  <MiningField name="Sepal.Width"  usageType="active"/>
  <MiningField name="Petal.Length" usageType="active"/>
  <MiningField name="Petal.Width"  usageType="active"/>
  <MiningField name="DiscretePlaceHolder" usageType="active"
               missingValueReplacement="pseudoValue"/>
 </MiningSchema>
 <Output>
  <OutputField name="Predicted_Species" feature="predictedValue"/>
  <OutputField name="Probability_setosa" optype="continuous"
               dataType="double" feature="probability" value="setosa"/>
  <OutputField name="Probability_versicolor" optype="continuous"
               dataType="double" feature="probability" value="versicolor"/>
  <OutputField name="Probability_virginica" optype="continuous"
               dataType="double" feature="probability" value="virginica"/>
 </Output>
 <BayesInputs>
  <Extension>
   <BayesInput fieldName="Sepal.Length">
    <TargetValueStats>
     <TargetValueStat value="setosa">
      <GaussianDistribution mean="5.006" variance="0.124248979591837"/>
     </TargetValueStat>
     <TargetValueStat value="versicolor">
      <GaussianDistribution mean="5.8953488372093" variance="0.283311184939092"/>
     </TargetValueStat>
     <TargetValueStat value="virginica">
      <GaussianDistribution mean="6.58163265306122" variance="0.410697278911565"/>
     </TargetValueStat>
    </TargetValueStats>
   </BayesInput>
  </Extension>
  <Extension>
   <BayesInput fieldName="Sepal.Width">
    <TargetValueStats>
     <TargetValueStat value="setosa">
...
...
      <GaussianDistribution mean="3.428" variance="0.143689795918367"/>
     </TargetValueStat>
     <TargetValueStat value="versicolor">
      <GaussianDistribution mean="2.76279069767442" variance="0.0966777408637874"/>
     </TargetValueStat>
     <TargetValueStat value="virginica">
      <GaussianDistribution mean="2.97142857142857" variance="0.105833333333333"/>
     </TargetValueStat>
    </TargetValueStats>
   </BayesInput>
  </Extension>
  <Extension>
   <BayesInput fieldName="Petal.Length">
    <TargetValueStats>
     <TargetValueStat value="setosa">
      <GaussianDistribution mean="1.462" variance="0.0301591836734694"/>
     </TargetValueStat>
     <TargetValueStat value="versicolor">
      <GaussianDistribution mean="4.21627906976744" variance="0.236633444075305"/>
     </TargetValueStat>
     <TargetValueStat value="virginica">
      <GaussianDistribution mean="5.55510204081633" variance="0.310442176870748"/>
     </TargetValueStat>
    </TargetValueStats>
   </BayesInput>
  </Extension>
  <Extension>
   <BayesInput fieldName="Petal.Width">
    <TargetValueStats>
     <TargetValueStat value="setosa">
      <GaussianDistribution mean="0.246" variance="0.0111061224489796"/>
     </TargetValueStat>
     <TargetValueStat value="versicolor">
      <GaussianDistribution mean="1.30697674418605" variance="0.042093023255814"/>
     </TargetValueStat>
     <TargetValueStat value="virginica">
      <GaussianDistribution mean="2.02448979591837" variance="0.0768877551020408"/>
     </TargetValueStat>
    </TargetValueStats>
   </BayesInput>
  </Extension>
  <BayesInput fieldName="DiscretePlaceHolder">
   <PairCounts value="pseudoValue">
    <TargetValueCounts>
     <TargetValueCount value="setosa" count="50"/>
     <TargetValueCount value="versicolor" count="43"/>
     <TargetValueCount value="virginica" count="49"/>
    </TargetValueCounts>
   </PairCounts>
  </BayesInput>
 </BayesInputs>
 <BayesOutput fieldName="Species">
  <TargetValueCounts>
   <TargetValueCount value="setosa" count="50"/>
   <TargetValueCount value="versicolor" count="43"/>
   <TargetValueCount value="virginica" count="49"/>
  </TargetValueCounts>
 </BayesOutput>
</NaiveBayesModel>
</PMML>

Evaluation

The above defined PMML model can be evaluated in a Spring XD stream definition by using the analytic-pmml module as a processor in your stream definition. The actual evaluation of the PMML is performed via the PmmlAnalytic which uses the jpmml-evaluator library.

Model Selection

The PMML standard allows multiple models to be defined within a single PMML document. The model to be used can be configured through the modelName option.

NOTE The PMML standard also supports other ways for selection models, e.g. based on a predicate. This is currently not supported.

In order to perform the evaluation in Spring XD you need to save the generated PMML document to some folder, typically the with the extension "pmml.xml". For this example we save the PMML document under the name iris-flower-classification-naive-bayes-1.pmml.xml.

In the following example we set up a stream definition with an http source that produces iris-flower-records that are piped to the analytic-pmml module which applies our iris flower classifier to predict the species of a given flower record. The result of that is a new record extended by a new attribute predictedSpecies which simply sent to a log sink.

The definition of the stream, which we call iris-flower-classification, looks as follows:

xd:>stream create --name iris-flower-classification
  --definition "http --outputType=application/x-xd-tuple | analytic-pmml
  --location=/models/iris-flower-classification-naive-bayes-1.pmml.xml
  --inputFieldMapping='sepalLength:Sepal.Length,
                       sepalWidth:Sepal.Width,
                       petalLength:Petal.Length,
                       petalWidth:Petal.Width'
  --outputFieldMapping='Predicted_Species:predictedSpecies' | log" --deploy
  • The location parameter can be used to specify the exact location of the pmml document. The value must be a valid spring resource location

  • The inputFieldMapping parameter defines a mapping of domain input fields to model input fields. It is just a list of fields or optional field:alias mappings to control which fields and how they are going to end up in the model-input. If no inputFieldMapping is defined then all domain input fields are used as model input.

  • The outputFieldMapping parameter defines a mapping of model output fields to domain output fields with semantics analog to the inputFieldMapping.

  • The optional modelName parameter of the analytic-pmml module can be used to refer to a particular named model within the PMML definition. If modelName is not defined the first model is selected by default.

NOTE Some analytical models like for instance association rules require a different typ of mapping. You can implement your own custom mapping strategies by implementing a custom InputMapper and OutputMapper and defining a new PmmlAnalytic or TuplePmmlAnalytic bean that uses your custom mappers.

After the stream has been successfully deployed to Spring XD we can eventually start to throw some data at it by issuing the following http request via the XD-Shell (or curl, or any other tool):

Note that our example record contains no information about which species the example belongs to - this will be added by our classifier.

xd:>http post --target http://localhost:9000 --contentType application/json --data "{ \"sepalLength\": 6.4, \"sepalWidth\": 3.2, \"petalLength\":4.5, \"petalWidth\":1.5 }"

After posting the above json document to the stream we should see the following output in the console:

   {
     "id":"1722ec00-baad-11e3-b988-005056c00008"
   , "timestamp":1396473833152
   , "sepalLength":"6.4"
   , "sepalWidth":"3.2"
   , "petalLength":"4.5"
   , "petalWidth":"1.5"
   , "predictedSpecies":"versicolor"
   }

NOTE the generated field predictedSpecies which now identifies our input as belonging to the iris species versicolor.

We verify that the generated PMML classifier produces the same result as R by executing the issuing the following commands in rproject:

datasets$testset[,1:4][1,]
# This is the first example record that we sent via the http post.
   Sepal.Length Sepal.Width Petal.Length Petal.Width
52          6.4         3.2          4.5         1.5

#Predict the class for the example record by using our naiveBayes model.
> predict(model, datasets$testset[,1:4][1,])
[1] versicolor

Counters and Gauges included with Spring XD

See the Counters and Gauges section for a list of available metrics included with Spring XD, along with configuration options and examples.

Tuples

Introduction

The Tuple class is a central data structure in Spring XD. It is an ordered list of values that can be retrieved by name or by index. Tuples are created by a TupleBuilder and are immutable. The values that are stored can be of any type and null values are allowed.

The underlying Message class that moves data from one processing step to the next can have an arbitrary data type as its payload. Instead of creating a custom Java class that encapsulates the properties of what is read or set in each processing step, the Tuple class can be used instead. Processing steps can be developed that read data from specific named values and write data to specific named values.

There are accessor methods that perform type conversion to the basic primitive types as well as BigDecimal and Date. This avoids you from having to cast the values to specific types. Instead you can rely on the Tuple’s type conversion infastructure to perform the conversion.

The Tuple’s types conversion is performed by Spring’s Type Conversion Infrastructure which supports commonly encountered type conversions and is extensible.

There are several overloads for getters that let you provide default values for primitive types should the field you are looking for not be found. Date format patterns and Locale aware NumberFormat conversion are also supported. A best effort has been made to preserve the functionality available in Spring Batch’s FieldSet class that has been extensively used for parsing String based data in files.

Creating a Tuple

The TupleBuilder class is how you create new Tuple instances. The most basic case is

Tuple tuple = TupleBuilder.tuple().of("foo", "bar");

This creates a Tuple with a single entry, a key of foo with a value of bar. You can also use a static import to shorten the syntax.

import static org.springframework.xd.tuple.TupleBuilder.tuple;

Tuple tuple = tuple().of("foo", "bar");

You can use the of method to create a Tuple with up to 4 key-value pairs.

Tuple tuple2 = tuple().of("up", 1, "down", 2);
Tuple tuple3 = tuple().of("up", 1, "down", 2, "charm", 3 );
Tuple tuple4 = tuple().of("up", 1, "down", 2, "charm", 3, "strange", 4);

To create a Tuple with more then 4 entries use the fluent API that strings together the put method and terminates with the build method

Tuple tuple6 = tuple().put("up", 1)
      	              .put("down", 2)
		      .put("charm", 3)
		      .put("strange", 4)
		      .put("bottom", 5)
		      .put("top", 6)
		      .build();

To customize the underlying type conversion system you can specify the DateFormat to use for converting String to Date as well as the NumberFormat to use based on a Locale. For more advanced customization of the type conversion system you can register an instance of a FormattingConversionService. Use the appropriate setter methods on TupleBuilder to make these customizations.

You can also create a Tuple from a list of String field names and a List of Object values.

Object[] tokens = new String[] { "TestString", "true", "C", "10", "-472", "354224", "543", "124.3", "424.3", "1,3245",
				null, "2007-10-12", "12-10-2007", "" };
String[] nameArray = new String[] { "String", "Boolean", "Char", "Byte", "Short", "Integer", "Long", "Float", "Double",
				"BigDecimal", "Null", "Date", "DatePattern", "BlankInput" };

List<String> names = Arrays.asList(nameArray);
List<Object> values = Arrays.asList(tokens);
tuple = tuple().ofNamesAndValues(names, values);

Getting Tuple values

There are getters for all the primitive types and also for BigDecimal and Date. The primitive types are

  • Boolean

  • Byte

  • Char

  • Double

  • Float

  • Int

  • Long

  • Short

  • String

Each getter has an overload for providing a default value. You can access the values either by field name or by index.

The overloaded methods for asking for a value to be converted into an integer are

  • int getInt(int index)

  • int getInt(String name)

  • int getInt(int index, int defaultValue)

  • int getInt(String name, int defaultValue)

There are similar methods for other primitive types. For Boolean there is a special case of providing the String value that represents a trueValue.

  • boolean getBoolean(int index, String trueValue)

  • boolean getBoolean(String name, String trueValue)

If the value that is stored for a given field or index is null and you ask for a primitive type, the standard Java defalt value for that type is returned.

The getString method will remove and leading and trailing whitespace. If you want to get the String and preserve whitespace use the methods getRawString

There is extra functionality for getting `Date`s. The are overloaded getters that take a String based date format

  • Date getDateWithPattern(int index, String pattern)

  • Date getDateWithPattern(int index, String pattern, Date defaultValue)

  • Date getDateWithPattern(String name, String pattern)

  • Date getDateWithPattern(String name, String pattern, Date defaultValue)

There are a few other more generic methods available. Their functionality should be obvious from their names

  • size()

  • getFieldCount()

  • getFieldNames()

  • getFieldTypes()

  • getTimestamp() - the time the tuple was created - milliseconds since epoch

  • getId() - the UUID of the tuple

  • Object getValue(int index)

  • Object getValue(String name)

  • T getValue(int index, Class<T> valueClass)

  • T getValue(String name, Class<T> valueClass)

  • List<Object> getValues()

  • List<String> getFieldNames()

  • boolean hasFieldName(String name)

Using SpEL expressions to filter a tuple

SpEL provides support to transform a source collection into another by selecting from its entries. We make use of this functionalty to select a elements of a the tuple into a new one.

Tuple tuple = tuple().put("red", "rot")
                     .put("brown", "braun")
		     .put("blue", "blau")
		     .put("yellow", "gelb")
		     .put("beige", "beige")
		     .build();

Tuple selectedTuple = tuple.select("?[key.startsWith('b')]");
assertThat(selectedTuple.size(), equalTo(3));

To select the first match use the ^ operator

selectedTuple = tuple.select("^[key.startsWith('b')]");

assertThat(selectedTuple.size(), equalTo(1));
assertThat(selectedTuple.getFieldNames().get(0), equalTo("brown"));
assertThat(selectedTuple.getString(0), equalTo("braun"));

Gradle Dependencies

If you wish to use Spring XD Tuples in you project add the following dependencies:

//Add this repo to your repositories if it does not already exist.
maven { url "http://repo.spring.io/libs-snapshot"}

//Add this dependency
compile 'org.springframework.xd:spring-xd-tuple:1.3.3.BUILD-SNAPSHOT'

Type Conversion

Introduction

Spring XD allows you to declaratively configure type conversion in stream definitions using the inputType and outputType module options. Note that general type conversion may also be accomplished easily within a transformer or a custom module. Currently, Spring XD natively supports the following type conversions commonly used in streams:

  • JSON to/from POJO

  • JSON to/from org.springframework.xd.tuple.Tuple

  • Object to/from byte[] : Either the raw bytes serialized for remote transport, bytes emitted by a module, or converted to bytes using Java serialization(requires the object to be Serializable)

  • String to/from byte[]

  • Object to plain text (invokes the object’s toString() method)

Where JSON represents either a byte array or String payload containing JSON. Currently, Objects may be converted from a JSON byte array or String. Converting to JSON always produces a String. Registration of custom type converters is covered in this section.

MIME types

inputType and outputType values are parsed as media types, e.g., application/json or text/plain;charset=UTF-8. MIME types are especially useful for indicating how to convert to String or byte[] content. Spring XD also uses MIME type format to represent Java types, using the general type application/x-java-object with a type parameter. For example, application/x-java-object;type=java.util.Map or application/x-java-object;type=com.bar.Foo can be used in an inputType convertion, --inputType='application/x-java-object;type=com.bar.Foo'. Note that the type is quoted with '' to avoid parsing errors produced by the ; character or other invalid characters. For convenience, you can use the class name by itself and Spring XD will translate a valid class name to the corresponding MIME type. In addition, Spring XD provides custom MIME types, notably, application/x-xd-tuple to specify a Tuple.

Stream Definition Examples

POJO to JSON

Type conversion will likely come up when implementing a custom module which produces or consumes a custom domain object. For example, you want to create a stream that integrates with a legacy system that includes custom domain types in its API. To process custom domain types directly minimally requires these types to be defined in Spring XD’s class path. This approach will be cumbersome to maintain when the domain model changes. The recommended approach is to convert such types to JSON at the source, or back to POJO at the sink. You can do this by declaring the required conversions in the stream definition:

customPojoSource --outputType=application/json |p1 | p2 | ... | customPojoSink --inputType=application/x-java-object;type=com.acme.MyDomainType

Note that the sink above does require the declared type to be in the module’s classpath to perform the JSON to POJO conversion. Generally, POJO to JSON does not require the Java class. Once the payload is converted to JSON, Spring XD provided transformers and filters (p1, p2, etc.) can evaluate the payload contents using JsonPath functions in SpEL expressions. Alternately, you can convert the JSON to a Tuple, as shown in the following example.

JSON to Tuple

Sometimes it is convenient to convert JSON content to a Tuple in order to evaluate and access individual field values.

xd:> stream create tuple --definition "http | filter --inputType=application/x-xd-tuple --expression=payload.hasFieldName('hello') | transform --expression=payload.getString('hello').toUpperCase() |  log" --deploy
Created and deployed new stream 'tuple'

Note inputType=application/x-xd-tuple on the filter module will cause the payload to be converted to a Tuple at the filter’s input channel. Thus, subsequent expressions are evaluated on a Tuple object. Here we invoke the Tuple methods hasFieldName('hello') on the filter and getString('hello') on the transformer. The output of the http source is expected to be JSON in this case. We set the Content-Type header to tell Spring XD that the payload is JSON.

xd:>http post --data {"hello":"world","foo":"bar"} --contentType application/json --target http://localhost:9000
> POST (application/json;charset=UTF-8) http://localhost:9000 {"hello":"world","foo":"bar"}
> 200 OK

In the Spring XD console log, you should see something like:

13:19:45,054  INFO pool-42-thread-4 sink.tuple - WORLD

Java Serialization

The following serializes a java.io.Serializable object to a file. Presumably the foo module outputs a Serializable type. If not, this will result in an exception. If remote transport is configured, the output of foo will be marshalled using Spring XD’s internal serialization. The object will be unmarshalled in the file module and then converted to a byte array using Java serialization.

foo  | --inputType=application/x-java-serialized-object file

MIME types and Java types

Internally, Spring XD implements type conversion using Spring Integration’s data type channels. The data type channel converts payloads to the configured data type using Spring’s MessageConverter.

Note
The use of MessageCoverter for data type channels was introduced in Spring Integration 4 to pass the Message to the converter method to allow it to access the Message’s content-type header. This provides greater flexibility. For example, it is now possible to support multiple strategies for converting a String or byte array to a POJO, based on the content-type header.

When Spring XD deploys a module with a declared type conversion, it modifies the module’s input and/or output channel definition to set the required Java type and registers MessageConverters associated with the target MIME type and Java type to the channel. The type conversions Spring XD provides out of the box are summarized in the following table:

Source Payload Target Payload content-type header outputType/inputType Comments

POJO

JSON String

ignored

application/json

Tuple

JSON String

ignored

application/json

JSON is tailored for Tuple

POJO

String (toString())

ignored

text/plain, java.lang.String

POJO

byte[] (java.io serialized)

ignored

application/x-java-serialized-object

JSON byte[] or String

POJO

application/json (or none)

application/x-java-object

byte[] or String

Serializable

application/x-java-serialized-object

application/x-java-object

JSON byte[] or String

Tuple

application/json (or none)

application/x-xd-tuple

byte[]

String

any

text/plain, java.lang.String

will apply any Charset specified in the content-type header

String

byte[]

any

application/octet-stream

will apply any Charset specified in the content-type header

Caveats

Note that inputType and outputType parameters only apply to payloads that require type conversion. For example, if a module produces an XML string with outputType=application/json, the payload will not be converted from XML to JSON. This is because the payload at the module’s output channel is already a String so no conversion will be applied at runtime.

Developing Modules and Extensions

Creating a Source Module

Introduction

As outlined in the modules document, Spring XD currently supports four types of modules: source, sink, and processor for stream processing and job for batch processing. This document walks through the creation of a custom source module.

The first module in a stream is always a source. Source modules are built with Spring Integration and are responsible for producing messages originating from an external data source on its output channel. These message can then be processed by the downstream modules in a stream. A source module is often fed data by a Spring Integration inbound channel adapter, configured with a poller.

Spring Integration provides a number of adapters out of the box to integrate with various transports and data stores, such as JMS, File, HTTP, Web Services, Mail, and more. Typically, it is straightforward to create a source module using an existing inbound channel adapter.

Here we walk through an example demonstrating how to create and register a source module using the Spring Integration Feed Inbound Channel Adapter. The complete code for this example is in the rss-feed-source sample project.

Create the module Application Context file

Configure the inbound channel adapter using an xml bean definition file in the config resource directory:

<beans...>

	<int-feed:inbound-channel-adapter id="xdFeed" channel="output" url="${url}" auto-startup="false" >
		<int:poller fixed-rate="${fixedRate}" max-messages-per-poll="${maxMessagesPerPoll}" />
	</int-feed:inbound-channel-adapter>

	<int:channel id="output"/>
</beans>

The adapter is configured to poll an RSS feed at a fixed rate (e.g., every 5 seconds). Note that auto-startup is set to false. This is a requirement for Spring XD modules. When a stream is deployed, the Spring XD runtime will create and start stream modules in reverse order to ensure that all modules are initialized before the source starts emiting messages. When an RSS Entry is retreived, it will create a message with a com.rometools.rome.feed.synd.SyndEntry payload type and send it to a message channel called output. The name output is a Spring XD convention indicating the module’s output channel. Any messages on the output channel will be consumed by the downstream processor or sink in a stream used by this module.

The module is configurable so that it may pull data from any feed URL, such as http://feeds.bbci.co.uk/news/rss.xml. Spring XD will automatically register a PropertyPlaceholderConfigurer in the module’s application context. These properties correspond to module options defined for this module (discussed below). Users supply option values when creating a stream using the DSL.

Users must provide a url option value when creating a stream that uses this source. The polling rate and maximum number of entries retrieved for each poll are also configurable and for these properties we should provide reasonable default values. The module’s properties file in the config resource directory contains Module Options Metadata including a description, type, and optional default value for each property. The metadata supports features like auto-completion in the Spring XD shell and option validation:

options.url.description = the URL of the RSS feed
options.url.type = java.lang.String

options.fixedRate.description = the fixed rate polling interval specified in milliseconds
options.fixedRate.default = 5000
options.fixedRate.type = int

options.maxMessagesPerPoll.description = the maximum number of messages per poll
options.maxMessagesPerPoll.default = 100
options.maxMessagesPerPoll.type = int

Alternately, you can write a POJO to define the metadata. Using a Java class provides better validation along with additional features and requires that the class be packaged as part of the module.

Create a Module Project

This section covers the setup of a standalone project containing the module configuration and some code for testing the module. This example uses Maven but Spring XD supports Gradle as well.

Take a look at the pom file for this example. You will see it declares spring-xd-module-parent as its parent and declares a dependency on spring-integration-feed which provides the inbound channel adapter. The parent pom provides everything else you need. We also need to configure repositories to access the parent pom and any other dependencies. The required xml file containing the bean definitions and properties file are located in src\main\resources\config. In this case, we have elected to use a custom transformer to convert the output of the feed inbound adapter to a JSON string.

<beans ...>
	<int-feed:inbound-channel-adapter id="xdFeed" channel="to.json" url="${url}" auto-startup="false">
		<int:poller fixed-rate="${fixedRate}" max-messages-per-poll="${maxMessagesPerPoll}" />
	</int-feed:inbound-channel-adapter>

	<int:transformer input-channel="to.json" output-channel="output">
		<bean class="com.acme.SyndEntryJsonTransformer"/>
	</int:transformer>

	<int:channel id="output"/>
</beans>

The project README contains a detailed explanation of why this transformer is needed, but such things are easily accomplished with Spring Integration.

Create a Spring Integration test

The first level of testing should ensure that the module’s Application Context is loaded and that the message flow works as expected independent of Spring XD. In this case, we need to wrap the module application context in a test context that provides a property placeholder (the Spring XD runtime does this for you). In addition, it is convenient to override the module’s output channel with a queue channel so that the test will block until a message is received from the feed.

Add the following configuration in the appropriate location under src/test/resources/:

<beans ...>

	<context:property-placeholder properties-ref="props"/>
	<util:properties id="props">
		<prop key="url">http://feeds.bbci.co.uk/news/rss.xml</prop>
		<prop key="fixedRate">5000</prop>
		<prop key="maxMessagesPerPoll">100</prop>
	</util:properties>

	<import resource="classpath:config/spring-module.xml"/>

	<!-- Override direct channel with a queue channel so the test will block until a message is received -->
	<int:channel id="output">
		<int:queue/>
	</int:channel>
</beans>

Next, create and run the test:

package com.acme;

import ...

@RunWith(SpringJUnit4ClassRunner.class)
@ContextConfiguration
public class FeedConfigurationTest {
	@Autowired
	PollableChannel output;

	@Autowired
	ConfigurableApplicationContext applicationContext;

	@Test
	public void test() {
		applicationContext.start();
		Message message = output.receive(10000);
		assertNotNull(message);
		assertTrue(message.getPayload() instanceof String);
	}
}

The test will load an Application Context using our feed and test context files. It will fail if a item is not received on the output channel within 10 seconds.e

Create an in-container test

Now that you have verified that the module is basically correct, you can write a test to use it in a stream deployed to an embedded Spring XD container.

Note

See test a module for some important tips abouts regarding in-container testing.

The spring-xd-module-parent pom provides the necessary dependencies to write such a test:

package com.acme;

import ...

public class FeedSourceModuleIntegrationTest {
	private static SingleNodeApplication application;

	private static int RECEIVE_TIMEOUT = 6000;

	/**
	 * Start the single node container, binding random unused ports, etc. to not conflict with any other instances
	 * running on this host. Configure the ModuleRegistry to include the project module.
	 */
	@BeforeClass
	public static void setUp() {
		RandomConfigurationSupport randomConfigSupport = new RandomConfigurationSupport();
		application = new SingleNodeApplication().run();
		SingleNodeIntegrationTestSupport singleNodeIntegrationTestSupport = new SingleNodeIntegrationTestSupport(application);
		singleNodeIntegrationTestSupport.addModuleRegistry(new SingletonModuleRegistry(ModuleType.source, "feed"));
	}

	@Test
	public void test() {
		String url = "http://feeds.bbci.co.uk/news/rss.xml";
		SingleNodeProcessingChainConsumer chain = chainConsumer(application, "feedStream", String.format("feed --url='%s'", url));

		Object payload = chain.receivePayload(RECEIVE_TIMEOUT);
		assertTrue(payload instanceof String);

		chain.destroy();
	}
}

The above test configures an and starts embedded Spring XD runtime (SingleNodeApplication) to deploy a stream that uses the module under test.

The SingleNodeProcessingChainConsumer can test a stream that does not include a sink. The chain itself provides an in-memory sink to access the stream’s output directly. In this case, we use the chain to test the source in isolation. The above test is equivalent to deploying following stream definition:

feed --url='http://feeds.bbci.co.uk/news/rss.xml' > queue:aNamedChannel

and the chain consumes messages on the named queue channel. At the end of each test method, the chain should be destroyed to destroy these internal resources and restore the initial state of the Spring XD container.

Note

The spring-xd-module-parent Maven pom includes a tasks to install a local message bus implementation under lib in the project root to enable a local transport provider for the embedded Spring XD container. It is necessary to run maven process-resources or a downstream goal (e.g., compile, test, package) once in order for this test to work correctly.

Install the Module

We have implemented and tested the module using Spring Integration directly and also by deploying the module to an embedded Spring XD container. Time to install the module to Spring XD!

The next step is to package the module as an uber-jar using maven:

$mvn package

This will build an uber-jar in target/rss-feed-source-1.0.0.BUILD-SNAPSHOT.jar. If you inspect the contents of this jar, you will see it includes the module configuration files, custom transformer class, and dependent jars. Fire up the Spring XD runtime if it is not already running and, using the Spring XD Shell, install the module as a source named feed using the module upload command:

xd:>module upload --file [path-to]/rss-source-feed/target/rss-source-feed-1.0.0.BUILD-SNAPSHOT.jar --name feed --type source

Also See registering a module for more details.

Test the source module

Once Spring XD is running, create a stream to test it the module. This stream will write SyndEntry objects rendered as JSON to the Spring XD log:

xd:> stream create --name feedtest --definition "feed --url='http://feeds.bbci.co.uk/news/rss.xml' | log" --deploy

You should start seeing messages like the following in the container log:

16:46:41,309 1.1.0.SNAP  INFO xdbus.feedTest.0-1 sink.feedTest - {"uri":"http://www.bbc.co.uk/sport/0/football/30700069","link":"http://www.bbc.co.uk/sport/0/football/30700069","comments":null,"updatedDate":null,"title":"Gerrard to seal move to LA Galaxy","description":{"type":"text/html","value":"Liverpool captain Steven Gerrard is on the brink of finalising an 18-month deal to join MLS side Los Angeles Galaxy.","mode":null,"interface":"com.rometools.rome.feed.synd.SyndContent"},"links":[],"contents":[],"modules":[{"uri":"http://purl.org/dc/elements/1.1/","title":null,"creator":null,"subject":null,"description":null,"publisher":null,"contributors":[],"date":1420580673000,"type":null,"format":null,"identifier":null,"source":null,"language":null,"relation":null,"coverage":null,"rights":null,"sources":[],"types":[],"formats":[],"identifiers":[],"interface":"com.rometools.rome.feed.module.DCModule","creators":[],"titles":[],"descriptions":[],"publishers":[],"contributor":null,"dates":[1420580673000],"languages":[],"relations":[],"coverages":[],"rightsList":[],"subjects":[]}],"enclosures":[],"authors":[],"contributors":[],"source":null,"wireEntry":null,"categories":[],"interface":"com.rometools.rome.feed.synd.SyndEntry","titleEx":{"type":null,"value":"Gerrard to seal move to LA Galaxy","mode":null,"interface":"com.rometools.rome.feed.synd.SyndContent"},"publishedDate":1420580673000,"author":""}

Creating a Data Stream Processor

Introduction

This section covers how to create a processor module that uses stream processing libraries and runtimes.module. Spring XD 1.2 provides integration with Project Reactor Stream, RxJava Observables, and Spark Streaming. Creating a data stream processor in XD allows you to use a functional programming model to filter, transform and aggregate data in a very concise and performant way. This section walks through implementing a custom processor module using each of these libraries.

Reactor Streams

Project Reactor provides a Stream API that is based on the Reactive Streams specification. The specification was jointly developed by twenty people from a dozen companies (Pivotal included) and has the goal of creating a standard for asynchronous stream processing with non-blocking back pressure on the JVM.

Tip
Explore Reactor with the reference guide.

To implement a Stream based processor module you need to implement the interface org.springframework.xd.reactor.Processor

public interface Processor<I, O> {

    /**
     * Process a stream of messages and return an output stream.  The input
     * and output stream will be mapped onto receive/send operations on the message bus.
     *
     * @param inputStream Input Stream the receives messages from the message bus
     * @return Output Publisher (Stream, Promise, or any valid Reactive Stream Publisher) of messages sent to the message bus
     */
    Publisher<O> process(Stream<I> inputStream);

}

Messages that are delivered on the Message Bus are accessed from the input Stream, which can be directly composed. The return value is the output Stream that is the result of applying various operations to the input stream. The content of the output Stream is sent to the message bus for consumption by other processors or sinks.

Examples of operations you can perform on the Stream are map, flatMap, buffer, window, and reduce. The parameterized data type can be a org.springframework.messaging.Message, org.springframework.xd.tuple.Tuple, java.lang.Map or any other POJO. The following example uses the Tuple object to compute the average value of a measurement from a sample size of 5.

import org.springframework.xd.reactor.Processor;
import org.springframework.xd.tuple.Tuple;
import reactor.rx.Stream;

import static com.acme.Math.avg;
import static org.springframework.xd.tuple.TupleBuilder.tuple;

public class MovingAverage implements Processor<Tuple, Tuple> {
    @Override
    public Publisher<Tuple> process(Stream<Tuple> inputStream) {
        return inputStream.map(tuple -> tuple.getDouble("measurement"))
                .buffer(5)
                .map(data -> tuple().of("average", avg(data)));
    }
}

You can now create unit tests for the Processor module just as you would for any other Java class. The module application context file can be in XML or in Java using a @Configuration class. The XML version is shown below.

<beans xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xmlns:int="http://www.springframework.org/schema/integration"
       xmlns="http://www.springframework.org/schema/beans"
       xsi:schemaLocation="http://www.springframework.org/schema/integration
         http://www.springframework.org/schema/integration/spring-integration.xsd
         http://www.springframework.org/schema/beans
	 http://www.springframework.org/schema/beans/spring-beans.xsd">


    <bean id="messageProcessor" class="com.acme.MovingAverage"/>


    <int:channel id="input"/>

    <bean name="messageHandler" class="org.springframework.xd.reactor.BroadcasterMessageHandler">
        <constructor-arg ref="messageProcessor"/>
    </bean>


    <int:service-activator input-channel="input" ref="messageHandler"
                           output-channel="output"/>

    <int:channel id="output"/>

</beans>

Examples of unit and integration testing a module are available in the reactor sample project. The sample project also shows how you can package your module into a single jar and upload it to the admin server.

RxJava Streams

RxJava provides the Observable API that is based on the Reactive Extensions .NET library.

To implement a Observable based XD processor module you need to implement the interface org.springframework.xd.rxjava.Processor

public interface Processor<I,O> {

    /**
     * Process a stream of messages and return an output stream.  The input
     * and output stream will be mapped onto receive/send operations on the message bus.
     *
     * @param inputStream Input stream the receives messages from the message bus
     * @return Output stream of messages sent to the message bus
     */
    Observable<O> process(Observable<I> inputStream);
}

Messages that are delivered on the Message Bus are accessed from the Observable input stream. The return value is the Observable output stream that contains the results of applying various operation to the input stream. The content of the output stream is sent to the message bus for consumption by other processors or sinks.

Examples of operations you can perform on the Stream are map, flatMap, buffer, window, and reduce. The parameterized data type can be a org.springframework.messaging.Message, org.springframework.xd.tuple.Tuple, java.lang.Map or any other POJO.

When used in combination with Data Partitioning on the Message Bus, this allows you to create an streaming application where Stream state is calculated based on those partitions where necessary.

Spring XD stream deployment with RxJava processing modules and data partitioning

In this deployment the data that is sent to the RxJava processing modules from the HTTP sources is partitioned such that the red data always goes to the red stream processing module and so on for the other colors. The next hop of processing, where writing to HDFS occurs, does not require data partitioning, so the message load can be shared across the HDFS sink instances.

There can be as many layers of RxJava Stream processing as you require, allowing you to collocate specific functional operations as you see fit within a single JVM or to distribute across multiple JVMs.

Spring XD stream deployment with multiple layers of RxJava processing modules

The following example uses the Tuple object to compute the average value of a measurement from a sample size of 5.

import org.springframework.xd.rxjava.Processor;
import org.springframework.xd.tuple.Tuple;
import rx.Observable;

import static com.acme.Math.avg;
import static org.springframework.xd.tuple.TupleBuilder.tuple;

public class MovingAverage implements Processor<Tuple, Tuple> {

    @Override
    public Observable<Tuple> process(Observable<Tuple> inputStream) {
        return inputStream.map(tuple -> tuple.getDouble("measurement"))
                .buffer(5)
                .map(data -> tuple().of("average", avg(data)));
    }
}

You can now create unit tests for the Processor module as you would for any other Java class. The module application context file can be in XML or in Java using a @Configuration class. The XML version is shown below.

<beans xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xmlns:int="http://www.springframework.org/schema/integration"
       xmlns="http://www.springframework.org/schema/beans"
       xsi:schemaLocation="http://www.springframework.org/schema/integration
         http://www.springframework.org/schema/integration/spring-integration.xsd
         http://www.springframework.org/schema/beans
	 http://www.springframework.org/schema/beans/spring-beans.xsd">


    <bean id="messageProcessor" class="com.acme.MovingAverage"/>


    <!-- Using a SubjectMessageHandler to share Observerable state across threads -->

    <int:channel id="input"/>

    <bean name="messageHandler" class="org.springframework.xd.rxjava.SubjectMessageHandler">
        <constructor-arg ref="messageProcessor"/>
    </bean>


    <int:service-activator input-channel="input" ref="messageHandler"
                           output-channel="output"/>

    <int:channel id="output"/>

</beans>

Examples of unit and integration testing a module are available in the RxJava Moving Average sample project. The sample project also shows how you can package your module into a single jar and upload it to the admin server.

Scheduling

There are two MessageHandler implementations that you can choose from, SubjectMessageHandler and MultipleSubjectMessageHandler.

SubjectMessageHandler uses a single SerializedSubject to process messages that were received from the Message Bus. This subject, downcast to Observable, is what is passed into the process method. Using SubjectMessageHandler has the advantage that the state of the Observabale input stream can be shared across all the Message Bus dispatcher threads that are invoking onNext. It has the disadvantage that the processing and consumption of the Observable output stream (that sends messages to the Message Bus) will execute serially on one of the dispatcher threads. Note you can modify what thread the Observable output stream will use by calling observeOn before returning the output stream from your processor.

MultipleSubjectMessageHandler uses multiple Subjects to perform processing. A Spring Expression Language (SpEL) expression is used to map the incoming message to a specific Subject to use for processing. Using MultipleSubjectMessageHandler has the advantage that it can use all Message Bus dispatcher threads. It has the disadvantage in that each Observable input stream has its own state, which may not be desirable for certain types of aggregate calculations that should see all of the data. A common partition expression to use is T(java.lang.Thread).currentThread().getId() so that a Subject will be created per thread.

    <bean name="messageHandler" class="org.springframework.xd.rxjava.MultipleSubjectMessageHandler">
        <constructor-arg ref="messageProcessor"/>
        <constructor-arg value="T(java.lang.Thread).currentThread().getId()"/>
    </bean>

The satisfies the contract to have single threaded access to a Subject. Another interesting partition expression to use in the case of the Kafka Message Bus is header['kafka_partition_id']. This will create a Subject per Kafka partition that represents an ordered sequence of events. The XD Kafka Message Bus statically maps partitions to dispatcher threads to there is only single threaded access toa Subject.

Spark streaming

Spring XD integrates with Spark streaming so that the streaming data computation logic can be run on a Spark cluster. Spring XD runs the Spark Driver as an XD module (processor or sink) in the XD container while the Spark streaming reliable receiver and the data computation is done at the Spark Cluster.

This provides advantage over connecting to various streaming sources while running the computation logic on Spark cluster. Running the spark driver on the XD container also provides automatic failover capabilities in case of driver failure.

With Spark Streaming, events are processed at the micro batch level via DStreams, which represent a continuous flow of partitioned RDDs. Setting up a Spark Streaming module within XD can be beneficial when adding streaming data computation logic for a tapped XD stream. While the primary stream processes events one at a time (through the regular XD modules), the tapped stream will become a source for the Spark Streaming module.

Lets discuss a real world scenario of data collection and doing some analytics on it.

stream create mainstream --definition "mqtt | filter1: <some filtering> | hdfs"
stream create sparkstream1 --definition "tap:stream:mainstream.filter1 > <some XD sink>"
stream create sparkstream2 --definition "tap:stream:mainstream.filter1 > spark-streaming-processor-module2 | spark-streaming-sinkmodule1"
stream create sparkstream3 --definition "tap:stream:mainstream.filter1 > spark-streaming-sinkmodule2"

In the above set of streams, consider a primary stream that collects data one at a time from various sensors and stores that raw data into HDFS, after only same basic filtering. At the same time, there are a few other streams that perform analytics on the data being collected at micro-batch level. Here, the tapped stream’s source can be reliable or durable based on the messagebus implementation, and this data is processed (at the micro batch level) by the Spark Streaming module. This allows the developer to choose the stream data processing based on the use case.

One can also add a tap at the output of the Spark streaming processor module. For instance, adding a tap at the output of sparkstream2's Spark stream processor would be:

stream create tapSparkStream --definition "tap:stream:sparkstream2.spark-streaming-processor-module2 > <some XD sink>"

Writing a spark streaming module

Spring XD provides Java and Scala based interfaces which expose a process method that the spark streaming developer would implement. This method processes the input DStream received by the spark streaming receiver. In case of XD processor module this method would return an output DStream. In case of XD sink module, it would write the computed data into file system, HDFS etc., (for example saveAsTextFiles(), saveAsHadoopFiles() using Spark APIs).

For Java based implementation, the interface org.springframework.xd.spark.streaming.java.Processor is defined

       public interface Processor<I extends JavaDStreamLike, O extends JavaDStreamLike> extends SparkStreamingSupport {

	/**
	 * Processes the input DStream and optionally returns an output DStream.
	 *
	 * @param input the input DStream
	 * @return output DStream (optional, may be null)
	 */
	O process(I input);
}

It is recommended to write the implementation in Java 8.

For Scala based implementation, the trait org.springframework.xd.spark.streaming.scala.Processor is defined

trait Processor[I, O] extends SparkStreamingSupport {

  /**
   * Processes the input DStream and optionally returns an output DStream.
   *
   * @param input the input DStream from the receiver
   * @return output DStream (optional, may be null)
   */
  def process(input: ReceiverInputDStream[I]): DStream[O]

When creating an XD processor/sink module, developer would implement this interface and make the module archive (along with its dependencies) available in the modules registry.

To set the Spark configuration properties when developing spark streaming module, the developer can use org.springframework.xd.spark.streaming.SparkConfig annotation on the method that returns type java.util.Properties.

To add default spark streaming command line options for the spark streaming module and to let XD admin know this is spark streaming module, following entry should be added in module registry module config properties (for example: modules/processor/spark-wordcount/config/spark-wordcount.properties):

options_class=org.springframework.xd.spark.streaming.DefaultSparkStreamingModuleOptionsMetadata

Developer can extend this to provide more custom command line options. By default, the following module options are supported for the spark streaming module:

  • batchInterval (the time interval in millis for batching the stream events)

  • storageLevel (the streaming data persistence storage level)

Note

If you are using Java7 to run Spring XD, then make sure to set the JAVA_OPTS to increase -XX:MaxPermSize to avoid PermGen issue on the XD container where the spark driver would be running.

How this works

When a Spark streaming processor (a processor or a sink) that implements Processor interface above is deployed, the SparkDriver sets up the streaming context and runs as an XD module inside the XD container.

This sets up Spark streaming receiver (in case of processor and sink) in spark cluster that connects to XD upstream module’s output channel in the message bus. Also note that this receiver is a reliable Spark streaming receiver (if you use rabbit or kafka as message-bus) out of the box. This is implemented using manual acknowledgement and explicit offset management on Rabbit and Kafka respectively. The MessageBusReceiver makes the incoming messages available for the computation in spark cluster as DStreams. If the streaming module is of XD processor type then the computed messages are pushed to the downstream module by MessageBusSender. The MessageBusSender binds to the downstream module’s input channel which subsequently connects to any of the XD processor or sink modules.

It is important to note that the MessageBusReceiver, streaming processor computation and the MessageBusSender run on Spark cluster.

Spring XD spark streaming modules and data partitioning

Failover and recovery

Spark streaming integration supports automatic failover capability on spark driver failure. In case of driver module failure, the module will get automatically re-deployed on another available XD container. Also, the underlying Spark streaming receiver is a reliable receiver when using RabbitMQ or Kafka as the messagebus. This will make sure all the messages are always acknowledged and processed reliably in Spark.

Module Type Conversion

Spark streaming modules can benefit from the out of the box module type conversion support from Spring XD. A spark streaming processor module can specify inputType and outputType while a spark streaming sink module can specify inputType to denote the contentType of the incoming/outgoing messages before they get ingested into/written out of spark streaming module.

stream create mainstream --definition "mqtt | filter1: <some filtering> | hdfs"
stream create sparkstream1 --definition "tap:mainstream:filter1 > spark-streaming-processor-module1 --inputType=application/json --outputType=application/x-xd-tuple | <some XD sink>"
stream create sparkstream2 --definition "tap:mainstream:filter1 > spark-streaming-processor-module2 | spark-streaming-sinkmodule1"
stream create sparkstream3 --definition "tap:mainstream:filter1 > spark-streaming-sinkmodule2 --inputType=text/plain"

For info on module type conversion, please refer here

XD processor module examples

Java based implementation

import java.util.Arrays;
import java.util.Properties;

import org.apache.spark.api.java.function.FlatMapFunction;
import org.apache.spark.api.java.function.Function2;
import org.apache.spark.api.java.function.PairFunction;
import org.apache.spark.streaming.api.java.JavaDStream;
import org.apache.spark.streaming.api.java.JavaPairDStream;

import org.springframework.xd.spark.streaming.SparkConfig;
import org.springframework.xd.spark.streaming.java.Processor;

import scala.Tuple2;

@SuppressWarnings({ "serial" })
public class WordCount implements Processor<JavaDStream<String>, JavaPairDStream<String, Integer>> {

	@Override
	public JavaPairDStream<String, Integer> process(JavaDStream<String> input) {
		JavaDStream<String> words = input.flatMap(new FlatMapFunction<String, String>() {

			@Override
			public Iterable<String> call(String x) {
				return Arrays.asList(x.split(" "));
			}
		});
		JavaPairDStream<String, Integer> wordCounts = words.mapToPair(new PairFunction<String, String, Integer>() {

			@Override
			public Tuple2<String, Integer> call(String s) {
				return new Tuple2<String, Integer>(s, 1);
			}
		}).reduceByKey(new Function2<Integer, Integer, Integer>() {

			@Override
			public Integer call(Integer i1, Integer i2) {
				return i1 + i2;
			}
		});
		return wordCounts;
	}

	@SparkConfig
	public Properties getSparkConfigProperties() {
		Properties props = new Properties();
		props.setProperty(SPARK_MASTER_URL_PROP, "local[4]");
		return props;
	}
}

Scala based implementation

import java.util.Properties

import org.apache.spark.streaming.StreamingContext._
import org.apache.spark.streaming.dstream.{DStream, ReceiverInputDStream}
import org.springframework.xd.spark.streaming.SparkConfig
import org.springframework.xd.spark.streaming.scala.Processor

class WordCount extends Processor[String, (String, Int)] {

  def process(input: ReceiverInputDStream[String]): DStream[(String, Int)] = {
      val words = input.flatMap(_.split(" "))
      val pairs = words.map(word => (word, 1))
      val wordCounts = pairs.reduceByKey(_ + _)
      wordCounts
  }

  @SparkConfig
  def properties : Properties = {
    val props = new Properties()
    props.setProperty("spark.master", "local[4]")
    props
  }

}

XD sink module example

Java based implementation

import java.io.BufferedWriter;
import java.io.File;
import java.io.FileWriter;
import java.io.IOException;
import java.util.Iterator;
import java.util.Properties;

import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.function.Function;
import org.apache.spark.api.java.function.VoidFunction;
import org.apache.spark.streaming.api.java.JavaDStream;

import org.springframework.xd.spark.streaming.SparkConfig;
import org.springframework.xd.spark.streaming.java.Processor;

@SuppressWarnings({ "serial" })
public class FileLogger implements Processor<JavaDStream<String>, JavaDStream<String>> {

	private File file;

	public void setPath(String filePath) {
		file = new File(filePath);
		if (!file.exists()) {
			try {
				file.createNewFile();
			}
			catch (IOException ioe) {
				throw new RuntimeException(ioe);
			}
		}
	}

	@SparkConfig
	public Properties getSparkConfigProperties() {
		Properties props = new Properties();
		props.setProperty("spark.master", "local[4]");
		return props;
	}

	@Override
	public JavaDStream<String> process(JavaDStream<String> input) {
		input.foreachRDD(new Function<JavaRDD<String>, Void>() {

			@Override
			public Void call(JavaRDD<String> rdd) {
				rdd.foreachPartition(new VoidFunction<Iterator<String>>() {

					@Override
					public void call(Iterator<String> items) throws Exception {
						FileWriter fw;
						BufferedWriter bw = null;
						try {
							fw = new FileWriter(file.getAbsoluteFile());
							bw = new BufferedWriter(fw);
							while (items.hasNext()) {
								bw.append(items.next() + System.lineSeparator());
							}
						}
						catch (IOException ioe) {
							throw new RuntimeException(ioe);
						}
						finally {
							if (bw != null) {
								bw.close();
							}
						}
					}
				});
				return null;
			}
		});
		return null;
	}

}

Scala based implementation

import java.io.{BufferedWriter, File, FileWriter, IOException}
import java.util.Properties

import org.apache.spark.streaming.dstream.{DStream, ReceiverInputDStream}
import org.springframework.xd.spark.streaming.SparkConfig
import org.springframework.xd.spark.streaming.scala.Processor

class FileLogger extends Processor[String, String] {

  var file: File = null

  def setPath(filePath: String) {
    file = new File(filePath)
    if (!file.exists) {
      try {
        file.createNewFile
      }
      catch {
        case ioe: IOException => {
          throw new RuntimeException(ioe)
        }
      }
    }
  }

  @SparkConfig def getSparkConfigProperties: Properties = {
    val props: Properties = new Properties
    props.setProperty("spark.master", "local[4]")
    return props
  }

  def process(input: ReceiverInputDStream[String]): DStream[String] = {
      input.foreachRDD(rdd => {
        rdd.foreachPartition(partition => {
          var fw: FileWriter = null
          var bw: BufferedWriter = null
          try {
            fw = new FileWriter(file.getAbsoluteFile)
            bw = new BufferedWriter(fw)
            while (partition.hasNext) {
              bw.append(partition.next.toString + System.lineSeparator)
            }
          }
          catch {
            case ioe: IOException => {
              throw new RuntimeException(ioe)
            }
          }
          finally {
            if (bw != null) {
              bw.close
            }
          }
        })
      })
    null
  }
}

Checkout some examples, module configurations and tests

Creating a Processor Module

Introduction

As outlined in the modules document, Spring XD currently supports four types of modules: source, sink, and processor for stream processing and job for batch processing. This document walks through implementing a custom processor module.

One or more processors can be included in a stream definition to modify the data as it passes on its way from the source to the sink. The architecture section covers the basics of stream processing. Processor modules provided out of the box are covered in the processors section.

Here we’ll look at how to create a simple processor module from scratch. This module will extract the text field from input messages from from a twittersearch source. The steps are essentially the same regardless of the module’s functionality. Note that Spring XD can perform this type of transformation without requiring a custom module. Rather than using the built-in functionality, we will implement a custom processor and wire it up with Spring Integration. The complete code for this example is here.

Write the Transformer Code

The tweet messages from twittersearch contain quite a lot of data (id, author, time, hash tags, and so on). The transformer we’ll write extracts the text of each tweet and outputs this as a string. The output messages from the twittersearch source are also strings, rendering the tweet data as JSON. We first load this into a map using Jackson library code, then extract the text field from the map.

package my.custom.transformer;

import java.io.IOException;
import java.util.Map;

import org.codehaus.jackson.map.ObjectMapper;
import org.codehaus.jackson.type.TypeReference;
import org.springframework.integration.transformer.MessageTransformationException;

public class TweetTransformer {
  private ObjectMapper mapper = new ObjectMapper();

  public String transform(String payload) {
    try {
      Map<String, Object> tweet = mapper.readValue(payload, new TypeReference<Map<String, Object>>() {});
      return tweet.get("text").toString();
    } catch (IOException e) {
      throw new MessageTransformationException("Unable to transform tweet: " + e.getMessage(), e);
    }
  }
}

Create the module Application Context File

Create the following file as spring-module.xml in the config resource directory:

<?xml version="1.0" encoding="UTF-8"?>

<beans:beans xmlns="http://www.springframework.org/schema/integration"
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
  xmlns:beans="http://www.springframework.org/schema/beans"
  xsi:schemaLocation="http://www.springframework.org/schema/beans
    http://www.springframework.org/schema/beans/spring-beans.xsd
    http://www.springframework.org/schema/integration
    http://www.springframework.org/schema/integration/spring-integration.xsd">
  <channel id="input"/>

  <transformer input-channel="input" output-channel="output">
    <beans:bean class="my.custom.transformer.TweetTransformer" />
  </transformer>

  <channel id="output"/>
</beans:beans>

Alternately, you can create the application context using an @Configuration class. In the example below, we’ve combined the configuration and the transformer into a single Java file for simplicity. Note that TweetTransformer now includes Spring Integration annotations:

package my.custom.transformer;

import java.io.IOException;
import java.util.Map;

import my.custom.transformer.TweetTransformer;
import org.codehaus.jackson.map.ObjectMapper;
import org.codehaus.jackson.type.TypeReference;

import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.integration.annotation.MessageEndpoint;
import org.springframework.integration.annotation.Transformer;
import org.springframework.integration.channel.DirectChannel;
import org.springframework.integration.config.EnableIntegration;
import org.springframework.integration.transformer.MessageTransformationException;
import org.springframework.messaging.MessageChannel;

@Configuration
@EnableIntegration
public class ModuleConfiguration {
  @Autowired
  TweetTransformer transformer;

  @Bean
  public MessageChannel input() {
    return new DirectChannel();
  }

  @Bean
  MessageChannel output() {
    return new DirectChannel();
  }
}

@MessageEndpoint
class TweetTransformer {
  private ObjectMapper mapper = new ObjectMapper();

  @Transformer(inputChannel = "input", outputChannel = "output")
  public String transform(String payload) {
    try {
      Map<String, Object> tweet = mapper.readValue(payload, new TypeReference<Map<String, Object>>() {
      });
      return tweet.get("text").toString();
    }
    catch (IOException e) {
      throw new MessageTransformationException("Unable to transform tweet: " + e.getMessage(), e);
    }
  }
}

To use @Configuration, you must also tell Spring which packages to scan in the module’s properties file spring-module.properties:

base_packages=my.custom.transformer

Write a Test

Writing a test to deploy the module in an embedded single node container requires the spring-xd-dirt and spring-xd-test libraries and a few other things. See the project pom or the gradle build script for details. The following code snippets are from TweetTransformerIntegrationTest

Note

See test a module for some important tips abouts regarding in-container testing.

First we start the SingleNodeApplication and register the module under test by adding a SingletonModuleRegistry providing the module name and type. This looks in the root classpath by default, so will find the module configuration in src/main/resources/config. SingleNodeIntegrationTestSupport provides programmatic access to major beans in the Admin and Container application contexts, as well as the contexts themselves.

/**
 * Unit tests a module deployed to an XD single node container.
 */
public class TweetTransformerIntegrationTest {

  private static SingleNodeApplication application;

  private static int RECEIVE_TIMEOUT = 5000;

  private static String moduleName = "tweet-transformer";

  /**
   * Start the single node container, binding random unused ports, etc. to not conflict with any other instances
   * running on this host. Configure the ModuleRegistry to include the project module.
   */
  @BeforeClass
  public static void setUp() {
    RandomConfigurationSupport randomConfigSupport = new RandomConfigurationSupport();
    application = new SingleNodeApplication().run();
    SingleNodeIntegrationTestSupport singleNodeIntegrationTestSupport = new SingleNodeIntegrationTestSupport
        (application);
    singleNodeIntegrationTestSupport.addModuleRegistry(new SingletonModuleRegistry(ModuleType.processor,
        moduleName));

  }

To implement ths test, we will use the SingleNodeProcessingChain test fixture. The chain is a partial stream definition, represented as Spring XD DSL, which may be a single module, a chain of processors separated by |. In this case we are testing a single module. The chain binds local message handlers that act as source and sink to complete the stream. Thus we can deploy the stream and send messages directly to the source and receive messages directly from the sink:

We could, in theory, test against the actual twittersearch source, but this is not advised because it would depend on connecting to Twitter, providing credentials, etc. So we will save that for when the module is actually installed to the target Spring XD runtime. Instead, we can simply send a message with a sample tweet and verify that we get the content of the text property as output, as expected.

  /**
   * This test creates a stream with the module under test, or in general a "chain" of processors. The
   * SingleNodeProcessingChain is a test fixture that allows the test to send and receive messages to verify each
   * message is processed as expected.
   */
  @Test
  public void test() {
    String streamName = "tweetTest";
    String tweet = "..." //JSON omitted here for clarity

    String processingChainUnderTest = moduleName;

    SingleNodeProcessingChain chain = chain(application, streamName, processingChainUnderTest);

    chain.sendPayload(tweet);

    String result = (String) chain.receivePayload(RECEIVE_TIMEOUT);

    assertEquals("Aggressive Ponytail #freebandnames", result);

    //Unbind the source and sink channels from the message bus
    chain.destroy();
  }

Register the Module

Since the module requires no external dependencies in this case, we can build the project as a simple jar file and install it using the module upload shell command:

xd:>module upload --file [path-to]/tweet-transformer-1.0.0.BUILD-SNAPSHOT.jar --name tweet-transformer --type processor
Successfully uploaded module 'processor:tweet-transformer'

If you make changes and need to re-install, you must first delete the existing module:

xd:>module delete processor:tweet-transformer
Note

A simple jar file works in this case because the module requires no additional library dependencies since the Spring XD class path already includes Jackson and Spring Integration. See Module Packaging for more details.

Test the custom module in the Spring XD runtime:

Start the Spring XD runtime and try creating a stream to test your processor:

xd:> stream create --name javatweets --definition "twittersearch --query=java --consumerKey=<your_key> --consumerSecret=<your_secret> | tweet-transformer | file" --deploy

If you haven’t already used twittersearch, read the sources section for more details. This command should stream tweets to the file /tmp/xd/output/javatweets but, unlike the normal twittersearch output, you should just see the text of the tweet rather than the full JSON document.

Also see si-dsl-module example for a more complex example of a processor module.

Creating a Sink Module

Introduction

As outlined in the modules document, Spring XD currently supports four types of modules: source, sink, and processor for stream processing and job for batch procesing. This document walks through implementing a custom sink module.

The last module in a stream is always a sink. A sink module is built with Spring Integration to consume messages on its input channel and send them to an external resource to terminate the stream.

Spring Integration provides a number of outbound channel adapters to integrate with various transports such as TCP, AMQP, JMS, Kafka, HTTP, web services, mail, or data stores such as file, Redis, MongoDB, JDBC, Splunk, Gemfire, and more. It is straightforward to create a sink module using an existing outbound channel adapters. Such outbound channel adapters are typically used to integrate streams with external data stores or legacy systems. Alternately, you may need to invoke a third party Java API to provide data to an external system. In this case, the sink can easily invoke a Java method using a Service Activator.

Here, we will demonstrate step-by-step how to create and install a sink module using the Spring Integration Redis Store Outbound Channel Adapter. The complete code for this example is redis-store-sink sample project.

Create the module Application Context

Configure the outbound channel adapter in an xml bean definition file under the config resource directoy:

<beans ...>

	<int:channel id="input" />

	<int-redis:store-outbound-channel-adapter
		id="redisListAdapter" collection-type="LIST" channel="input" key="${collection}" auto-startup="false"/>

	<beans:bean id="redisConnectionFactory"
		class="org.springframework.data.redis.connection.jedis.JedisConnectionFactory">
		<beans:property name="hostName" value="${host}" />
		<beans:property name="port" value="${port}" />
	</beans:bean>

</beans>

The adapter, as required by Spring XD, is configured as an endpoint on a channel named input. When a message is consumed, the Redis Store outbound channel adapter will write the payload to a Redis list with a key given by the ${collection} property. By default, the Redis Store outbound channel adapter uses a bean named redisConnectionFactory to connect to the Redis server. Here the connection factory is configured with property placeholders ${host}, ${port} which will be provided as module options in stream definitions that use this sink. Note that auto-startup is set to false. This is a requirement for Spring XD modules. When a stream is deployed, the Spring XD runtime will create and start the modules in the correct order to ensure that everything is initialized before the stream starts processing messages.

Note

By default, the adapter uses a StringRedisTemplate. Therefore, this module will store all payloads directly as Strings. You may configure a RedisTemplate with a different value Serializer to serialize other data types, such as Java objects, to the Redis collection.

Spring XD will automatically register a PropertyPlaceholderConfigurer to your application context, so there is no need to declare one here. These properties correspond to module options defined for this module (discussed below). Users supply option values when creating a stream using the DSL.

The module’s properties file in the config resource directory contains Module Options Metadata including a description, type, and optional default value for each property. The metadata supports features like auto-completion in the Spring XD shell and option validation:

options.collection.description = the name of the list
options.collection.default= ${xd.stream.name}
options.collection.type = java.lang.String
#
options.host.description = the host name for the Redis store
options.host.default= localhost
options.host.type = java.lang.String
#
options.port.description = the port for the Redis store
options.port.default= 6379
options.port.type = java.lang.Integer

Note that the collection defaults to the stream name, referencing a common property provided by Spring XD.

Alternately, you can write a POJO to define the metadata. Using a Java class provides better validation along with additional features and requires that the class be packaged as part of the module.

Create a module project

This section covers creating the module as a standalone project containing some code to test the module. This example uses Maven but Spring XD supports Gradle as well

Take a look at the pom file for this example. You will see it declares spring-xd-module-parent as its parent and declares a dependency on spring-integration-redis which provides the outbound channel adapter. The parent pom provides everything else you need. We also need to configure repositories to access the parent pom and any other dependencies. The xml file containing the bean definitions and the properties file are located in src\main\resources\config.

Create the Spring integration test

The main objective of the test is to ensure that messages are stored in a Redis list once the module’s Application Context is loaded. In order to test the module stand-alone, we need to enhance the module context with property values and a RedisTemplate to retrieve the stored messages.

<beans...>

	<context:property-placeholder properties-ref="props"/>

	<util:properties id="props">
		<prop key="collection">mycollection</prop>
		<prop key="host">localhost</prop>
		<prop key="port">6379</prop>
	</util:properties>

	<import resource="classpath:config/spring-module.xml"/>

	<bean id="redisTemplate" class="org.springframework.data.redis.core.StringRedisTemplate">
		<property name="connectionFactory" ref="redisConnectionFactory"/>
	</bean>
</beans>

Next, create and run the RedisStoreSinkTest:

package org.springframework.xd.samples;

import ...

@RunWith(SpringJUnit4ClassRunner.class)
@ContextConfiguration
public class RedisStoreSinkTest {

	@Autowired
	ConfigurableApplicationContext applicationContext;

	@Autowired
	MessageChannel input;

	@Autowired
	RedisTemplate<String,String> redisTemplate;

	@Test
	public void test() {
		applicationContext.start();
		input.send(new GenericMessage<String>("hello"));
		assertEquals("hello", redisTemplate.boundListOps("mycollection").leftPop(5, TimeUnit.SECONDS));
	}

	@After
	public void cleanUp() {
		redisTemplate.delete("mycollection");
	}

}

The test will load the module application context using our test context and send a message to the module’s input channel. It will fail if the input payload "hello" is not added to the Redis list within 5 seconds.

Run the test

The test requires a running Redis server. See Getting Started for information on installing and starting Redis.

Test the Module Options

Another test you may want to include is one to verify the module options metadata, as defined in spring-module.properties Here is an example ModuleOptionsTest that uses Spring XD’s DefaultModuleOptionsMetadataResolver

package org.springframework.xd.samples;

import ...

/**
 * Tests expected module properties are present.
 */
 public class ModuleOptionsTest {

	@Test
	public void testModuleOptions() {
		ModuleOptionsMetadataResolver moduleOptionsMetadataResolver = new DefaultModuleOptionsMetadataResolver();
		String resource = "classpath:/";
		ModuleDefinition definition = ModuleDefinitions.simple("redis-store", sink, resource);
		ModuleOptionsMetadata metadata = moduleOptionsMetadataResolver.resolve(definition);

		assertThat(
				metadata,
				containsInAnyOrder(moduleOptionNamed("collection"), moduleOptionNamed("host"),
						moduleOptionNamed("port")));

		for (ModuleOption moduleOption : metadata) {
			if (moduleOption.getName().equals("collection")) {
				assertEquals("${xd.stream.name}", moduleOption.getDefaultValue());
			}
			if (moduleOption.getName().equals("port")) {
				assertEquals("6379", moduleOption.getDefaultValue());
			}
			if (moduleOption.getName().equals("host")) {
				assertEquals("localhost", moduleOption.getDefaultValue());
			}
		}
	}

	public static Matcher<ModuleOption> moduleOptionNamed(String name) {
		return hasProperty("name", equalTo(name));
	}
}

Install the module

The next step is to package the module as an uber-jar using maven:

$mvn package

This will build an uber-jar in target/redis-store-sink-1.0.0.BUILD-SNAPSHOT.jar. If you inspect the contents of this jar, you will see it includes the module configuration files and dependent jars (spring-integration-redis in this case). Fire up the Spring XD runtime if it is not already running and, using the Spring XD Shell, install the module as a sink named redis-store using the module upload command:

xd:>module upload --file [path-to]/redis-store-sink/target/redis-store-sink-1.0.0.BUILD-SNAPSHOT.jar --name redis-store --type sink

See registering a module for more details.

Test the module

Once the XD server is running, create a stream to test your new module. This stream will write tweets containing the word "java" to Redis as a JSON string:

xd:> stream create --name javasearch --definition "twittersearch --consumerKey=<your_key> --consumerSecret=<your_secret> --query=java | redis-store --collection=javatweets" --deploy

Note that you need to have a consumer key and secret to use the twittersearch module. See the description in the sources section for more information.

Fire up the redis-cli and verify that tweets are being stored:

$ redis-cli
redis 127.0.0.1:6379> lrange javatweets 0 -1
1) "{\"id\":342386150738120704,\"text\":\"Now Hiring: Senior Java Developer\",\"createdAt\":1370466194000,\"fromUser\":\"jencompgeek\",...\"}"

If you prefer a simpler test, you can create a stream using the http source and manually post data to it:

xd:> stream create --name redisTest --definition "http | redis-store" --deploy
xd:> http post --target http://localhost:9000 --data hello
redis 127.0.0.1:6379> lrange redisTest 0 -1
1) "hello"

Creating a Job Module

Introduction

As outlined in the modules document, XD currently supports four types of modules: source, sink, and processor for stream processing and job for batch procesing. This document walks through creation of a simple job module.

Developing your Job

The Job definitions provided as part of the Spring XD distribution as well as those included in the Spring XD Samples repository can be used a basis for building your own custom Jobs. The development of a Job largely follows the development of a Spring Batch job, for which there are several references.

For help developing Job steps specific to Hadoop, e.g. HDFS, Pig, Hive, the Spring XD Samples is useful as well as the following resources

Creating a Simple Job

First we’ll look at how to create a job module from scratch. The complete working example is here.

Create a Module Project

This section covers the setup of a standalone project containing the module configuration and custom code. This example uses Maven but Spring XD supports Gradle as well.

Take a look at the pom file for this example. You will see it declares spring-xd-module-parent as its parent. The parent pom provides support for building and packaging Spring XD modules, including spring-batch libraries. We also need to configure repositories to access the parent pom and its dependencies.

First create a java project for your module, named batch-simple in your favorite IDE.

Create the Spring Batch Job Definition

Create a The job definition file in src\main\resources\config. In this case, we use a custom Tasklet. In this example there is only one step and it simply prints out the job parameters.

<?xml version="1.0" encoding="UTF-8"?>
<beans ...>

	<batch:job id="job">
		<batch:step id="helloSpringXDStep">
			<batch:tasklet ref="helloSpringXDTasklet" />
		</batch:step>
	</batch:job>

	<bean id="helloSpringXDTasklet"
		class="org.springframework.springxd.samples.batch.HelloSpringXDTasklet" />

</beans>

Write the Tasklet

Write a HelloSpringXDTasklet java class that implements Tasklet. This will retrieve the job parameters and print them to stdout.

package org.springframework.springxd.samples.batch;

import ...

public class HelloSpringXDTasklet implements Tasklet, StepExecutionListener {

	private volatile AtomicInteger counter = new AtomicInteger(0);

	public HelloSpringXDTasklet() {
		super();
	}

	public RepeatStatus execute(StepContribution contribution,
			ChunkContext chunkContext) throws Exception {

		final JobParameters jobParameters = chunkContext.getStepContext().getStepExecution().getJobParameters();
		final ExecutionContext stepExecutionContext = chunkContext.getStepContext().getStepExecution().getExecutionContext();

		System.out.println("Hello Spring XD!");

		if (jobParameters != null && !jobParameters.isEmpty()) {

			final Set<Entry<String, JobParameter>> parameterEntries = jobParameters.getParameters().entrySet();

			System.out.println(String.format("The following %s Job Parameter(s) is/are present:", parameterEntries.size()));

			for (Entry<String, JobParameter> jobParameterEntry : parameterEntries) {
				System.out.println(String.format(
						"Parameter name: %s; isIdentifying: %s; type: %s; value: %s",
						jobParameterEntry.getKey(),
						jobParameterEntry.getValue().isIdentifying(),
						jobParameterEntry.getValue().getType().toString(),
						jobParameterEntry.getValue().getValue()));

				if (jobParameterEntry.getKey().startsWith("context")) {
					stepExecutionContext.put(jobParameterEntry.getKey(), jobParameterEntry.getValue().getValue());
				}
			}

			if (jobParameters.getString("throwError") != null
					&& Boolean.TRUE.toString().equalsIgnoreCase(jobParameters.getString("throwError"))) {

				if (this.counter.compareAndSet(3, 0)) {
					System.out.println("Counter reset to 0. Execution will succeed.");
				}
				else {
					this.counter.incrementAndGet();
					throw new IllegalStateException("Exception triggered by user.");
				}

			}
		}
		return RepeatStatus.FINISHED;
	}

	@Override
	public void beforeStep(StepExecution stepExecution) {
	}

	@Override
	public ExitStatus afterStep(StepExecution stepExecution) {
		// To make the job execution fail, set the step execution to fail
		// and return failed ExitStatus
		// stepExecution.setStatus(BatchStatus.FAILED);
		// return ExitStatus.FAILED;
		return ExitStatus.COMPLETED;
	}
}

Package and install the Module:

Follow the instructions in the project README for more details. The steps are summarized here.

Build the project with maven:

$mvn package

Uupload the jar file to Spring XD and register it as a job module named myjob using the Spring XD shell module upload command:

xd:>module upload --type job --name myjob --file [path-to]/batch-simple/target/springxd-batch-simple-1.0.0.BUILD-SNAPSHOT.jar

Modules can reside in an expanded directory named after the module, e.g. modules/job/myjob or as a single uber-jar, e.g., modules/job/myjob.jar. See module packaging and registering a modulefor more details.

Run the job

Start the Spring XD container if it is not already running.

xd:> job create --name helloSpringXD --definition "myjob" --deploy
xd:> job launch helloSpringXD --params {"myStringParameter":"foobar","-secondParam(long)":"123456"}
Note

By default, deploy is set to false. "--deploy" or "--deploy true" will deploy the job along with job creation.

In the console log of the Spring XD container you should see the following:

Hello Spring XD!
The following 3 Job Parameter(s) is/are present:
Parameter name: secondParam; isIdentifying: false; type: LONG; value: 123456
Parameter name: myStringParameter; isIdentifying: true; type: STRING; value: foobar
Parameter name: random; isIdentifying: true; type: STRING; value: 0.06893349621991496

Creating a read-write processing Job

To create a job in the XD shell, execute the job create command specifying:

  • name - the "name" that will be associated with the Job

  • definition - the name of the job module

Often a batch job will involve reading batches of data from a source, tranforming or processing that data and then wrting the batch of data to a destination. This kind of flow is implemented using Chunk-oriented processing, represented in the job configuration using the <chunk/> element containing reader, writer and optional processor elements. Other attributes define the size of the chunck and various policies for handling failure cases.

You will usually be able to reuse existing reader and writer implementations. The filejdbc job provided with the Spring XD distribution shows an example of this using the standard File reader and JDBC writer.

The processor is based on the ItemProcessor interface. It has a generic signature that lets you operate on a record at at time. The batch of records is handled as a collection in reader and writer implementations. In the filejdbc job, the reader converts input records into a Spring XD Tuple. The tuple serves as a generic data structure but you can also use or write another converter to convert the input record to your own custom POJO object.

Orchestrating Hadoop Jobs

There are several tasklet implementation that will run various types of Hadoop Jobs

The Spring Hadoop Samples project provides examples of how to create batch jobs that orchestate various hadoop jobs at each step. You can also mix and match steps related to work that is executed on the Hadoop cluster and work that is executed on the Spring XD cluster.

Creating a Python Module

Introduction

Spring XD provides support for processor and sink modules that invoke an external shell command. You can use these to integrate a Python script with a Spring XD stream. The following echo.py script is a simple example which can implement a processor to simply echo the input.

#echo.py
import sys

#=====================
# Write data to stdout
#=====================
def send(data):
  sys.stdout.write(data)
  sys.stdout.flush()

#===========================================
# Terminate a message using the default CRLF
#===========================================
def eod():
  send("\r\n")

#===========================
# Main - Echo the input
#===========================

while True:
  try:
    data = raw_input()
    if data:
      send(data)
      eod()
  except:
    break

To use this in a stream, create a stream definition like this:

xd:>stream create pytest --definition "time | shell --command='python <absolute-path-to>/echo.py' | log" --deploy
Created and deployed new stream 'pytest'
Note

Python must be installed on the host of any container to which the processor module is deployed.

You should see the time messages echoed in the Spring XD container log. The shell processor works by binding its message channels to the external process' stdin and stdout. Behind the scenes, the shell modules use java.lang.ProcessBuilder to connect to the shell process. As you can see, most of echo.py is boilerplate code. To make things easier, Spring XD provides a python module to handle all of the low level I/O.

from springxd.stream import Processor

def echo(data):
    return data

process = Processor()
process.start(echo)

As you can see, this creates a Processor object which has a start method to which you may pass any function that accepts a single argument and returns a value. Currently, both the input and output data must be strings. Processor uses Encoders.CRLF (\r\n) by default. This is how the Spring XD module delimits individual messages in the stream. Encoders.LF is also supported. The shell command processor also uses CRLF by default.

xd:>stream create pytest --definition "time | shell --command='python <absolute-path-to>/echo.py' | log" --deploy
Created and deployed new stream 'pytest'

Alternately, you can specify the LF encoder in the Python script and the stream definition:

from springxd.stream import Processor, Encoders

def echo(data):
    return data

process = Processor(Encoders.LF)
process.start(echo)
xd:>stream create pytest --definition "time | shell --command='python <absolute-path-to>/echo.py' --encoder=LF | log" --deploy

The stream module also provides a similar Sink object which accepts a function that need not return a value (Sink will ignore the returned value).

Note

In order to import the springxd.stream module into your script, you must include it in your Python module search path. Python provides several ways to do this as described here. Spring XD python modules are included in the distribution in the python directory. The stream module is designed to be version agnostic and has been tested against Python 2.7.6 and Python 3.4.2

Providing Module Options Metadata

Introduction

Each available module can expose metadata about the options it accepts. This is useful to enhance the user experience, and is the foundation to advanced features like contextual help and code completion. the For example, provided that the file source module has been enriched with options metadata (and it has), one can use the module info command in the shell to get information about the module:

xd:> module info source:file
Information about source module 'file':

  Option Name        Description                                                                  Default  Type
  -----------------  ---------------------------------------------------------------------------  -------  ---------
  dir                the absolute path to the directory to monitor for files                      <none>   String
  pattern            a filter expression (Ant style) to accept only files that match the pattern  *        String
  outputType         how this module should emit messages it produces                             <none>   MediaType
  preventDuplicates  whether to prevent the same file from being processed twice                  true     boolean
  ref                set to true to output the File object itself                                 false    boolean
  fixedDelay         the fixed delay polling interval specified in seconds                        5        int

For this to be available, module authors have to provide a little bit of extra information, known as "Module Options Metadata". That metadata can take two forms, depending on the needs of the module: one can either use the "simple" approach, or the "POJO" approach. If one does not need advanced features like profile activation, validation or options encapsulation, then the "simple" approach is sufficient.

Using the "Simple" approach

To use the simple approach, simply create a file named <module>.properties right next to the <module>.xml file for your module.

Declaring and documenting an option

In that file, each option <option> is declared by adding a line of the form

options.<option>.description = the description

The description for the option is the only required part, and is a very important piece of information for the end user, so pay special attention to it (see also Style remarks)

That sole line in the properties file makes a --<option>= construct available in the definition of a stream using your module.

Note
About plugin provided options metadata

Some options are automatically added to a module, depending on its type. For example, every source module automatically inherits a outputType option, that controls the type conversion feature between modules. You don’t have to do anything for that to happen.

Similarly, every job module benefits from a handful of job specific options.

Here is a recap of those automatically provided options:

Module Type Options

Source

outputType

Processor

outputType, inputType

Sink

inputType

Job

makeUnique, numberFormat, dateFormat

Advertising default values

In addition to this, one can also provide a default value for the option, using

options.<option>.default = SomeDefault

Doing this, the default value should not be used in the placeholder syntax in the xml file. Assuming this is the contents of foo.properties:

options.bar.description = a very useful option
options.bar.default = 5

then in foo.xml:

<!-- this is correct -->
<feature the-bar="${bar}"" />

<!-- this is incorrect/not needed -->
<feature the-bar="${bar:5}" />

Exposing the option type

Lastly, one can document the option type using a construct like

options.<option>.type = fully.qualified.class.Name

For simple "primitive" types, one can use short names, like so:

options.<option>.type = String
or
options.<option>.type = boolean
or
options.<option>.type = Integer

Note that there is support for both wrapper types (e.g. Integer) and primitive types (e.g. int). Although this is used for documentation purposes only, the primitive type would typically be used to indicate a required option (null being prohibited).

Using the "POJO" approach

To use advanced features such as profile activation driven by the values provided by the end user, one would need to leverage the "POJO" approach.

Instead of writing a properties file, you will need to write a custom java class that will hold the values at runtime. That class is also introspected to derive metadata about your module.

Declaring options to the module

For the simplest cases, the class you need to write does not need to implement or inherit from anything. The only thing you need to do is to reference it in a properties file named after your module (the same file location you would have used had you been leveraging the "simple" approach):

options_class = fully.qualified.name.of.your.Pojo

Note that the key is options_class, with an s and an underscore (not to be confused with option.<optionname> that is used in the "simple" approach)

For each option you want available using the --<option>= syntax, you must write a public setter annotated with @ModuleOption, providing the option description in the annotation.

The type accepted by that setter will be used as the documented type.

That setter will typically be used to store the value in a private field. How the module application can get ahold of the value is the topic of the next section.

Exposing values to the context

For a provided value to be used in the module definition (using the ${foo} syntax), your POJO class needs to expose a getFoo() getter.

At runtime, an instance of the POJO class will be created (it requires a no-arg constructor, by the way) and values given by the user will be bound (using setters). The POJO class thus acts as an intermediate PropertySource to provide values to ${foo} constructs.

Providing defaults

To provide default values, one would most certainly simply store a default value in the backing field of a getter/setter pair. That value (actually, the result of invoking the matching getter to a setter on a newly instanciated object) is what is advertised as the default.

Encapsulating options

Although one would typically use the combination of a foo field and a getFoo(), setFoo(x) pair, one does not have to.

In particular, if your module definition requires some "complex" (all things being relative here) value to be computed from "simpler" ones (e.g. a suffix value would be computed from an extension option, that would take care of adding a dot, depending on whether it is blank or not), then you’d simply do the following:

public class MyOptions {
	private String extension;

	@ModuleOption("the file extension to use")
	public void setExtension(String extension) {
	    this.extension = extension;
	}

	public String getSuffix() {
	    return extension == null ? null : "." + extension;
	}
}

This would expose a --extension= option, being surfaced as a ${suffix} placeholder construct.

The astute reader will have realized that the default can not be computed then, because there is no getExtension() (and there should not be, as this could be mistakenly used in ${extension}). To provide the default value, you should use the defaultValue attribute of the @ModuleOption annotation.

Using profiles

The real benefit of using a POJO class for options metadata comes with advanced features though, one of which is dynamic profile activation.

If the set of beans (or xml namespaced elements) you would define in the module definition file depends on the value that the user provided for one or several options, then you can make your POJO class implement ProfileNamesProvider. That interface brings one contract method, profilesToActivate() that you must implement, returning the names of the profiles you want to use (this method is invoked after user option values have been bound, so you can use any logic involving those to compute the list of profile names).

As an example of this feature, see e.g. TriggerSourceOptionsMetadata.

Using validation

Your POJO class can optionally bear JSR303 annotations. If it does, then validation will occur after values have been successfully bound (understand that injection can fail early due to type incoherence by the way. This comes for free and does not require JSR303 annotations).

This can be used to validate a set of options passed in (some are often mutually exclusive) or to catch misconfiguration earlier than deployment time (e.g. a port number cannot be negative).

Metadata style remarks

To provide a uniform user experience, it is better if your options metadata information adheres to the following style:

  • option names should follow the camelCase syntax, as this is easier with the POJO approach. If we later decide to switch to a more unix-style, this will be taken care of by XD itself, with no change to the metadata artifacts described here

  • description sentences should be concise

  • descriptions should start with a lowercase letter and should not end with a dot

  • use primitive types for required numbers

  • descriptions should mention the unit for numbers (e.g ms)

  • descriptions should not describe the default value, to the best extent possible (this is surfaced thru the actual default metadata awareness)

  • options metadata should know about the default, rather than relying on the ${foo:default} construct

Extending Spring XD

Introduction

This document describes how to customize or extend the Spring XD Container. Spring XD is a distributed runtime platform delivered as executable components including XD Admin, XD Container, and XD Shell. The XD Container is a Spring application combining XML resources, Java @Configuration classes, and Spring Boot auto configuration for its internal configuration, initialized via the Spring Boot SpringApplicationBuilder. Since Spring XD is open source, the curious user can see exactly how it is configured. However, all Spring XD’s configuration is bundled in jar files and therefore not directly accessible to end users. Most users do not need to customize or extend the XD Container. For those that do, Spring XD provides hooks to:

  • Provide additional bean definitions

  • Override selected bean definitions with custom implementations

Customization scenarios might include:

This following sections provide an overview of XD Container internals and explain how to extend Spring XD for each of these scenarios. The reader is expected to have a working knowledge of both the Spring Framework and Spring Integration.

Spring XD Application Contexts

The diagram below shows how Spring XD is organized into several Spring application contexts. Some understanding of the Spring XD application context hierarchy is necessary for extending XD. In the diagram, solid arrows indicate a parent-child relationship. As with any Spring application a child application context may reference beans defined in its parent application context, but the parent context cannot access beans defined in the child context. It is important to keep in mind that a bean definition registered in a child context with the same id as a bean in the parent context will create a separate instance in the child context. Similarly, any bean definition will override an earlier bean definition in the same application context registered with the same id (Sometimes referred to as "last one wins").

Spring XD’s primary extension mechanism targets the Plugin Context highlighted in the diagram. Using a separate convention, it is also possible to register an alternate MessageBus implementation in the Shared Server Context.

Extension contexts
Figure 32. The Spring XD Application Context Hierarchy

While this arrangement of application contexts is more complex than the typical Spring application, XD is designed this way for the following reasons:

  • Bean isolation - Some beans are "global" in that they are shared by all XD runtime components: Admin, Container, and Modules. Those allocated to the Shared Server Context are shared only by Admin and Container. Some beans must be available to Plugins ,used to configure Modules. However Plugins and Modules should be isolated from critical internal components. While complete isolation has proven difficult to achieve, the intention is to minimize any undesirable side effects when introducing extensions.

  • Bean scoping - To ensure that single node and distributed configurations of the Spring XD runtime are logically equivalent, the Spring configuration is identical in both cases, avoiding unnecessary duplication of bean definitions.

  • Lifecycle management - Plugins and other beans used to configure these application contexts are also Spring beans which Spring XD dynamically "discovers" during initialization. Such components must be fully instantiated prior to the creation of the application context to which they are targeted. To ensure initialization happens in the desired order, such beans may be either defined in an isolated application context (i.e., not part of the hierarchy) or in a parent context which Spring initializes before any of its descendants.

Plugin Architecture

The XD Container at its core is simply a runtime environment for hosting and managing micro Spring applications called Modules. Each module runs in its own application context (Module Context). The Module Context is a child of Global Context, as modules share some bean definitions, but otherwise is logically isolated from beans defined in the XD Container. The Module Context is fundamental to the Spring XD design. In fact, this is what allows each module to define its own input and output channels, and in general, enables beans be uniquely configured via property placeholders evaluated for each deployed instance of a Module. The Module interface and its default implementation provide a thin wrapper around a Spring Application Context for which properties are bound, profiles activated, and beans added or enhanced in order to "plug" the module into the XD Container.

The ModuleDeployer, shown in the diagram, is a core component of the Container Context, responsible for initializing modules during deployment, and shutting them down during undeployment. The ModuleDeployer sees the module as a "black box", unaware of its purpose or runtime requirements. Binding a module’s channels to XD’s data transport, for instance, is the responsibility of the MessageBus implementation configured for the transport. The MessageBus binding methods are actually invoked by the StreamPlugin during the initialization of a stream module. To support jobs, XD provides a JobPlugin to wire the Spring Batch components defined in the module during deployment. The JobPlugin also invokes the MessageBus to support communications between XD and job modules. These, and other functions critical to Spring XD are performed by classes that implement the Plugin interface. A Plugin operates on every deployed Module which it is implemented to support. Thus the ModuleDeployer simply invokes the deployment life cycle methods provided by every Plugin registered in the Plugin Context.

The ModuleDeployer discovers registered Plugins by calling getBeansOfType(Plugin.class) for the Plugin Context (its parent context). This means that adding your own Plugin requires these steps:

  • Implement the Plugin interface

  • Add your Plugin implementation and any dependent classes to Spring XD’s class path

  • Follow conventions that Spring XD uses to register Plugins

The next section covers these steps in more detail.

How to Add a Spring bean to the XD Container

This section applies to adding a Plugin, which is generally useful since a Plugin has access to every module as it is being deployed (see the previous section on Plugin Architecture). Furthermore, this section describes a generic mechanism for adding any bean definition to the Plugin Context. Spring XD uses both Spring Framework’s class path component scanning and resource resolution to find any components that you add to specified locations in the class path. This means you may provide Java @Configuration and/or any classes annotated with the @Component stereotype in a configured base package in addition to bean definitions defined in any XML or Groovy resource placed under a configured resource location. These locations are given by the properties xd.extensions.locations and xd.extensions.basepackages, optionally configured in servers.yml down at the bottom:

# User Extensions: Where XD scans the class path to discover extended container configuration to adds beans to the Plugins context.
# Each property may be a comma delimited string. 'basepackages' refers to package names used for
# annotated component (@Configuration or @Component stereotypes) scanning. 'locations' is a list of root resource directories containing XML or Groovy configuration.
# XD prepends classpath:* if no prefix included and appends **/*.* to each location
#xd:
#  extensions:
#      basepackages: com.acme.xd.extensions
#      locations: META-INF/spring-xd/ext

As the pluralization of these property names suggests, you may represent multiple values as a comma delimited string. Also note that there is no default for xd.extensions.basepackages. So if you want to use annotation based configuration, you must first set up one or more base package locations. The resource location(s) define the root locations where any XML or Groovy Spring bean definition file found in the given root or any of its subdirectories will be loaded. The root location defaults to META-INF/spring-xd/ext

The Container loads any bean definitions found in these configured locations on the class path and adds them to the Plugin Context. This is the appropriate application context since in order to apply custom logic to modules, you will most likely need to provide a custom Plugin.

Note
The extension mechanism is very flexible. In theory, one can define BeanPostProcessors, BeanFactoryPostProcessors, or ApplicationListeners to manipulate Spring XD application contexts. Do so at your own risk as the Spring XD initialization process is fairly complex, and not all beans are intended to be extensible.

Extensions are packaged in a jar file which must be added to Spring XD’s class path. Currently, you must manually copy the jar to $XD_HOME/lib for each container instance. To implement a Plugin, you will need to include a compile time dependency on spring-xd-module in your build. To access other container classes and to test your code in a container you will also require spring-xd-dirt.

Providing A new Type Converter

Spring XD supports automatic type conversion to convert payloads declaratively. For example, to convert an object to JSON, you provide the module option --outputType=application/json to a module used in a stream definition. The conversion is enabled by a Plugin that binds a Spring MessageConverter to a media type. The default type converters are currently configured in streams.xml, packaged in spring-xd-dirt-<version>.jar. If you look at that file, you can see an empty list registered as customMessageConverters.

<!-- Users can override this to add converters.-->
	<util:list id="customMessageConverters"/>

So registering new type converters is a matter of registering an alternate list as customMessageConverters to the application context. Spring XD will replace the default empty list with yours. xd.messageConverters and customMessageConverters are two lists injected into the ModuleTypeConversionPlugin to build an instance of CompositeMessageConverter which delegates to the first converter in list order that is able to perform the necessary conversion. The Plugin injects the CompositeMessageConverter into the module’s input or output the MessageChannel, corresponding to the inputType or outputType options declared for any module in the stream definition (or defined as the module’s default inputType).

The CompositeMessageConverter is desirable because a module does not generally know what payload type it will get from its predecessor. For example, the converters that Spring XD provides out of the box can convert any Java object, including a Tuple and a byte array to a JSON String. However the methods for converting a byte array or a Tuple are each optimized for the respective type. The CompositeMessageConverter for --outputType=application/json must provide all three methods and the Data Type channel chooses the first converter that applies to both the incoming payload type and the media type (e.g., application/json). Note that the order that the converters appear in the list is significant. In general, converters for specific payload types precede more general converters for the same media type. The customMessageConverters are added after the standard converters in the order defined. So it is generally easier to add converters for new media types than to replace existing converters.

For example, a member of the Spring XD community inquired about Spring XD’s support for Google protocol buffers. This user was interested in integrating Spring XD with an existing messaging system that uses GPB heavily and needed a way to convert incoming and outgoing GPB payloads to interoperate with XD streams. This could be accomplished by providing a customMessageConverters bean containing a list of required message converters. Writing a custom converter to work with XD requires extending AbstractFromMessageConverter provided by spring-xd-dirt. It is recommended to review the existing implementations listed in streams.xml to get a feel for how to do this. In addition, you would likely define a custom MimeType such as application/gpb.

Note
It is worth mentioning that GPB is commonly used for marshaling objects over the network. In the context of Spring XD marshaling is treated as a separate concern from payload conversion. In Spring XD, marshaling happens at the "pipe" indicated by the | symbol using a different serialization mechanism, described below. In this case, the GPB payloads are produced and consumed by systems external to Spring XD and need to be converted in order that a GPB payload can work with XD streams. In this scenario, if the GPB is represented as a byte array, the bytes are transmitted over the network directly and marshaling is unnecessary.

As an illustration, suppose this user has developed a source module that emits GPB payloads from a legacy service. Spring XD provides transform and filter modules that accept SpEL expressions to perform their respective tasks. These modules are useful in many situations but the SpEL expressions generally require a POJO representing a domain type, or a JSON string. In this case it would be convenient to support stream definitions such as

gpb-source --outputType=application/x-java-object | transform --expression=...

where gpb-source represents a custom module that emits a GPB payload and expression references some specific object property. The media type application/x-java-object is a convention used by XD to indicate that the payload should be converted to a Java type embedded in the serialized representation (GPB in this example). Alternately, converting to JSON could be performed if the stream definition were:

gpb-source --outputType=application/json | transform --expression=...

To convert an XD stream result to GPB to be consumed by an external service might look like:

source | P1 ... | Pn | gpb-sink --inputType=application/gpb

These examples would require registering custom MessageConverters to handle the indicated conversions. Alternately, this may be accomplished by writing custom processor modules to perform the required conversion. The above examples would then have stream definitions that look more like:

gpb-source | gpb-to-pojo | transform --expression=...
source | P1 ... | Pn | json-to-gpb | gpb-sink
Tip
While custom processor modules are easier to implement, they add unnecessary complexity to stream definitions that use them. If such conversions are required everywhere, enabling automatic conversion may be worth the effort. Also, note that using a separate module generally requires additional network hops (at each pipe). If a processor module is necessary only to perform a common payload conversion, it is more efficient to install a custom converter.

Adding a New Data Transport

Spring XD offers Redis and Rabbit MQ for data transport out of the box. Transport is configured simply by setting the property xd.transport to redis or rabbit. In addition xd-singlenode supports a --transport command line option that can accept local(the single node default) in addition. This simple configuration mechanism is supported internally by an import declaration that binds the transport implementation to a name.

<import resource="classpath*:/META-INF/spring-xd/transports/${XD_TRANSPORT}-bus.xml"/>

The above snippet is from an internal Spring configuration file loaded into the Shared Server Context. Spring XD provides MessageBus implementations in META-INF/spring-xd/transports/redis-bus.xml and META-INF/spring-xd/transports/rabbit-bus.xml

This makes it relatively simple for Spring XD developers and advanced users to provide alternate MessageBus implementations to enable a new transport and activate that transport by setting the xd.transport property. For example, to implement a JMS MessageBus you would add a jar containing /META-INF/spring-xd/transports/jms-bus.xml in the class path. This file must register a bean of type MessageBus with the ID messageBus. A jar providing the above configuration file along with the MessageBus implementation and any dependencies must be installed $XD_HOME/lib.

When implementing a MessageBus, it is advisable to review and understand the existing implementations which extend MessageBusSupport. This base class performs some common tasks including payload marshaling. Spring XD uses the term codec to connote a component that performs both serialization and deserialization and provides a bean with the same name. In the example above, the JMS MessageBus configuration`/META-INF/spring-xd/transports/jms-bus.xml` might look something like:

<bean id="messageBus" class="my.example.JmsMessageBus">
	<constructor-arg ref="jmsConnectionFactory" />
	<constructor-arg ref="codec"/>
</bean>

where JmsMessageBus extends MessageBusSupport and the developer is responsible for configuring any dependent JMS resources appropriately.

Optimizing Serialization

Introduction

Spring XD is configured by default to use Kryo to serialize and deserialize POJO message payloads when using remote transport such as Redis, Rabbit, or Kafka. Note that if the payload is already in the form of a byte array, then XD will transport it as is. Also if the payload is a String, XD will use getBytes() and bypass Kryo. Kryo performs favorably compared to other JVM serialization libraries but may become a bottleneck in high-throughput applications. In rare cases, custom serialization may be needed to address functional issues. This section offers some tips and techniques for customizing and optimizing serialization to address such situations. In accordance with the usual caveats about premature optimization, don’t apply these techniques unless you really need to. Furthermore, the time it takes to serialize and deserialize an object is proportional to factors such the number of fields, composition, and collection sizes. Prerequisite to the techniques presented here, consider colocating modules (via local binding or module composition), or using lighter payloads.

Serialization Performance

Published serialization benchmarks have shown that Kryo serialization operations take less than 1000 nanoseconds for reasonably sized objects. The Spring XD team has independently verified these results. Serializing large objects may take many orders of magnitude longer. To put this in perspective, let’s suppose that serialization and deserialization of your payload each take 5000 nanoseconds. For a stream ,e.g., some-source | some-sink running in a distributed XD runtime, moving the payload from the source to the sink requires serialization at the producer and deserialization at the consumer. Both operations would account for 10,000 ns or 10 microseconds. So the maximum theoretical throughput for this simple stream, with no partitioning or parallel deployment, is at most 100,000 messages/sec. This does not account for network latency (transporting the serialized bytes), overhead in the messaging middleware, or computations performed by the modules. If your application requires a higher level of throughput than mentioned above then serialization is a potential bottleneck. However, if your throughput requirements are not so demanding, serialization performance becomes less of an issue. In this hypothetical scenario, At a throughput of 10,000 messages/second, 10 microseconds is only 10% of the maximum time required to process the message. However if you trying to stream large objects at this rate, serialization time, along with transport and processing time, will be significantly higher.

Serialization in XD

In Spring XD 1.2.x, MessageBus implementations inject a bean of type MultiTypeCodec, where the term Codec refers to an component providing methods to serialize and deserialize objects. MultiType means that the component can handle any object. Spring XD is internally configured with a PojoCodec that delegates to Kryo and provides hooks to register custom Kryo serializers for specific types.

Customizing Kryo

By default, Kryo uses delegates unknown Java types to its FieldSerializer. Kryo also registers default serializers for each primitive type along with String, Collection and Map serializers. FieldSerializer uses reflection to navigate the object graph. A more efficient approach is to implement a custom serializer that is aware of the object’s structure and can directly serialize selected primitive fields:

public class AddressSerializer extends Serializer<Address> {
	@Override
	public void write(Kryo kryo, Output output, Address address) {
		output.writeString(address.getStreet());
		output.writeString(address.getCity());
		output.writeString(address.getCountry());
	}

	@Override
	public Address read(Kryo kryo, Input input, Class<Address> type) {
		return new Address(input.readString(),input.readString(),input.readString());
	}
}

The Serializer interface exposes Kryo, Input, and Output which provide complete control over which fields are included and other internal settings as described in the documentation.

Disabling References

A simple setting may be applied to boost performance if the payload types do not contain any cyclical references. If the payload contains cyclical references, then Kryo needs to do some extra work in this case and the references setting is enabled by default. If not needed, this code can be disabled and can have a measurable impact on performance. In Spring XD this is controlled by the property xd.codec.kryo.references in servers.yml. Set this property to false to disable references.

Registering a Custom Kryo Serializer in XD

If custom serialization is indicated, please consult the Kryo documentation since you will be using the native API. The general requirement is to provide a jar on the Spring XD class path (copy it to <XD_INSTALL_DIR>/xd/lib). This may be the same jar that contains the domain objects or a separate jar that contains the custom Kryo serializer(s) along with a bit of Spring configuration to be imported into the XD runtime.

First provide one or more classes that extend com.esotericsoftware.kryo.Serializer. Next provide a Spring @Configuration in the package spring.xd.bus.ext. For example:

package spring.xd.bus.ext;
...

@Configuration
public class CustomKryoRegistrarConfig {
	@Bean
	public KryoRegistrar myCustomRegistration() {
		List<Registration> registrations = new ArrayList<>();
		registrations.add(new Registration(MyObject.class, new MySerializer(),62));
		return new KryoRegistrationRegistrar(registrations);
	}

// Every thing self contained here for illustration purposes only.
	public static class MyObject {}

	public static class MySerializer extends Serializer<MyObject> {
		@Override
		public void write(Kryo kryo, Output output, MyObject object) {
			...
		}

		@Override
		public MyObject read(Kryo kryo, Input input, Class<MyObject> type) {
			MyObject result;
			...
			return result;
		}
	}
}

The above example works by configuring a KryoRegistrationRegistrar. This class holds a list of com.esotericsoftware.kryo.Registration each of which associates a Java class to a Serializer and a unique integer. The integer is the registration ID for the type which allows Kryo to encode the serialized type as an integer instead of writing the fully qualified class name. This significantly reduces the size of the serialized payload. Spring XD will inject any beans of type KryoRegistrar found on the class path into the PojoCodec. Hence a jar containing a Spring configuration similar to the above and installed in xd/lib will register the custom serializer. One caveat is that multiple KryoRegistrars may contain conflicting registrations. The ID assigned must be unique, and there may not be multiple registrations for the same class. PojoCodec will merge and validate all registrations during container initialization so that any conflicts will result in an exception during container initialization.

Note
The codec must be configured exactly the same way in every container instance. This it is important to keep custom jars and other related runtime configuration consistent. The container logs include the Kryo registration settings.

Implementing KryoSerializable

If you have write access to the domain object source code it may implement KryoSerializable as described here. In this case the class provides the serialization methods itself and no further configuration is required for Spring XD. This has the advantage of being much simpler to use with XD, however benchmarks have shown this is not quite as efficient as registering a custom serializer explicitly:

public class Address implements KryoSerializable {
	...

	@Override
	public void write(Kryo kryo, Output output) {
		output.writeString(this.street);
		output.writeString(this.city);
		output.writeString(this.country);
	}

	@Override
	public void read(Kryo kryo, Input input) {
		this.street = input.readString();
		this.city = input.readString();
		this.country = input.readString();
	}
}

Note that this technique can also be used to wrap a serialization library other than Kryo.

Using DefaultSerializer Annotation

Kryo also provides an annotation as described here.

@DefaultSerializer(SomeClassSerializer.class)
public class SomeClass {
       // ...
}

If you have write access to the domain object this may be a simpler alternative to specify a custom serializer. Note this does not register the class with an ID, so your mileage may vary. This may be combined with using a KryoClassMapRegistrar or KryoClassListRegistrar to register objects if necessary, but then there is less benefit to using the annotation.

Replacing PojoCodec

It is also possible to replace PojoCodec with an implementation of MultiTypeCodec that uses another serialization library in place of Kryo. XD does not provide an alternate implementation, but if one were inclined to write one, a configuration similar to this, in the spring.xd.bus.ext package, is required:

package spring.xd.bus.ext;
...

@Configuration
public class CustomCodecConfig {
	@Bean //must be named 'codec'
	public MultiTypeCodec<Object> codec() {
		return new CustomCodec(...);
	}
}

Benchmarking

Prior to adding any serialization configuration to XD, we highly recommend running some benchmark tests to measure serialization of your data in isolation. It is important to first establish a baseline measurement. Once the baseline performance is known, you can readily measure the impact of optimizations. Serialization has been measured on the order of few hundred nanoseconds. At this scale, it is important to test in an environment which does not have external processes competing for resources. This type of microbenchmark must also account for JVM optimizations and garbage collection by "warming up" (letting the test run for a while before starting the timer) , requesting GC and pausing between runs, and the like. Such tests can also be run with a JVM profiling tool such as Yourkit to get to the finest level of detail.

An excellent resource on JVM serialization benchmarks is the jvm-serializers project which incidentally demonstrates that manually optimized Kryo is the fastest among the libraries tested. The Spring XD samples repository includes a serialization-benchmarks project that has co-opted some of the jvm-serializer techniques and contains sample benchmarks, including one which closely matches results for the jvm-serializers kryo_manual test using XD’s PojoCodec as an entry point. You can use one of these samples as a template to benchmark your custom serializer.

Samples

We have a number of sample projects in the Spring XD Samples GitHub repository. Below are some additional examples for ingesting syslog data to HDFS.

Syslog ingestion into HDFS

In this section we will show a simple example on how to setup syslog ingestion from multiple hosts into HDFS.

Create the streams with syslog as source and HDFS as sink (Please refer to source and sink)

If you’re using syslog over TCP and need the highest throughput, use the Reactor-backed syslog module.

xd:> stream create --definition "reactor-syslog --port=<tcp-port> | hdfs" --name <stream-name>

The reactor-syslog module doesn’t yet support UDP (though it soon will), so if you’re using syslog over UDP you’ll want to use the standard syslog module.

xd:> stream create --definition "syslog-udp --port=<udp-port> | hdfs" --name <stream-name>
xd:> stream create --definition "syslog-tcp --port=<tcp-port> | hdfs" --name <stream-name>

Please note for hdfs sink, set rollover parameter to a smaller value to avoid buffering and to see the data has made to HDFS (incase of smaller volume of log).

Configure the external hosts’ syslog daemons forward their messages to the xd-container host’s UDP/TCP port (where the syslog-udp/syslog-tcp source module is deployed).

A sample configuration using syslog-ng

Edit syslog-ng configuration (for example: /etc/syslog-ng/syslog-ng.conf):

1) Add destination

destination <destinationName> {
      tcp("<host>" port("<tcp-port>"));
};

or,

destination <destinationName> {
      udp("<host>" port("<udp-port>"));
};

where "host" is the container(launcher) host where the syslog module is deployed.

2) Add log rule to log message sources:

log {
  source(<message_source>); destination(<destinationName>);
};

3) Make sure to restart the service after the change:

sudo service syslog-ng restart

Now, the syslog messages from the syslog message sources are written into HDFS /xd/<stream-name>/

Configuration Guidelines

Overview

When running a distributed Spring XD runtime, there are a number of considerations related to performance and reliability. In most cases, these involve settings that have tradeoffs, but in this section we provide some background so you know what the options are and how to configure them.

In the Deployment section that follows, we provide detailed information about various properties that can be passed along with the stream deploy command. That section also describes a scenario that is common for minimizing network hops, where direct binding can occur between modules rather than having each pipe within a stream correspond to a send and receive over the Message Bus. For more detail see the Direct Binding subsection.

Another relevant topic for minimizing network hops is the ability to compose modules. That is a useful technique where a subset of the stream’s contiguous modules can be grouped together as if a single module. All of the pipes within the composed module will rely upon a local transport rather than sending and receiving via the Message Bus. For more detail read the Composing Modules section.

For production use, high availability will typically be a requirement for the data transport. In the Message Bus Configuration section below, we provide details on the relevant HA configuration settings as well as other reliability settings, security settings (including enabling SSL), and error-handling capabilities.

When configuring a RabbitMQ Message Bus, you will also want to consider several performance settings. For example, unless strict sequential ordering is required, the prefetch and concurrency values should be overridden (the default for each is 1). That can lead to a significant performance improvement. In the less likely case that performance concerns completely outweigh reliability, you can disable acknowledgements and even disable the persistence of messages. For a listing of these settings and more, refer to the RabbitMQ Configuration section. Several performance related configuration settings exist on the broker itself, and those are well-documented in the RabbitMQ Admin Guide. For example, the vm_memory_high_watermark and vm_memory_high_watermark_paging_ratio are both explained within the Flow Control subsection of the guide.

If you are using the HTTP source module in a stream and want to scale, you can deploy multiple instances by specifying the module.http.count property as described in the Deployment Properties section. Keep in mind that each instance will share the same port value. The default is 9000, but that can be overridden, for all instances, by including --port as an option for the HTTP module in the stream definition. That means you would want to ensure that each container that may be a candidate for deploying one of the HTTP module instances (taking into account the criteria deployment property if provided), is running on a different host, either physically or on separate virtual machines. Of course, in a production environment, you would likely want to add a load balancer in front of those HTTP endpoints.

Also when using the HTTP source module, you may want to consider enabling support for HTTPS. An example is provided in the documentation for that module’s options.

Deployment

Introduction

This section covers topics related to deployment, including:

When you deploy a Stream or Job, the Spring XD Runtime performs the following steps:

  • parse the stream or job definition (DSL Guide) to resolve each Module reference along with its options

  • set option values assigned to each component Module

  • parse and store the deployment request including the Deployment Manifest

  • allocate each Module to an available Container instance in accordance with the Deployment Manifest

  • binding Module channel(s), either to the MessageBus or directly using Direct Binding

  • track the state of each deployed module

  • track the overall stream or job’s Deployment State

Deployment Manifest

A stream is composed of modules. Each module is deployed to one or more Container instance(s). In this way, stream processing is distributed among multiple containers. By default, deploying a stream to a distributed runtime configuration uses simple round robin logic. For example if there are three containers and three modules in a stream definition, s1= m1 | m2 | m3, then Spring XD will attempt to distribute the work load evenly among each container. This is a very simplistic strategy and does not take into account things like:

  • server load - how many modules are already deployed to a container? What is the current memory and CPU utilization?

  • server affinity - some containers may have external software installed and specific modules will benefit from co-location. For example, an hdfs sink might be deployed only to hosts running Hadoop. Or perhaps a file sink should be deployed to hosts configured with extra disk space.

  • scalability - Suppose the stream s1, above, can achieve higher throughput with multiple instances of m2 running, so we want to deploy m2 to every available container.

  • fault tolerance - the ability to target physical servers on redundant networks, routers, racks, etc.

Generally, more complex deployment strategies are needed to tune and operate XD. Additionally, we must consider various features and constraints when deploying to a PaaS, Yarn or some other cluster manager. Additionally, Spring XD allows supports Stream Partitioning and Direct Binding.

To address such deployment concerns, Spring XD provides a Deployment Manifest which is submitted with the deployment request, in the form of in-line deployment properties (or potentially a reference to a separate document containing deployment properties).

Deployment Properties

When you execute the stream deploy shell command, you can optionally provide a comma delimited list of key=value pairs known as deployment properties. Examples for the key include module.[modulename].count and module.[modulename].criteria (for a full list of properties, see below). The value for the count is a positive integer, and the value for criteria is a valid SpEL expression. The Spring XD runtime matches an available container for each module according to the deployment manifest.

The deployment properties allow you to specify deployment instructions for each module. Currently this includes:

  • The number of module instances

  • A target server or server group

  • MessageBus attributes required for a specific module

  • Stream Partitioning

  • Direct Binding

  • History Tracking

Spring XD Shell interaction

When using the Spring XD Shell, there are two ways to provide deployment properties: either inline or via a file reference. Those two ways are exclusive and documented below:

Inline properties

use the --properties shell option and list properties as a comma separated list of key=value pairs, like so:

stream deploy foo --properties "module.transform.count=2,module.log.criteria=groups.contains('group1')"
Using a file reference

use the --propertiesFile option and point it to a local Java .properties file (i.e. that lives in the filesystem of the machine running the shell). Being read as a .properties file, normal rules apply (ISO 8859-1 encoding, =, <space> or : delimiter, etc.) although we recommend using = as a key-value pair delimiter for consistency:

stream deploy foo --propertiesFile myprops.properties

where myprops.properties contains

# this is a comment
module.transform.count=2
module.log.criteria = groups.contains('group1')

Those two options apply to the stream deploy and job deploy commands.

General Properties
Note
You can apply criteria to all modules in the stream by using the wildcard * for [modulename]
module.[modulename].count

The number of module instances (see above).

module.[modulename].criteria

A boolean SpEL expression using the Container Attributes as an evaluation context.

module.[modulename].trackHistory

A boolean value indicating whether history should be tracked in a message header for this module. Usually used during stream development or for debugging, with module.*.trackHistory=true to track all modules. The xdHistory message header contains an entry for each module that processes the message; each entry includes useful information including the stream name, module label, host, container id, thread name, etc. This enables the determination of exactly how a message was processed through the stream(s).

Example:

xd:>stream deploy --name test1 --properties "module.transform.count=3,module.log.criteria=groups.contains('group1')"
Bus Properties
Common Bus Properties
Note
The following properties are only allowed when using a RabbitMessageBus or a RedisMessageBus; the LocalMessageBus does not support properties.
module.[modulename].consumer.backOffInitialInterval

The number of milliseconds to wait for the first delivery retry (default 1000)

module.[modulename].consumer.backOffMaxInterval

The maximum number of milliseconds to wait between retries (default 10000)

module.[modulename].consumer.backOffMultiplier

The previous retry interval is multiplied by this to determine the current interval (but see backOffMaxInterval) (default 2.0)

module.[modulename].consumer.concurrency

The number of concurrent consumers for the module (default 1).

module.[modulename].consumer.maxAttempts

The maximum number of attempts to make a delivery when a failure occurs (default 3)

RabbitMQ Bus Properties
Note
The following properties are only allowed when using a RabbitMessageBus.

See the Spring AMQP reference documentation for information about the RabbitMQ-specific attributes.

module.[modulename].consumer.ackMode

Controls message acknowledgements (default AUTO)

module.[modulename].consumer.maxConcurrency

The maximum number of concurrent consumers for the module (default 1).

module.[modulename].consumer.prefetch

The number of messages prefetched from the RabbitMQ broker (default 1)

module.[modulename].consumer.prefix

A prefix applied to all queues/exchanges that are declared by the bus - allows policies to be applied (default xdbus.)

module.[modulename].consumer.requestHeaderPatterns

Controls which message headers are passed between modules (default STANDARD_REQUEST_HEADERS,*)

module.[modulename].consumer.replyHeaderPatterns

Controls which message headers are passed between modules (only used in partitioned jobs) (default STANDARD_REPLY_HEADERS,*)

module.[modulename].consumer.requeue

Whether messages will be requeued (and retried) on failure (default true)

module.[modulename].consumer.transacted

Whether consumers use transacted channels (default false)

module.[modulename].consumer.txSize

The number of delivered messages between acknowledgements (when ackMode=AUTO) (default 1)

module.[modulename].consumer.durableSubscription

When true, publish/subscribe named channels (tap:, topic:) will be backed by a durable queue and will be eligible for dead-letter configuration, accoring to the autBindDLQ setting. Note that, since RabbitMQ doesn’t permit queue attributes to be changed, changing the durableSubscription property from true to false between deployments, without first removing the queue, will not have any effect. If a stream is deployed with durableSubscription=true, and you wish to change it to a non-durable subscription, you will need to remove the queue from RabbitMQ before redeploying. Spring XD will create the queue the with the appropriate settings, unless the queue exists already. Changing from a non-durable subscription to a durable subscription will not have this problem because, for a non-durable subscription, the queue will be automatically deleted when the stream is undeployed.

module.[modulename].producer.deliveryMode

The delivery mode of messages sent to RabbitMQ (PERSISTENT or NON_PERSISTENT) (default PERSISTENT)

module.[modulename].producer.requestHeaderPatterns

Controls which message headers are passed between modules (default STANDARD_REQUEST_HEADERS,*)

module.[modulename].producer.replyHeaderPatterns

Controls which message headers are passed between modules (only used in partitioned jobs) (default STANDARD_REPLY_HEADERS,*)

module.[modulename].consumer.autoBindDLQ

When true, the bus will automatically declare dead letter queues and binding for each bus queue. The user is responsible for setting a policy on the broker to enable dead-lettering; see Message Bus Configuration for more information. The bus will configure a dead-letter-exchange (<prefix>DLX) and bind a queue with the name <original queue name>.dlq and route using the original queue name..

module.[modulename].consumer.republishToDLQ

By default, failed messages after retries are exhausted are rejected. If a dead-letter queue (DLQ) is configured, rabbitmq will route the failed message (unchanged) to the DLQ. Setting this property to true instructs the bus to republish failed messages to the DLQ, with additional headers, including the exception message and stack trace from the cause of the final failure. Note that the republish will occur even if maxAttempts is only set to 1. Also see autoBindDLQ (default false)

module.[modulename].producer.batchingEnbled

Batch messages sent to the bus (default false)

module.[modulename].producer.batchSize

The normal batch size, may be preempted by batchBufferLimit or batchTimeout (default 100)

module.[modulename].producer.batchBufferLimit

If a batch will exceed this limit, the batch will be sent prematurely (default 10000)

module.[modulename].producer.batchTimeout

If no messages are received in this time (ms), the batch will be sent (default 5000)

module.[modulename].producer.compress

When true, compress the message before sending to rabbit; (default false) see RabbitMQ Message Bus Properties for information about the compression level

Stream Partitioning
Note
Partitioning is not supported with the local transport.

A common pattern in stream processing is to partition the data as it is streamed. This entails deploying multiple instances of a message consuming module and using content-based routing so that messages containing the identical data value(s) are always routed to the same module instance. You can use the Deployment Manifest to declaratively configure a partitioning strategy to route each message to a specific consumer instance.

Partition Properties

See below for examples of deploying partitioned streams.

module.[modulename].producer.partitionKeyExtractorClass

The class name of a PartitionKeyExtractorStrategy (default null)

module.[modulename].producer.partitionKeyExpression

A SpEL expression, evaluated against the message, to determine the partition key; only applies if partitionKeyExtractorClass is null. If both are null, the module is not partitioned (default null)

module.[modulename].producer.partitionSelectorClass

The class name of a PartitionSelectorStrategy (default null)

module.[modulename].producer.partitionSelectorExpression

A SpEL expression, evaluated against the partition key, to determine the partition index to which the message will be routed. The final partition index will be the return value (an integer) modulo [nextModule].count If both the class and expression are null, the bus’s default PartitionSelectorStrategy will be applied to the key (default null)

In summary, a module is partitioned if its count is > 1 and the previous module has a partitionKeyExtractorClass or partitionKeyExpression (class takes precedence). When a partition key is extracted, the partitioned module instance is determined by invoking the partitionSelectorClass, if present, or the partitionSelectorExpression % partitionCount , where partitionCount is count in the case of Redis and RabbitMQ, and the underlying partition count of the topic in the case of Kafka (see the Message Bus section on Kafka partition configuration for details). If neither a partitionSelectorClass nor a partitionSelectorExpression is present the result is key.hashCode() % partitionCount.

For Redis and Rabbit, the use of partitionKeyExpression and partitionKeyExtractorClass is restricted to sending data modules that have count > 1. Any other use (i.e. sending data to a module with count = 1, or to a named channel) will result in a error at deployment time.

In the case of Kafka, partitionKeyExpression and partitionKeyExtractorClass may be used for sending data to any modules, including the ones with count = 1, as well as to named channels, since partitioning is based on the partition count of the target topic, and not the receiving module count.

Direct Binding

Sometimes it is desirable to allow co-located, contiguous modules to communicate directly, rather than using the configured remote transport, to eliminate network latency. Spring XD creates direct bindings by default only in cases where every "pair" of producer and consumer (modules bound on either side of a pipe) are guaranteed to be co-located.

Currently Spring XD implements no conditional logic to force modules to be co-located. The only way to guarantee that every producer-consumer pair is co-located is to specify that the pair be deployed to every available container instance, in other words, the module counts must be 0. The figure below illustrates this concept. In the first hypothetical case, we deploy one instance (the default)of producer m1, and two instances of the consumer m2. In this case, enabling direct binding would isolate one of the consumer instances. Spring XD will not create direct bindings in this case. The second case guarantees co-location of the pairs and will result in direct binding.

Direct Binding

In addition, direct binding requires that the producer is not configured for partitioning since partitioning is implemented by the Message Bus.

Using module.*.count=0 is the most straightforward way to enable direct binding. Direct binding may be disabled for the stream using module.*.producer.directBindingAllowed=false. Additional direct binding deployment examples are shown below.

Deployment States

The ability to specify criteria to match container instances and deploy multiple instances for each module leads to one of several possible deployment states for the stream as a whole. Consider a stream in an initial undeployed state.

Deploy States

After executing the stream deployment request, the stream will be one of the following states:

  • Deployed - All modules deployed successfully as specified in the deployment manifest.

  • Incomplete - One of the requested module instances could not be deployed, but at least one instance of each module definition was successfully deployed. The stream is operational and can process messages end-to-end but the deployment manifest was not completely satisfied.

  • Failed - At least one of the module definitions was not deployed. The stream is not operational.

Note
The state diagram above represents these states as final. This is an over-simplification since these states are affected by container arrivals and departures that occur during or after the execution of a deployment request. Such transitions have been omitted intentionally but are worth considering. Also, there is an analogous state machine for undeploying a stream, initially in any of these states, which is left as an exercise for the reader.

Example

xd:>stream create test1 --definition "http | transform --expression=payload.toUpperCase() | log"
Created new stream 'test1'

Next, deploy it requesting three transformer instances:

xd:>stream deploy --name test1 --properties "module.transform.count=3"
Deployed stream 'test1'

xd:>stream list
  Stream Name  Stream Definition                                          Status
  -----------  ---------------------------------------------------------  ----------
  test1        http | transform --expression=payload.toUpperCase() | log  incomplete

If there are only two container instances available, only two instances of transform will be deployed. The stream deployment state is incomplete and the stream is functional. However the unfulfilled deployment request remains active and the third instance will be deployed if a new container comes on line that matches the criteria.

Container Attributes

The SpEL context (root object) for module.[modulename].criteria is ContainerAttributes, basically a map derivative that contains some standard attributes:

  • id - the generated container ID

  • pid - the process ID of the container instance

  • host - the host name of the machine running the container instance

  • ip — the IP address of the machine running the container instance

ContainerAttributes also includes any user-defined attribute values configured for the container. These attributes are configured by editing xd/config/servers.yml the file included in the XD distribution contains some commented out sections as examples. In this case, the container attributes configuration looks something like:

xd:
  container:
      groups: group2
      color: red

Groups

Groups may be assigned to a container via the optional command line argument --groups or by setting the environment variable XD_CONTAINER_GROUPS. As the property name suggests, a container may belong to more than one group, represented as comma-delimited string. The concept of server groups is considered an especially useful convention for targeting groups of servers for deployment to support many common scenarios, so it enjoys special status. Internally, groups is simply a user defined attribute.

IP Address

The IP address of the container can also be optionally set via the command argument --containerIp or by setting the environment variable XD_CONTAINER_IP. If not specified, the IP address will be automatically set. Please be aware of the limitations, though, particularly in cases where the physically machine has multiple IP addresses assigned.

For the automatic assignment of the IP address, XD internally loops through the available network interfaces and assigned IP addresses and will pick the first available IPv4 address that is not a loopback address.

Depending on your underlying server or network infrastructure, you may prefer specifying the IP address explicitly.

Hostname

The hostname of the container can be optionally set as well via the command argument --containerHostname or by setting the environment variable XD_CONTAINER_HOSTNAME. If not specified, the hostname will be automatically set. Please be aware of the limitations, though. You may prefer specifying the hostname address explicitly.

Tip
While there is no command line option to set the container hostname and IP address when running in Single Node mode, you can still specify the values via environment variables or by customizing the respective settings in application.yml

Stream Deployment Examples

To Illustrate how to use the Deployment Manifest, We will use a runtime configuration with 3 container instances, as displayed in the XD shell:

xd:>runtime containers
  Container Id                          Host              IP Address     PID   Groups  Custom Attributes
  ------------------------------------  ----------------  -------------  ----  ------  -----------------
  bc624816-f8a8-4f35-83f6-a125ed147b7c  ip-10-110-18-10   10.110.18.10   1708  group2  {color=red}
  018b7c8d-6fa9-4759-8471-76899766f892  ip-10-139-36-168  10.139.36.168  1852  group2  {color=blue}
  afc3741c-217a-415a-9d86-a1f62de03613  ip-10-139-17-116  10.139.17.116  1861  group1  {color=green}

Each of the three containers is running on a different host and has configured Groups and Custom Attributes as shown.

First, create a stream:

xd:>stream create test1 --definition "http | transform --expression=payload.toUpperCase() | log"
Created new stream 'test1'

Next, deploy it using a manifest:

xd:>stream deploy --name test1 --properties "module.transform.count=3,module.log.criteria=groups.contains('group1')"
Deployed stream 'test1'

Verify the deployment:

xd:>runtime modules
  Module                       Container Id                          Options                                         Deployment Properties
  ---------------------------  ------------------------------------  ----------------------------------------------  ---------------------------------------------------------
  test1.processor.transform.1  bc624816-f8a8-4f35-83f6-a125ed147b7c  {valid=true, expression=payload.toUpperCase()}  {count=3, sequence=1}
  test1.processor.transform.2  018b7c8d-6fa9-4759-8471-76899766f892  {valid=true, expression=payload.toUpperCase()}  {count=3, sequence=2}
  test1.processor.transform.3  afc3741c-217a-415a-9d86-a1f62de03613  {valid=true, expression=payload.toUpperCase()}  {count=3, sequence=3}
  test1.sink.log.1             afc3741c-217a-415a-9d86-a1f62de03613  {name=test1, expression=payload, level=INFO}    {count=1, sequence=1, criteria=groups.contains('group1')}
  test1.source.http.1          bc624816-f8a8-4f35-83f6-a125ed147b7c  {port=9000}                                     {count=1, sequence=1}

We can see that three instances of the transform processor have been deployed, one to each container instance. Also the log module has been deployed to the container assigned to group1. Now we can undeploy and deploy the stream using a different manifest:

xd:>stream undeploy test1
Un-deployed stream 'test1'
xd:>runtime modules
  Module  Container Id  Properties
  ------  ------------  ----------

xd:>stream deploy --name test1 --properties "module.log.count=3,module.log.criteria=!groups.contains('group1')"
Deployed stream 'test1'

xd:>stream list
  Stream Name  Stream Definition                                          Status
  -----------  ---------------------------------------------------------  ----------
  test1        http | transform --expression=payload.toUpperCase() | log  incomplete

xd:>runtime modules
  Module                       Container Id                          Options                                         Deployment Properties
  ---------------------------  ------------------------------------  ----------------------------------------------  ----------------------------------------------------------
  test1.processor.transform.1  018b7c8d-6fa9-4759-8471-76899766f892  {valid=true, expression=payload.toUpperCase()}  {count=1, sequence=1}
  test1.sink.log.1             bc624816-f8a8-4f35-83f6-a125ed147b7c  {name=test1, expression=payload, level=INFO}    {count=3, sequence=1, criteria=!groups.contains('group1')}
  test1.sink.log.2             018b7c8d-6fa9-4759-8471-76899766f892  {name=test1, expression=payload, level=INFO}    {count=3, sequence=2, criteria=!groups.contains('group1')}
  test1.source.http.1          afc3741c-217a-415a-9d86-a1f62de03613  {port=9000}                                     {count=1, sequence=1}

Now there are only two instances of the log module deployed. We asked for three however the deployment criteria specifies only containers not in group1 are eligible. The log module is deployed only to the two containers matching the criteria. The deployment status of stream test1 is shown as incomplete. The stream is functional even though the deployment manifest is not completely satisfied. If we fire up a new container not in group1, the DeploymentSupervisor will handle any outstanding deployment requests by comparing xd/deployments/modules/requested to xd/deployments/modules/allocated, and will deploy the third log instance and update the stream state to deployed.

Partitioned Stream Deployment Examples

Using SpEL Expressions

First, create a stream:

xd:>stream create --name partitioned --definition "jms | transform --expression=#expensiveTransformation(payload) | log"

Created new stream 'partitioned'

The hypothetical SpEL function expensiveTransformation represents a resource intensive processor which we want to load balance by running on multiple containers. In this case, we also want to partition the stream so that payloads containing the same customerId are always routed to the same processor instance. Perhaps the processor aggregates data by customerId and this step needs to run using co-located resources.

Next, deploy it using a manifest:

xd:>stream deploy --name partitioned --properties "module.jms.producer.partitionKeyExpression=payload.customerId,module.transform.count=3"

Deployed stream 'partitioned'

In this example three instances of the transformer will be created (with partition index of 0, 1, and 2). When the jms module sends a message it will take the customerId property on the message payload, invoke its hashCode() method and apply the modulo function with the divisor being the transform.count property to determine which instance of the transform will process the message (payload.getCustomerId().hashCode() % 3). Messages with the same customerId will always be processed by the same instance.

Direct Binding Deployment Examples

In the simplest case, we enforce direct binding by setting the instance count to 0 for all modules in the stream. A count of 0 means deploy the module to all available containers:

xd:>runtime containers
  Container Id                          Host            IP Address    PID    Groups  Custom Attributes
  ------------------------------------  --------------  ------------  -----  ------  -----------------
  8e814924-15de-4ca1-82d3-ddfe851668ab  ultrafox.local  192.168.1.18  81532
  a2b89274-2d40-46e4-afc5-4988bea28a16  ultrafox.local  192.168.1.9   4605   group1

We start with two container instances. One belongs to the group group1.

xd:>stream create direct --definition "time | log"
Created new stream 'direct'
xd:>stream deploy direct --properties module.*.count=0
Deployed stream 'direct'
xd:>runtime modules
  Module                Container Id                          Options                                        Deployment Properties
  --------------------  ------------------------------------  ---------------------------------------------  ---------------------------------------------------------
  direct.sink.log.0     a2b89274-2d40-46e4-afc5-4988bea28a16  {name=direct, expression=payload, level=INFO}  {count=0, sequence=0}
  direct.sink.log.0     8e814924-15de-4ca1-82d3-ddfe851668ab  {name=direct, expression=payload, level=INFO}  {count=0, sequence=0}
  direct.source.time.0  a2b89274-2d40-46e4-afc5-4988bea28a16  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}     {producer.directBindingAllowed=true, count=0, sequence=0}
  direct.source.time.0  8e814924-15de-4ca1-82d3-ddfe851668ab  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}     {producer.directBindingAllowed=true, count=0, sequence=0}

Note that we have two containers and two instances of each module deployed to each. Spring XD automatically sets the bus properties needed to allow direct binding, producer.directBindingAllowed=true on the time module.

Suppose we only want one instance of this stream and we want it to use direct binding. Here we can add deployment criteria to restrict the available containers to group1.

xd:>stream undeploy direct
Un-deployed stream 'direct'
xd:>stream deploy direct --properties "module.*.count=0, module.*.criteria=groups.contains('group1')"
Deployed stream 'direct'
xd:>runtime modules
  Module                Container Id                          Options                                        Deployment Properties
  --------------------  ------------------------------------  ---------------------------------------------  ---------------------------------------------------------------------------------------------
  direct.sink.log.0     a2b89274-2d40-46e4-afc5-4988bea28a16  {name=direct, expression=payload, level=INFO}  {count=0, sequence=0, criteria=groups.contains('group1')}
  direct.source.time.0  a2b89274-2d40-46e4-afc5-4988bea28a16  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}     {producer.directBindingAllowed=true, count=0, sequence=0, criteria=groups.contains('group1')}

Direct binding eliminates latency between modules but sacrifices some of the resiliency provided by the messaging middleware. In the scenario above, if we lose one of the containers, we lose messages. To disable direct binding when module counts are set to 0, set module.*.producer.directBindingAllowed=false.

xd:>stream undeploy direct
Un-deployed stream 'direct'
xd:>stream deploy direct --properties "module.*.count=0, module.*.producer.directBindingAllowed=false"
Deployed stream 'direct'
xd:>runtime modules
  Module                Container Id                          Options                                        Deployment Properties
  --------------------  ------------------------------------  ---------------------------------------------  ----------------------------------------------------------
  direct.sink.log.0     a2b89274-2d40-46e4-afc5-4988bea28a16  {name=direct, expression=payload, level=INFO}  {producer.directBindingAllowed=false, count=0, sequence=0}
  direct.sink.log.0     8e814924-15de-4ca1-82d3-ddfe851668ab  {name=direct, expression=payload, level=INFO}  {producer.directBindingAllowed=false, count=0, sequence=0}
  direct.source.time.0  a2b89274-2d40-46e4-afc5-4988bea28a16  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}     {producer.directBindingAllowed=false, count=0, sequence=0}
  direct.source.time.0  8e814924-15de-4ca1-82d3-ddfe851668ab  {fixedDelay=1, format=yyyy-MM-dd HH:mm:ss}     {producer.directBindingAllowed=false, count=0, sequence=0}

Finally, we can still have the best of both worlds by enabling guaranteed delivery at one point in the stream, usually the source. If the tail of the stream is co-located and the source uses the message bus, the message bus may be configured so that if a container instance goes down, any unacknowledged messages will be retried until the container comes back or its modules are redeployed.

TDB: A realistic example

An alternate scenario with similar characteristics would be if the stream uses a rabbit or jms source. In this case, guaranteed delivery would be configured in the external messaging system instead of the Spring XD transport.

Troubleshooting

Debugging a distributed system to diagnose problems can be challenging. While using Spring XD, if you encounter

ZooKeeper disconnects

Problem: Spring XD processes disconnecting from ZooKeeper

Recommendation: Depending on your setup, modify either xd-singlenode or xd-container scripts by setting the environment variable export JAVA_OPTS=-verbose:gc before launching them.

Reason: ZooKeeper requires a heartbeat at a regular interval to test liveness of connected processes. Full "stop the world" GCs can result in connection and session timeouts from ZooKeeper. While verbose, GC logs are helpful for diagnosing this and other performance issues.

Debugging Slowness

Problem: Slow or unresponsive application

Recommendation: Capture multiple thread dumps several seconds apart using jstack.

Reason: Examination of thread dumps can reveal stuck or slow moving threads. This data is useful for determining the root cause of a slow or unresponsive application.

File Descriptors and limit violation

Problem: java.io.FileNotFoundException: (Too many open files)

Recommendation: Default ulimit setting in most UNIX based operating systems is 1024. Raise ulimit setting to at least 10000.

Reason: Stream and job modules in Spring XD are loaded and unloaded dynamically on demand. When a module is unloaded, the associated class loaders may not be garbage collected right away, resulting in open file handles for the jar files used by the module. Depending on the number of modules in use, the file handle limit of 1024 may be exceeded.

Message Bus Configuration

Introduction

This section contains additional information about configuring the Message Bus, including High Availability, SSL, Error handling and partitioning.

Rabbit Message Bus High Availability (HA) Configuration

Introduction

The RabbitMessageBus allows for HA configuration using normal RabbitMQ HA Configuration.

First, use the addresses property in servers.yml to include the host/port for each server in the cluster. See Application Configuration.

By default, queues and exchanges declared by the bus are prefixed with xdbus. (this prefix can be changed as described in Application Configuration).

To configure the entire bus for HA, create a policy:

rabbitmqctl set_policy ha-xdbus "^xdbus\." '{"ha-mode":"all"}'

Connection Management and HA Queues

When consuming from HA queues, there might be some performance advantage in consuming from the node that actually hosts the queue. Starting with version 1.2 it is now possible to configure the Rabbit Message Bus to do that.

Caution
To utilize this mechanism, the rabbit management plugin must be enabled on each node in the cluster. The plugin’s REST API is used to determine the location of the queue.

This feature is enabled by adding more than one node to the spring.rabbitmq.node property. See RabbitMQ Configuration for configuration details.

When a node fails and a queue is moved to one of the mirrors, the bus will automatically reconnect to the right node.

Error Handling (Message Delivery Failures)

RabbitMQ Message Bus

Note
The following applies to normally deployed streams. When direct binding between modules is being used, exceptions thrown by the consumer are thrown back to the producer.

When a consuming module (processor, sink) fails to handle a message, the bus will retry delivery based on the module (or default bus) retry configuration. The default configuration will make 3 attempts to deliver the message. The retry configuration can be modified at the bus level (in servers.yml), or for an individual stream/module using the deployment manifest.

When retries are exhausted, by default, messages are discarded. However, using RabbitMQ, you can configure such messages to be routed to a dead-letter exchange/dead letter queue. See the RabbitMQ Documentation for more information.

Note
The following configuration examples assume you are using the default bus prefix used for naming rabbit elements: "xdbus."

Consider a stream: stream create foo --definition "source | processor | sink"

The first pipe (by default) will be backed by a queue named xdbus.foo.0, the second by xdbus.foo.1. Messages are routed to these queues using the default exchange (with routing keys equal to the queue names).

To enable dead lettering just for this stream, first configure a policy:

rabbitmqctl set_policy foo.DLX "^xdbus\.foo\..*" '{"dead-letter-exchange":"foo.dlx"}' --apply-to queues

To configure dead-lettering for all streams:

rabbitmqctl set_policy DLX "^xdbus\..*" '{"dead-letter-exchange":"dlx"}' --apply-to queues

The next step is to declare the dead letter exchange, and bind dead letter queues with the appropriate routing keys.

For example, for the second "pipe" in the stream above we might bind a queue foo.sink.dlq to exchange foo.dlx with a routing key xdbus.foo.1 (remember, the original routing key was the queue name).

Now, when the sink fails to handle a message, after the configured retries are exhausted, the failed message will be routed to foo.sink.dlq.

There is no automated mechanism provided to move dead lettered messages back to the bus queue.

Automatic Dead Lettering Queue Binding

Starting with version 1.1, the dead letter queue and binding can be automatically configured by the system. A new property autoBindDLQ has been added; it can be set at the bus level (in servers.yml) or using deployment properties, e.g. --properties module.*.consumer.autoBindDLQ=true for all modules in the stream. When true, the dead letter queue will be declared (if necessary) and bound to a dead letter exchange named xdbus.DLX (again, assuming the default prefix) using the queue name as the routing key.

In the above example, where we have queues xdbus.foo.0 and xdbus.foo.1, the system will also create xdbus.foo.0.dlq, bound to xdbus.DLX with routing key xdbus.foo.0 and xdbus.foo.1.dlq, bound to xdbus.DLX with routing key xdbus.foo.1.

Note
Starting with version 1.2, any queues that are deployed with autoBindDLQ will automatically be configured to enable dead-lettering, routing to the DLX with the proper routing key. It is no longer necessary to use a policy to set up dead-lettering when using autoBindDLQ.

Also, starting with version 1.2, the provision of dead-lettering on publish/subscribe named channels (tap: or topic:) depends on a new deployment property durable. This property is similar to a JMS durable subscription to a topic and is false by default. When false (default), the queue backing such a named channel is declared auto-delete and is removed when the stream is undeployed. A DLQ will not be created for such queues. When true, the queue becomes permanent (durable) and is not removed when the stream is undeployed. Also, when true, the queue is eligible for DLQ provisioning, according to the autoBindDLQ deployment property. durable can be set at the bus level, or in an individual deployment property, such as:

stream create ticktock --definition="time --fixedDelay=5 | log" --deploy stream create tttap --definition="tap:stream:ticktock > log" stream deploy tttap --properties=module.log.consumer.durableSubscription=true

Redis Message Bus

When Redis is the transport, the failed messages (after retries are exhausted) are LPUSH+ed to a +LIST ERRORS:<stream>.n (e.g. ERRORS:foo.1 in the above example in the RabbitMQ Message Bus section).

This is unconditional; the data in the ERRORS LIST is in "bus" format; again, as with the RabbitMQ Message Bus, some external mechanism would be needed to move the data from the ERRORS LIST back to the bus’s foo.1 LIST.

Note
When moving errored messages back to the main stream, it is important to understand that these messages contain binary data and are unlikely to survive conversion to and from Unicode (such as with Java String variables). If you use Java to move these messages, we recommend that you use a RedisTemplate configured as follows:
<bean id="redisTemplate"
         class="org.springframework.data.redis.core.RedisTemplate">
  <property name="connectionFactory" ref="jedisConnectionFactory" />
  <property name="keySerializer">
    <bean
      class="org.springframework.data.redis.serializer.StringRedisSerializer" />
  </property>
  <property name="enableDefaultSerializer" value="false" />
</bean>

or

@Bean
public RedisTemplate<String, byte[]> redisTemplate() {
    RedisTemplate<String, byte[]> template = new RedisTemplate<String, byte[]>();
    template.setConnectionFactory(connectionFactory());
    template.setKeySerializer(new StringRedisSerializer());
    template.setEnableDefaultSerializer(false);
    return template;
}

This enables the message payload to be retained as byte[] with no conversion; you would then use something like…​

byte[] errorEvt = redisTemplate.opsForList().rightPop(errorQueue);
redisTemplate.opsForList().leftPush(destinationQueue, errorEvt);

If, after moving a message, you see an error such as:

redis.RedisMessageBus$ReceivingHandler - Could not convert message: EFBFBD...

This is a sure sign that a UTF-8 → Unicode → UTF-8 conversion was performed on the message.

Rabbit Message Bus Secure Sockets Layer (SSL)

If you wish to use SSL for communications with the RabbitMQ server, consult the RabbitMQ SSL Support Documentation.

First configure the broker as described there. The message bus is a client of the broker and supports both of the described configurations for connecting clients (SSL without certificate validation and with certficate validation).

To use SSL without certificate validation, simply set

spring:
  rabbitmq:
    useSSL: true

In servers.yml (and set the port(s) in the addresses property appropriately).

To use SSL with certificate validation, set

spring:
  rabbitmq:
    useSSL: true
    sslProperties: file:path/to/secret/ssl.properties
    #ssl:
      #   keyStore:
      #   keyStorePassphrase:
      #   trustStore:
      #   trustStorePassphrase:

The sslProperties property is a Spring resource (file:, classpath: etc) that points to a properties file, Typically, this file would be secured by the operating system (and readable by the XD container) because it contains security information. Specifically:

keyStore=file:/secret/client/keycert.p12
trustStore=file:/secret/trustStore
keyStore.passPhrase=secret
trustStore.passPhrase=secret

Where the pkcs12 keystore contains the client certificate and the truststore contains the server’s certificate as described in the rabbit documentation. The key/trust store properties are Spring resources.

Alternately, you may specify these properties in-line in lieu of using an external properties file. In this case,the passphrase properties may be encrypted. See Encrypted Properties for more details. When both sslProperties and in-line ssl properties are configured,the in-line properties take precedence.

Note
By default, the rabbit source and sink modules inherit their default configuration from the container, but it can be overridden, either using modules.yml or with specific module definitions.

Rabbit Message Bus Batching and Compression

See RabbitMQ Message Bus Properties for information about batching and compressing messages passing through the bus.

Removing RabbitMQ MessageBus Resources

When a stream or job is undeployed, the broker resources (queues, exchanges) are NOT removed from RabbitMQ. This is due to the possibility that a stream might be being undeployed temporarily, and avoids message loss.

If you wish to completely remove these resources, a REST API is provided for this purpose. In addition, the SpringXDTemplate provides a Java binding for this REST api via its streamOperations().cleanBusResources(String name) and jobOperations().cleanBusResources(String name) APIs.

Or, you can use the REST API directly; for example:

curl 'http://localhost:9393/streams/clean/rabbit/foo\
     ?user=guest&pw=guest&vhost=/&busPrefix=xdbus.&adminUri=http://localhost:15672'

curl 'http://localhost:9393/jobs/clean/rabbit/bar\
    ?user=guest&pw=guest&vhost=/&busPrefix=xdbus.&adminUri=http://localhost:15672'

Where foo is the stream name and bar is the job name.

The stream or job name supports a simple wildcard syntax; if it ends with *, then all streams beginning with the name (excluding the *) will be cleaned.

These operations remove the inter-module stream queues, any tap exchanges created for the stream, the job queue and request queue for partitioned jobs.

The operation will fail if any queue currently has a consumer; similarly, the operation will fail if any exchange has a binding. Under either condition, no changes will be made to RabbitMQ.

The following query params are supported:

  • adminUri - location of the RabbitMQ Admin (default http://localhost:15672)

  • user - admin user (default guest)

  • pw - admin password (default guest)

  • vhost - the vhost used for the bus resources (default /)

  • busPrefix - the prefix used for all bus resources (default xdbus.)

Kafka Message Bus Partition Control

This section describes how topic partitioning functions when using Kafka as transport.

Controlling the partition count of a transport topic

The KafkaMessageBus will attempt to set the number of partitions in a transport topic as consumerCount * consumerConcurrency, either by creating the topic with the required number of partitions, or by repartitioning it, in case it exists.

For example, let’s consider a stream with the following definition:

stream create ingest --definition="http | hdfs"

A default deployment will result in the creation of a single topic with a single partition.

stream deploy ingest

A deployment (or redeployment) of the same stream with a different module count and concurrency will result in 6 partitions, evenly distributed across the 3 module instances:

stream deploy ingest --properties module.hdfs.count=3,module.hdfs.concurrency=2

Besides relying on defaults, you can customize the number of Kafka partitions used by transport topics by indicating a minimum value to be used by deployments (by minimum, it is understood that, if smaller than consumerCount * consumerConcurrency, the latter value will be used instead).

This can be done globally by changing the xd.messagebus.kafka.default.minPartitionCount property in servers.yml:

xd:
  messagebus:
    kafka:
      default:
        minPartitionCount:   5

This will result in creating at least 5 partitions for each transport topic.

Alternatively, and for more granular control, the property can be specified for specific deployments and modules, through the producer.minPartitionCount property in the deployment manifest, as in the following example, where 10 partitions will be created:

stream deploy ingest --properties module.http.producer.minPartitionCount=10,module.hdfs.count=3

Overpartitioning can serve a number of purposes, such as load balancing and distributing data among brokers, as well as allowing for scaling up by increasing the number of concurrent consumers in the future.

Note
If the Kafka topic already exists and it already has a number of partitions larger than either minPartitionCount or consumerCount * consumerConcurrency, its partition count will remain unchanged, and the Kafka transport will operate with all the existing partitions.

Administration

Monitoring and Management

Spring XD uses Spring Boot’s monitoring and management support over HTTP and JMX along with Spring Integration’s MBean Exporters

Monitoring XD Admin, Container and Single-node servers

JMX is disabled by default. To enable JMX, set XD_JMX_ENABLED=true. JMX is disabled by default due to performance issues when message rates are over 100K (for ~100 byte messages). Peformance related issue will be addressed in a future release.

Spring integration components are exposed over JMX using IntegrationMBeanExporter

Once JMX is enabled, all the available MBeans can be accessed over HTTP using Jolokia.

If you want to disable Jolokia endpoints but still want to use JMX, then you can set this property in config/servers.yml:

endpoints:
  jolokia:
    enabled: false

To enable boot provided management endpoints over HTTP

The Spring Boot management endpoints are exposed over HTTP.

When starting admin, container or singlenode server, the command-line option --mgmtPort can be specified to use an explicit port for management server. With the given valid management port, the management endpoints can be accessed from that port. Please refer Spring Boot document here for more details on the endpoints.

For instance, once XD admin is started on localhost and the management port set to use the admin port (9393)

http://localhost:9393/management/health
http://localhost:9393/management/env
http://localhost:9393/management/beans

etc..

To enable the container shutdown operation in the UI

Add the following configuration to config/servers.yml. This configuration is available as a commented section in config/servers.yml.

---
spring:
  profiles: container
management:
  port: 0

To disable boot endpoints over HTTP

Set management.port=-1 for both default and container profiles in config/servers.yml

Management over JMX

All the boot endpoints are exposed over JMX with the domain name org.springframework.boot The MBeans that are exposed within XD admin, container server level are available with the domain names xd.admin (for XD admin), xd.container (for XD container), xd.shared.server and xd.parent representing the application contexts common to both XD admin and container. Singlenode server will have all these domain names exposed. When the stream/job gets deployed into the XD container, the stream/job MBeans are exposed with specific domain/object naming strategy.

Monitoring deployed modules in XD container

When a module is deployed (with JMX is enabled on the XD container), the IntegrationMBeanExporter is injected into module’s context via MBeanExportingPlugin and this exposes all the spring integration components inside the module. For the given module, the IntegrationMBeanExporter uses a specific object naming strategy that assigns domain name as xd.<stream/job name> and, object name as <module name>.<module index>.

Streams

For a stream name mystream with DSL http | log will have

MBeans with domain name xd.mystream with two objects http.0 and log.1

Source, processor, and sink modules will generally have the following attributes and operations

Module Type Attributes and Operations

Source

MessageSourceMetrics

Processor,Sink

MessageHandlerMetrics

In addition, each module has channel attributes and operations defined by MessageChannelMetrics.

Jobs

For a job name myjob with DSL jdbchdfs will have

MBeans with domain name xd.myjob with an object jdbchdfs.0

You can also obtain monitoring information for Jobs using the UI or accessing the Job management REST API. Documentation for the Job Management REST API is forthcoming, but until then please reference the request mappings in BatchJobsController, BatchJobExecutionsController, BatchStepExecutionsController, and BatchJobInstancesController.

Using Jolokia to access JMX over http

When JMX is enabled (after setting XD_JMX_ENABLED to true), Jolokia is auto-configured to expose the XD admin, container and singenode server MBeans.

For example, with XD singlenode running management port 9080

http://localhost:9393/management/jolokia/search/xd*:*,component=MessageChannel

will list all the MessageChannel MBeans exposed in XD container. Apart from this, other available domain and types can be accessed via Jolokia endpoints.

REST API

Introduction

The Spring XD Administrator process (Admin) provides a REST API to access various Spring XD resources such as streams, jobs, metrics, modules, Spring batch resources, and container runtime information. The REST API is used internally by the XD Shell and Admin UI and can support any custom client application that requires interaction with XD.

The HTTP port is configurable and may be set as a command line argument when starting the Admin server, or set in $XD_HOME/config/servers.yml. The default port is 9393:

> $XD_HOME/bin/xd-admin --httpPort <port>

The Admin server also exposes runtime management resources enabled by Spring Boot under the /management context path, e.g., http://localhost:9393/management/metrics. These resources are covered in the Spring Boot documentation.

Note
There is also a mgmtPort command line argument which assigns a separate port for management services. Normally the same port is used for everything.

XD Resources

Table 1. Table XD REST endpoints

stream definitions

/streams/definitions

stream deployments

/streams/deployments

job definitions

/jobs/definitions

job deployments

/jobs/deployments

batch job configurations

/jobs/configurations

batch job executions

/jobs/executions

batch job instances

/jobs/instances

module definitions

/modules

deployed modules

/runtime/modules

containers

/runtime/containers

counters

/metrics/counters

field value counters

/metrics/field-value-counters

aggregate counters

/metrics/aggregate-counters

gauges

/metrics/gauges

rich-gauges

/metrics/rich-gauges

completions

/completions

Stream Definitions

Table 2. Table Stream Definitions
Resource URL Request Method Description

/streams/definitions

GET

list defined streams along with deployment state

/streams/definitions

DELETE

delete all stream definitions, undeploying deployed streams

/streams/definitions/{name}

GET

get a stream definition (currently no deployment information is included)

/streams/definitions

POST

create a new stream, optionally deploying it if deploy=true(default). The request body is application/x-www-form-urlencoded and requires two parameters, name and definition (DSL)

/streams/definitions/{name}

DELETE

delete a stream, undeploying if deployed.

Stream Deployments

Table 3. Table Stream Deployments
Resource URL Request Method Description

/streams/deployments/

GET

get detailed deployment state for all streams (TBD)

/streams/deployments

DELETE

undeploy all streams

/streams/deployments/{name}

GET

get detailed deployment state for a stream (TBD)

/streams/deployments/{name}

POST

deploy a stream, where the request body contains the deployment properties application/x-www-form-urlencoded

/streams/deployments/{name}

DELETE

undeploy a stream

Job Definitions

Table 4. Table Job Definitions
Resource URL Request Method Description

/jobs/definitions

GET

list defined jobs along with deployment state

/jobs/definitions

DELETE

delete all job definitions, undeploying deployed jobs

/jobs/definitions/{name}

GET

get a job definition

/jobs/definitions

POST

create a new job, where the The request body is application/x-www-form-urlencoded and requires two parameters, name and definition (DSL)

/jobs/definitions/{name}

DELETE

delete a job, undeploying if deployed

Job Deployments

Table 5. Table Job Deployments
Resource URL Request Method Description

/jobs/deployments/

GET

get detailed deployment state for all jobs (TBD)

/jobs/deployments

DELETE

undeploy all jobs

/jobs/deployments/{name}

GET

get detailed deployment state for a job (TBD. Probably not in 1.0)

/jobs/deployments/{name}

POST

deploy a job, where the request body contains the deployment properties

/jobs/deployments/{name}

DELETE

undeploy a job

Batch Job Configurations

Spring Batch configured jobs stored in the Spring Batch Repository

Table 6. Table Batch Jobs
Resource URL Request Method Description

/jobs/configurations

GET

get configuration information about all batch jobs

/jobs/configurations/{jobName}

GET

get configuration information about a batch job

Batch Job Executions

Table 7. Table Batch Executions
Resource URL Request Method Description

/jobs/executions

GET

list all job executions (Only application/json mediatype supported)

/jobs/executions?stop=true

PUT

stop all jobs

/jobs/executions?jobname={jobName}

GET

get information about all executions of a job (Only application/json accept header is supported)

/jobs/executions?jobname={jobName}

POST

request the launch of a job

/jobs/executions/{jobExecutionId}

GET

get information about a particular execution of a job

/jobs/executions/{jobExecutionId}?restart=true

PUT

restart a job

/jobs/executions/{jobExecutionId}?stop=true

PUT

stop a job

/jobs/executions/{jobExecutionId}/steps

GET

list the steps for a job execution (Only application/json accept header is supported)

/jobs/executions/{jobExecutionId}/steps/{stepExecutionId}

GET

get a step execution

/jobs/executions/{jobExecutionId}/steps/{stepExecutionId}/progress

GET

get the step execution progress