Version 1.6.0.RELEASE

© 2012-2018 Pivotal Software, Inc.

Copies of this document may be made for your own use and for distribution to others, provided that you do not charge any fee for such copies and further provided that each copy contains this Copyright Notice, whether distributed in print or electronically.

Getting Started

Spring Cloud Data Flow is a toolkit for building data integration and real-time data processing pipelines.

Pipelines consist of Spring Boot apps, built using the Spring Cloud Stream or Spring Cloud Task microservice frameworks. This makes Spring Cloud Data Flow suitable for a range of data processing use cases, from import/export to event streaming and predictive analytics.

This project provides support for using Spring Cloud Data Flow with Kubernetes as the runtime for these pipelines with apps packaged as Docker images.

1. Installation

In this section we will install the Spring Cloud Data Flow Server on a Kubernetes cluster. Spring Cloud Data Flow depends on a few services and their availability. For example, we need an RDBMS service for the app registry, stream/task repositories and task management. For streaming pipelines, we also need a messaging middleware option such as Apache Kafka or RabbitMQ. In addition to this, we need a Redis service if the analytics features are in use.

This guide describes setting up an environment for testing Spring Cloud Data Flow on Google Kubernetes Engine and is not meant to be a definitive guide for setting up a production environment. Feel free to adjust the suggestions to fit your test set-up. Please remember that a production environment requires much more consideration for persistent storage of message queues, high availability, security etc.

Currently, only apps registered with a --uri property pointing to a Docker resource are supported by the Data Flow Server for Kubernetes.

Note that we do support Maven resources for the --metadata-uri property.

E.g. the below app registration is valid:

dataflow:>app register --type source --name time --uri docker://springcloudstream/time-source-rabbit:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:time-source-rabbit:jar:metadata:1.3.1.RELEASE

but any app registered with a Maven, HTTP or File resource for the executable jar (using a --uri property prefixed with maven://, http:// or file://) is not supported.

1.1. Kubernetes Compatibility

The Spring Cloud Data Flow implementation for Kubernetes uses Spring Cloud Deployer Kubernetes library for orchestration. Before you begin setting up Kubermetes cluster, refer to the compatibility-matrix to learn more about the deployer/server compatibility against Kubernetes release versions.

1.2. Create a Kubernetes cluster

The Kubernetes Picking the Right Solution guide lets you choose among many options so you can pick one that you are most comfortable using.

All our testing is done using the Google Kubernetes Engine that is part of the Google Cloud Platform. That is a also the target platform for this section. We have also successfully deployed using Minikube and we will note where you need to adjust for deploying on Minikube.

When starting Minikube you should allocate some extra resources since we will be deploying several services. We have used minikube start --cpus=4 --memory=4096 to start.

The rest of this getting started guide assumes that you have a working Kubernetes cluster and a kubectl command line utility. See the docs for installation instructions: Installing and Setting up kubectl.

1.3. Deploying using kubectl

  1. Get the Kubernetes configuration files.

    There are sample deployment and service YAML files in the https://github.com/spring-cloud/spring-cloud-dataflow-server-kubernetes repository that you can use as a starting point. They have the required metadata set for service discovery by the different apps and services deployed. To check out the code enter the following commands:

    $ git clone https://github.com/spring-cloud/spring-cloud-dataflow-server-kubernetes
    $ cd spring-cloud-dataflow-server-kubernetes
    $ git checkout v1.6.0.RELEASE

1.3.1. Choosing a Message Broker

For deployed applications to communicate with each other, a message broker needs to be selected. The sample deployment and service YAML files provide configurations for RabbitMQ and Kafka. Only one message broker needs to be configured.

  1. RabbitMQ

    Run the following command to start the RabbitMQ service:

    $ kubectl create -f src/kubernetes/rabbitmq/

    You can use the command kubectl get all -l app=rabbitmq to verify that the deployment, pod and service resources are running. Use the command kubectl delete all -l app=rabbitmq to clean up afterwards.

  2. Kafka

    Run the following command to start the Kafka service:

    $ kubectl create -f src/kubernetes/kafka/

    You can use the command kubectl get all -l app=kafka to verify that the deployment, pod and service resources are running. Use the command kubectl delete all -l app=kafka to clean up afterwards.

1.3.2. Deploy Services and DataFlow

  1. Deploy MySQL.

    We are using MySQL for this guide, but you could use Postgres or H2 database instead. We include JDBC drivers for all three of these databases, you would just have to adjust the database URL and driver class name settings.

    You can modify the password in the src/kubernetes/mysql/mysql-deployment.yaml files if you prefer to be more secure. If you do modify the password you will also have to provide it base64 encoded in the src/kubernetes/mysql/mysql-secrets.yaml file.

    Run the following commands to start the MySQL service:

    $ kubectl create -f src/kubernetes/mysql/

    You can use the command kubectl get all -l app=mysql to verify that the deployment, pod and service resources are running. Use the command kubectl delete all,pvc,secrets -l app=mysql to clean up afterwards.

  2. Deploy Redis.

    The Redis service will be used for the analytics functionality. Run the following commands to start the Redis service:

    $ kubectl create -f src/kubernetes/redis/

    If you don’t need the analytics functionality you can turn this feature off by changing SPRING_CLOUD_DATAFLOW_FEATURES_ANALYTICS_ENABLED to false in the src/kubernetes/server/server-deployment.yaml file. If you don’t install the Redis service then you should also remove the Redis configuration settings from the following files depending on your message broker of choice:

    • If using RabbitMQ: src/kubernetes/server/server-config-rabbit.yaml

    • If using Kafka: src/kubernetes/server/server-config-kafka.yaml

    You can use the command kubectl get all -l app=redis to verify that the deployment, pod and service resources are running. Use the command kubectl delete all -l app=redis to clean up afterwards.

  3. Deploy the Metrics Collector.

    The Metrics Collector will provide message rates for all deployed stream apps. These message rates will be visible in the Dashboard UI. Run the following commands to start the Metrics Collector:

    • If using RabbitMQ:

      $ kubectl create -f src/kubernetes/metrics/metrics-deployment-rabbit.yaml
    • If using Kafka:

      $ kubectl create -f src/kubernetes/metrics/metrics-deployment-kafka.yaml

      Create the metrics service:

      $ kubectl create -f src/kubernetes/metrics/metrics-svc.yaml

      You can use the command kubectl get all -l app=metrics to verify that the deployment, pod and service resources are running. Use the command kubectl delete all -l app=metrics to clean up afterwards.

  4. Deploy Skipper

    Optionally, you can deploy Skipper to leverage the features of upgrading and rolling back Streams since Data Flow delegates to Skipper for those features. For more details, review Spring Cloud Skipper’s reference guide for a complete overview and its feature capabilities.

    You should specify the version Skipper that you want to deploy.

    The deployment is defined in the src/kubernetes/skipper/skipper-deployment.yaml file. To control what version of Skipper that gets deployed you should modify the tag used for the Docker image in the container spec:

        spec:
          containers:
          - name: skipper
            image: springcloud/spring-cloud-skipper-server:1.0.8.RELEASE   (1)
            imagePullPolicy: Always
    1 You may change the version as you like.
    Skipper includes the concept of platforms, so it is important to define the "accounts" based on the project preferences. In the above YAML file, the accounts map to minikube as the platform. This can be modified, and of course, you can have any number of platform definitions. More details are in Spring Cloud Skipper reference guide.

    If you’d like to orchestrate stream processing pipelines with Apache Kafka as the messaging middleware using Skipper, you must change the SPRING_APPLICATION_JSON environment variable value in the src/kubernetes/skipper/skipper-deployment.yaml file to:

    "{\"spring.cloud.skipper.server.enableLocalPlatform\" : false, \"spring.cloud.skipper.server.platform.kubernetes.accounts.minikube.environmentVariables\": \"SPRING_CLOUD_STREAM_KAFKA_BINDER_BROKERS=${KAFKA_SERVICE_HOST}:${KAFKA_SERVICE_PORT}, SPRING_CLOUD_STREAM_KAFKA_BINDER_ZK_NODES=${KAFKA_ZK_SERVICE_HOST}:${KAFKA_ZK_SERVICE_PORT}\",\"spring.cloud.skipper.server.platform.kubernetes.accounts.minikube.memory\" : \"1024Mi\"}"

    Additionally, if you would like to use the Apache Kafka Streams Binder, the SPRING_APPLICATION_JSON environment variable in src/kubernetes/skipper/skipper-deployment.yaml would be configured as follows:

    "{\"spring.cloud.skipper.server.enableLocalPlatform\" : false, \"spring.cloud.skipper.server.platform.kubernetes.accounts.minikube.environmentVariables\": \"SPRING_CLOUD_STREAM_KAFKA_BINDER_BROKERS=${KAFKA_SERVICE_HOST}:${KAFKA_SERVICE_PORT}, SPRING_CLOUD_STREAM_KAFKA_BINDER_ZK_NODES=${KAFKA_ZK_SERVICE_HOST}:${KAFKA_ZK_SERVICE_PORT}, SPRING_CLOUD_STREAM_KAFKA_STREAMS_BINDER_BROKERS=${KAFKA_SERVICE_HOST}:${KAFKA_SERVICE_PORT}, SPRING_CLOUD_STREAM_KAFKA_STREAMS_BINDER_ZK_NODES=${KAFKA_ZK_SERVICE_HOST}:${KAFKA_ZK_SERVICE_PORT}\",\"spring.cloud.skipper.server.platform.kubernetes.accounts.minikube.memory\" : \"1024Mi\"}"

    Run the following commands to start Skipper as the companion server for Spring Cloud Data Flow:

    $ kubectl create -f src/kubernetes/skipper/skipper-deployment.yaml
    $ kubectl create -f src/kubernetes/skipper/skipper-svc.yaml

    You can use the command kubectl get all -l app=skipper to verify that the deployment, pod and service resources are running. Use the command kubectl delete all -l app=skipper to clean up afterwards.

  5. Deploy the Data Flow Server.

    You should specify the version of the Spring Cloud Data Flow server that you want to deploy.

    The deployment is defined in the src/kubernetes/server/server-deployment.yaml file. To control what version of the Spring Cloud Data Flow server that gets deployed you should modify the tag used for the Docker image in the container spec:

        spec:
          containers:
          - name: scdf-server
            image: springcloud/spring-cloud-dataflow-server-kubernetes:1.6.0.RELEASE   (1)
            imagePullPolicy: Always
    1 Change the version as you like. This document is based on the 1.6.0.RELEASE release. The docker tag latest can be used for BUILD-SNAPSHOT releases.

    To use Skipper, you must uncomment the following properties to src/kubernetes/server/server-deployment.yaml. under the env: section

     - name: SPRING_CLOUD_SKIPPER_CLIENT_SERVER_URI
       value: 'http://${SKIPPER_SERVICE_HOST}/api'
     - name: SPRING_CLOUD_DATAFLOW_FEATURES_SKIPPER_ENABLED
       value: 'true'

    The Data Flow Server uses the Fabric8 Java client library to connect to the Kubernetes cluster. We are using environment variables to set the values needed when deploying the Data Flow server to Kubernetes. We are also using the Fabric8 Spring Cloud integration with Kubernetes library to access Kubernetes ConfigMap and Secrets settings. The ConfigMap settings for RabbitMQ are specified in the src/kubernetes/server/server-config-rabbit.yaml file and Kafka in the src/kubernetes/server/server-config-kafka.yaml file. MySQL secrets are located in the src/kubernetes/mysql/mysql-secrets.yaml file. If you modified the password for MySQL you should have changed it in the src/kubernetes/mysql/mysql-secrets.yaml file. Any secrets have to be provided base64 encoded.

    We are now configuring the Data Flow server with file based security and the default user is 'user' with a password of 'password'. Feel free to change this in the src/kubernetes/server/server-config-rabbit.yaml file for RabbitMQ and the src/kubernetes/server/server-config-kafka.yaml file when using Kafka.
    The default memory for the pods is set to 1024Mi. Update the value in the src/kubernetes/server/server-deployment.yaml file if you expect most of your apps to require more memory.
    The latest releases of kubernetes have enabled RBAC on the api-server. If your target platform has RBAC enabled you must ask a cluster-admin to create the roles and role-bindings for you before deploying the dataflow server. They associate the dataflow service account with the roles it needs to be run with.

    Create Role Bindings and Service account:

    $ kubectl create -f src/kubernetes/server/server-roles.yaml
    $ kubectl create -f src/kubernetes/server/server-rolebinding.yaml
    $ kubectl create -f src/kubernetes/server/service-account.yaml

    When using RabbitMQ:

    $ kubectl create -f src/kubernetes/server/server-config-rabbit.yaml

    When using Kafka:

    $ kubectl create -f src/kubernetes/server/server-config-kafka.yaml

    Create server deployment:

    $ kubectl create -f src/kubernetes/server/server-svc.yaml
    $ kubectl create -f src/kubernetes/server/server-deployment.yaml

    You can use the command kubectl get all -l app=scdf-server to verify that the deployment, pod and service resources are running. Use the command kubectl delete all,cm -l app=scdf-server to clean up afterwards. To cleanup roles, bindings and the service account, use the following commands:

    $ kubectl delete role scdf-role
    $ kubectl delete rolebinding scdf-rb
    $ kubectl delete serviceaccount scdf-sa

    Use the kubectl get svc scdf-server command to locate the EXTERNAL_IP address assigned to scdf-server, we will use that later to connect from the shell.

    $ kubectl get svc scdf-server
    NAME         CLUSTER-IP       EXTERNAL-IP       PORT(S)    AGE
    scdf-server  10.103.246.82    130.211.203.246   80/TCP     4m

    So the URL you need to use is in this case 130.211.203.246

    If you are using Minikube then you don’t have an external load balancer and the EXTERNAL-IP will show as <pending>. You need to use the NodePort assigned for the scdf-server service. Use this command to look up the URL to use:

    $ minikube service --url scdf-server
    http://192.168.99.100:31991

2. Helm Installation

Spring Cloud DataFlow offers a Helm Chart for deploying the Spring Cloud Data Flow server and its required services to a Kubernetes Cluster.

The helm chart is available since the 1.2 GA release of Spring Cloud Data Flow for Kubernetes.

The following instructions cover how to initialize Helm and install Spring Cloud Data Flow on a Kubernetes cluster.

  1. Installing Helm

    Helm is comprised of two components: one is the client (Helm) the other is the server (Tiller). The Helm client is run on your local machine and can be installed using the following instructions found here. If Tiller has not been installed on your cluster, execute the following Helm client command:

    $ helm init
    To verify that the Tiller pod is running execute the following command: kubectl get pod --namespace kube-system and you should see the Tiller pod running.
  2. Installing the Spring Cloud Data Flow Server and required services.

    Before we can run the Spring Cloud Data Flow Chart, we need to access the incubator repository where it currently resides. To add this repository to our Helm install, execute the following commands:

    helm repo add incubator https://kubernetes-charts-incubator.storage.googleapis.com
    helm repo update

    To install Spring Cloud Data Flow and its required services execute the following:

    helm install --name my-release incubator/spring-cloud-data-flow

    If you are running on a Kubernetes cluster without a load balancer, such as in Minikube, then you should override the service type to use NodePort. Add the --set server.service.type=NodePort override:

    helm install --name my-release --set server.service.type=NodePort \
        incubator/spring-cloud-data-flow

    If you are running on a Kubernetes cluster without RBAC, such as in minikube, then you should override rbac.create to false. By default, it is set to true based on best practices. Add the --set rbac.create=false override:

    helm install --name my-release --set server.service.type=NodePort \
        --set rbac.create=false \
        incubator/spring-cloud-data-flow

    If you wish to specify a different version of Spring Cloud Data Flow besides the current GA release, you can set the server.version as shown below:

    helm install --name my-release incubator/spring-cloud-data-flow --set server.version=<version-you-want>
    To see all of the settings that can be configured on the Spring Cloud Data Flow chart, check out the README.

    Here’s Spring Cloud Data Flow’s Kubernetes version compatibility with the respective Helm Chart releases.

    | SCDF-K8S-Server Version \ Chart Version | 0.1.x | 0.2.x |
    |-----------------------------------------|-------|-------|
    |1.2.x                                    |✓      |✕     |
    |1.3.x                                    |✕      |✓     |
    |1.4.x                                    |✕      |✓     |
    |1.5.x                                    |✕      |✓     |
    |1.6.x                                    |✕      |✓     |
    |---------------------------------------------------------|

    You should see the following output:

    NAME:   my-release
    LAST DEPLOYED: Sat Mar 10 11:33:29 2018
    NAMESPACE: default
    STATUS: DEPLOYED
    
    RESOURCES:
    ==> v1/Secret
    NAME                  TYPE    DATA  AGE
    my-release-mysql      Opaque  2     1s
    my-release-data-flow  Opaque  2     1s
    my-release-redis      Opaque  1     1s
    my-release-rabbitmq   Opaque  2     1s
    
    ==> v1/ConfigMap
    NAME                          DATA  AGE
    my-release-data-flow-server   1     1s
    my-release-data-flow-skipper  1     1s
    
    ==> v1/PersistentVolumeClaim
    NAME                 STATUS   VOLUME                                    CAPACITY  ACCESSMODES  STORAGECLASS  AGE
    my-release-rabbitmq  Bound    pvc-e9ed7f55-2499-11e8-886f-08002799df04  8Gi       RWO          standard      1s
    my-release-mysql     Pending  standard                                  1s
    my-release-redis     Pending  standard                                  1s
    
    ==> v1/ServiceAccount
    NAME                  SECRETS  AGE
    my-release-data-flow  1        1s
    
    ==> v1/Service
    NAME                          CLUSTER-IP      EXTERNAL-IP  PORT(S)                                AGE
    my-release-mysql              10.110.98.253   <none>       3306/TCP                               1s
    my-release-data-flow-server   10.105.216.155  <pending>    80:32626/TCP                           1s
    my-release-redis              10.111.63.33    <none>       6379/TCP                               1s
    my-release-data-flow-metrics  10.107.157.1    <none>       80/TCP                                 1s
    my-release-rabbitmq           10.106.76.215   <none>       4369/TCP,5672/TCP,25672/TCP,15672/TCP  1s
    my-release-data-flow-skipper  10.100.28.64    <none>       80/TCP                                 1s
    
    ==> v1beta1/Deployment
    NAME                          DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE
    my-release-mysql              1        1        1           0          1s
    my-release-rabbitmq           1        1        1           0          1s
    my-release-data-flow-metrics  1        1        1           0          1s
    my-release-data-flow-skipper  1        1        1           0          1s
    my-release-redis              1        1        1           0          1s
    my-release-data-flow-server   1        1        1           0          1s
    
    
    NOTES:
    1. Get the application URL by running these commands:
         NOTE: It may take a few minutes for the LoadBalancer IP to be available.
               You can watch the status of the server by running 'kubectl get svc -w my-release-data-flow-server'
      export SERVICE_IP=$(kubectl get svc --namespace default my-release-data-flow-server -o jsonpath='{.status.loadBalancer.ingress[0].ip}')
      echo http://$SERVICE_IP:80

    You have just created a new release in the default namespace of your Kubernetes cluster. The notes section gives instructions for connecting to the newly installed server. It takes a couple of minutes for the application and its required services to start up. You can check on the status by issuing a kubectl get pod -w command. Wait for the READY column to show "1/1" for all pods. Once that is done, you can connect to the Data Flow server using the external ip listed via a kubectl get svc my-release-data-flow-server command. The default username is user, and the password is password.

    If you are running on Minikube then you can use the following command to get the URL for the server:

    minikube service --url my-release-data-flow-server

    To see what Helm releases you have running, you can use the helm list command. When it is time to delete the release, run helm delete my-release. This removes any resources created for the release but keeps release information so you can rollback any changes using a helm rollback my-release 1 command. To completely delete the release and purge any release metadata, use helm delete my-release --purge.

    There is an issue with generated secrets used for the required services getting rotated on chart upgrades. To avoid this set the password for these services when installing the chart. You can use:

    helm install --name my-release \
        --set rabbitmq.rabbitmqPassword=rabbitpwd \
        --set mysql.mysqlRootPassword=mysqlpwd \
        --set redis.redisPassword=redispwd incubator/spring-cloud-data-flow

3. Deploying Streams

3.1. Create Streams without Skipper

  1. Download and run the Spring Cloud Data Flow shell.

    wget http://repo.spring.io/release/org/springframework/cloud/spring-cloud-dataflow-shell/1.6.0.RELEASE/spring-cloud-dataflow-shell-1.6.0.RELEASE.jar
    
    $ java -jar spring-cloud-dataflow-shell-1.6.0.RELEASE.jar

    That should give you the following startup message from the shell:

      ____                              ____ _                __
     / ___| _ __  _ __(_)_ __   __ _   / ___| | ___  _   _  __| |
     \___ \| '_ \| '__| | '_ \ / _` | | |   | |/ _ \| | | |/ _` |
      ___) | |_) | |  | | | | | (_| | | |___| | (_) | |_| | (_| |
     |____/| .__/|_|  |_|_| |_|\__, |  \____|_|\___/ \__,_|\__,_|
      ____ |_|    _          __|___/                 __________
     |  _ \  __ _| |_ __ _  |  ___| | _____      __  \ \ \ \ \ \
     | | | |/ _` | __/ _` | | |_  | |/ _ \ \ /\ / /   \ \ \ \ \ \
     | |_| | (_| | || (_| | |  _| | | (_) \ V  V /    / / / / / /
     |____/ \__,_|\__\__,_| |_|   |_|\___/ \_/\_/    /_/_/_/_/_/
    
    1.6.0.RELEASE
    
    Welcome to the Spring Cloud Data Flow shell. For assistance hit TAB or type "help".
    server-unknown:>

    Configure the Data Flow server URI with the following command (use the URL determined above in the previous step) using the default user and password settings:

    server-unknown:>dataflow config server --username user --password password --uri http://130.211.203.246/
    Successfully targeted http://130.211.203.246/
    dataflow:>
  2. Register the Docker images of the Rabbit binder based time and log apps using the shell.

    When using RabbitMQ:

    dataflow:>app register --type source --name time --uri docker://springcloudstream/time-source-rabbit:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:time-source-rabbit:jar:metadata:1.3.1.RELEASE
    dataflow:>app register --type sink --name log --uri docker://springcloudstream/log-sink-rabbit:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:log-sink-rabbit:jar:metadata:1.3.1.RELEASE

    When using Kafka:

    dataflow:>app register --type source --name time --uri docker://springcloudstream/time-source-kafka-10:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:time-source-kafka-10:jar:metadata:1.3.1.RELEASE
    dataflow:>app register --type sink --name log --uri docker://springcloudstream/log-sink-kafka-10:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:log-sink-kafka-10:jar:metadata:1.3.1.RELEASE
  3. Alternatively, if you would like to register all out-of-the-box stream applications for a particular binder in bulk, you can use one of the following commands. For more details, review how to register applications.

    When using RabbitMQ:

    dataflow:>app import --uri http://bit.ly/Celsius-SR3-stream-applications-rabbit-docker

    When using Kafka:

    dataflow:>app import --uri http://bit.ly/Celsius-SR3-stream-applications-kafka-10-docker
  4. Deploy a simple stream in the shell

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

    You can use the command kubectl get pods to check on the state of the pods corresponding to this stream. We can run this from the shell by running it as an OS command by adding a "!" before the command.

    dataflow:>! kubectl get pods -l role=spring-app
    command is:kubectl get pods -l role=spring-app
    NAME                  READY     STATUS    RESTARTS   AGE
    ticktock-log-0-qnk72  1/1       Running   0          2m
    ticktock-time-r65cn   1/1       Running   0          2m

    Look at the logs for the pod deployed for the log sink.

    dataflow:>! kubectl logs ticktock-log-0-qnk72
    command is:kubectl logs ticktock-log-0-qnk72
    ...
    2017-07-20 04:34:37.369  INFO 1 --- [time.ticktock-1] log-sink                                 : 07/20/17 04:34:37
    2017-07-20 04:34:38.371  INFO 1 --- [time.ticktock-1] log-sink                                 : 07/20/17 04:34:38
    2017-07-20 04:34:39.373  INFO 1 --- [time.ticktock-1] log-sink                                 : 07/20/17 04:34:39
    2017-07-20 04:34:40.380  INFO 1 --- [time.ticktock-1] log-sink                                 : 07/20/17 04:34:40
    2017-07-20 04:34:41.381  INFO 1 --- [time.ticktock-1] log-sink                                 : 07/20/17 04:34:41
  5. Destroy the stream

    dataflow:>stream destroy --name ticktock

    A useful command to help in troubleshooting issues, such as a container that has a fatal error starting up, add the options --previous to view last terminated container log. You can also get more detailed information about the pods by using the kubctl describe like:

    kubectl describe pods/ticktock-log-qnk72
    If you need to specify any of the app specific configuration properties then you might use "long-form" of them including the app specific prefix like --jdbc.tableName=TEST_DATA. This form is required if you didn’t register the --metadata-uri for the Docker based starter apps. In this case you will also not see the configuration properties listed when using the app info command or in the Dashboard GUI.

3.2. Create Streams with Skipper

Refer to the section Streams deployed using Skipper for more information.

3.3. Accessing app from outside the cluster

If you need to be able to connect to from outside of the Kubernetes cluster to an app that you deploy, like the http-source, then you need to use either an external load balancer for the incoming connections or you need to use a NodePort configuration that will expose a proxy port on each Kubetnetes Node. If your cluster doesn’t support external load balancers, like the Minikube, then you must use the NodePort approach. You can use deployment properties for configuring the access. Use deployer.http.kubernetes.createLoadBalancer=true for the app to specify that you want to have a LoadBalancer with an external IP address created for your app’s service. For the NodePort configuration use deployer.http.kubernetes.createNodePort=<port> where <port> should be a number between 30000 and 32767.

  1. Register the http-source, you can use one of the following commands:

    When using RabbitMQ:

    dataflow:>app register --type source --name http --uri docker//springcloudstream/http-source-rabbit:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:http-source-rabbit:jar:metadata:1.3.1.RELEASE

    When using Kafka:

    dataflow:>app register --type source --name http --uri docker//springcloudstream/http-source-kafka:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:http-source-kafka:jar:metadata:1.3.1.RELEASE
  2. Create the http | log stream without deploying it using the following command:

    dataflow:>stream create --name test --definition "http | log"
  3. If your cluster supports an External LoadBalancer for the http-source, then you can use the following command to deploy the stream:

    dataflow:>stream deploy test --properties "deployer.http.kubernetes.createLoadBalancer=true"

    Wait for the pods to be started showing 1/1 in the READY column by using this command:

    dataflow:>! kubectl get pods -l role=spring-app
    command is:kubectl get pods -l role=spring-app
    NAME               READY     STATUS    RESTARTS   AGE
    test-http-2bqx7    1/1       Running   0          3m
    test-log-0-tg1m4   1/1       Running   0          3m

    Now, look up the external IP address for the http app (it can sometimes take a minute or two for the external IP to get assigned):

    dataflow:>! kubectl get service test-http
    command is:kubectl get service test-http
    NAME         CLUSTER-IP       EXTERNAL-IP      PORT(S)    AGE
    test-http    10.103.251.157   130.211.200.96   8080/TCP   58s
  4. If you are using Minikube, or any cluster that doesn’t support an External LoadBalancer, then you should deploy the stream with a NodePort in the range of 30000-32767. Use the following command to deploy it:

    dataflow:>stream deploy test --properties "deployer.http.kubernetes.createNodePort=32123"

    Wait for the pods to be started showing 1/1 in the READY column by using this command:

    dataflow:>! kubectl get pods -l role=spring-app
    command is:kubectl get pods -l role=spring-app
    NAME               READY     STATUS    RESTARTS   AGE
    test-http-9obkq    1/1       Running   0          3m
    test-log-0-ysiz3   1/1       Running   0          3m

    Now look up the URL to use with the following command:

    dataflow:>! minikube service --url test-http
    command is:minikube service --url test-http
    http://192.168.99.100:32123
  5. Post some data to the test-http app either using the EXTERNAL-IP address from above with port 8080 or the URL provided by the minikube command:

    dataflow:>http post --target http://130.211.200.96:8080 --data "Hello"
  6. Finally, look at the logs for the test-log pod:

    dataflow:>! kubectl get pods-l role=spring-app
    command is:kubectl get pods-l role=spring-app
    NAME              READY     STATUS             RESTARTS   AGE
    test-http-9obkq   1/1       Running            0          2m
    test-log-0-ysiz3  1/1       Running            0          2m
    dataflow:>! kubectl logs test-log-0-ysiz3
    command is:kubectl logs test-log-0-ysiz3
    ...
    2016-04-27 16:54:29.789  INFO 1 --- [           main] o.s.c.s.b.k.KafkaMessageChannelBinder$3  : started inbound.test.http.test
    2016-04-27 16:54:29.799  INFO 1 --- [           main] o.s.c.support.DefaultLifecycleProcessor  : Starting beans in phase 0
    2016-04-27 16:54:29.799  INFO 1 --- [           main] o.s.c.support.DefaultLifecycleProcessor  : Starting beans in phase 2147482647
    2016-04-27 16:54:29.895  INFO 1 --- [           main] s.b.c.e.t.TomcatEmbeddedServletContainer : Tomcat started on port(s): 8080 (http)
    2016-04-27 16:54:29.896  INFO 1 --- [  kafka-binder-] log.sink                                 : Hello
  7. Destroy the stream

    dataflow:>stream destroy --name test

4. Deploying Tasks

  1. Create a task and launch it

    Let’s register the timestamp task app and create a simple task definition and launch it.

    dataflow:>app register --type task --name timestamp --uri docker:springcloudtask/timestamp-task:1.3.1.RELEASE --metadata-uri maven://org.springframework.cloud.task.app:timestamp-task:jar:metadata:1.3.1.RELEASE
    dataflow:>task create task1 --definition "timestamp"
    dataflow:>task launch task1

    We can now list the tasks and executions using these commands:

    dataflow:>task list
    ╔═════════╤═══════════════╤═══════════╗
    ║Task Name│Task Definition│Task Status║
    ╠═════════╪═══════════════╪═══════════╣
    ║task1    │timestamp      │running    ║
    ╚═════════╧═══════════════╧═══════════╝
    
    dataflow:>task execution list
    ╔═════════╤══╤════════════════════════════╤════════════════════════════╤═════════╗
    ║Task Name│ID│         Start Time         │          End Time          │Exit Code║
    ╠═════════╪══╪════════════════════════════╪════════════════════════════╪═════════╣
    ║task1    │1 │Fri May 05 18:12:05 EDT 2017│Fri May 05 18:12:05 EDT 2017│0        ║
    ╚═════════╧══╧════════════════════════════╧════════════════════════════╧═════════╝
  2. Destroy the task

    dataflow:>task destroy --name task1

5. Application and Server Properties

This section covers how you can customize the deployment of your applications. You can use a number of properties to influence settings for the applications that are deployed. Properties can be applied on a per application basis or in the server configuration for all deployed applications.

Properties set on a per application basis will always take precedence over properties set as the server configuration. This allows for the ability to override global server level properties on a per-application basis.

See KubernetesDeployerProperties for more of the supported options.

5.1. Using Deployments

The deployer uses Replication Controllers by default. To use Deployments instead you can set the following option as part of the container env section in a deployment YAML file. This is now the preferred setting and will be the default in future releases of the deployer.

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_CREATE_DEPLOYMENT
          value: 'true'

5.2. Memory and CPU Settings

The apps are deployed by default with the following "Limits" and "Requests" settings:

    Limits:
      cpu:	500m
      memory:	512Mi
    Requests:
      cpu:	500m
      memory:	512Mi

You might find that the 512Mi memory limit is too low and to increase it you can provide a common spring.cloud.deployer.memory deployer property like this (replace <app> with the name of the app you would like to set this for):

deployer.<app>.memory=640m

This property affects both the Requests and Limits memory value set for the container.

If you would like to set the Requests and Limits values separately you would have to use the deployer properties that are specific to the Kubernetes deployer. To set the Limits to 1000m for cpu, 1024Mi for memory and Requests to 800m for cpu, 640Mi for memory you can use the following properties:

deployer.<app>.kubernetes.limits.cpu=1000m
deployer.<app>.kubernetes.limits.memory=1024Mi
deployer.<app>.kubernetes.requests.cpu=800m
deployer.<app>.kubernetes.requests.memory=640Mi

That should result in the following container settings being used:

    Limits:
      cpu:	1
      memory:	1Gi
    Requests:
      cpu:	800m
      memory:	640Mi
When using the common memory property you should use and m suffix for the value while when using the Kubernetes specific properties you should use the Kubernetes Mi style suffix.

You can also control the default values to set the cpu and memory requirements for the pods that are created as part of app deployments. The following can be declared as part of the container env section in a deployment YAML file:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_CPU
          value: 500m
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_MEMORY
          value: 640Mi

The settings we have used so far only affect the settings for the container, they do not affect the memory setting for the JVM process in the container. If you would like to set JVM memory settings you can provide an environment variable for this, see the next section for details.

5.3. Environment Variables

To influence the environment settings for a given app, you can take advantage of the spring.cloud.deployer.kubernetes.environmentVariables deployer property. For example, a common requirement in production settings is to influence the JVM memory arguments. This can be achieved by using the JAVA_TOOL_OPTIONS environment variable:

deployer.<app>.kubernetes.environmentVariables=JAVA_TOOL_OPTIONS=-Xmx1024m
The environmentVariables property accepts a comma delimited string. If an environment variable contains a value which is also a comma delimited string, then it must be enclosed in single quotes, e.g., spring.cloud.deployer.kubernetes.environmentVariables=spring.cloud.stream.kafka.binder.brokers='somehost:9092, anotherhost:9093'

This overrides the JVM memory setting for the desired <app> (just replace <app> with the name of your app).

5.4. Liveness and Readiness Probes

The liveness and readiness probes are using the paths /health and /info respectively. They use a delay of 10 for both and a period of 60 and 10 respectively. You can change these defaults when you deploy the stream by using deployer properties.

Here is an example changing the liveness probe (just replace <app> with the name of your app) via deployer properties:

deployer.<app>.kubernetes.livenessProbePath=/health
deployer.<app>.kubernetes.livenessProbeDelay=120
deployer.<app>.kubernetes.livenessProbePeriod=20

The same can be declared as part of the container env section in a deployment YAML file:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_LIVENESS_PROBE_PATH
          value: '/health'
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_LIVENESS_PROBE_DELAY
          value: '120'
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_LIVENESS_PROBE_PERIOD
          value: '20'

Similarly, swap liveness for readiness to override the default readiness settings.

By default, port 8080 is used as the probe port. You can change the defaults for both liveness and readiness probe ports by using deployer properties, for example:

deployer.<app>.kubernetes.readinessProbePort=7000
deployer.<app>.kubernetes.livenessProbePort=7000

As well as in the container env section of a deployment YAML file:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_READINESS_PROBE_PORT
          value: '7000'
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_LIVENESS_PROBE_PORT
          value: '7000'

If you intend to use Spring Boot 2.x+, please note that all Actuator endpoints in Spring Boot 2.x have been moved under /actuator by default. You must adjust the liveness and readiness probe paths to the new defaults. Here is an example of configuring Spring Boot 2.x+ liveness and readiness endpoint paths (just replace <app> with the name of your app):

deployer.<app>.kubernetes.livenessProbePath=/actuator/health
deployer.<app>.kubernetes.readinessProbePath=/actuator/info

To automatically set both liveness and readiness endpoints per application to the default Spring Boot 2.x paths, the following property can be set:

deployer.<app>.kubernetes.bootMajorVersion=2

If desired, see the Spring Boot 2.0 Migration Guide for more information and how to restore the Spring Boot 1.x base path behavior.

Probe endpoints that are secured can be accessed by using credentials stored in a Kubernetes secret. An existing Secret can be used providing the credentials are contained under the "credentials" key name of the Secret’s data block. Probe authentication can be configured on a per application basis. When enabled, it is applied to both the liveness and readiness probe endpoints using the same credentials and authentication type. Currently only Basic authentication is supported. A new secret can be created as follows:

Basic authentication encodes a username and password as a base64 string in the format of username:password. First generate the base64 string with the credentials used to access the secured probe endpoints, for example:

$ echo -n "user:pass" | base64
dXNlcjpwYXNz
$

Replacing user and pass with the appropriate values. With the encoded credentials, create a file for example myprobesecret.yml with the following contents:

apiVersion: v1
kind: Secret
metadata:
  name: myprobesecret
type: Opaque
data:
  credentials: GENERATED_BASE64_STRING

Replacing GENERATED_BASE64_STRING with the base64 encoded value generated above. Now create the Secret using kubectl:

$ kubectl create -f ./myprobesecret.yml
secret "myprobesecret" created
$

Then set the following deployer properties to use authentication when accessing probe endpoints:

deployer.<app>.kubernetes.probeCredentialsSecret=myprobesecret

Replacing <app> with the name of the application to apply authentication to.

5.5. Using SPRING_APPLICATION_JSON

Data Flow Server properties that are common across all of the Data Flow Server implementations including the configuration of maven repository settings can be set at the server level in the container env section of a deployment YAML using a SPRING_APPLICATION_JSON environment variable as shown:

        env:
        - name: SPRING_APPLICATION_JSON
          value: "{ \"maven\": { \"local-repository\": null, \"remote-repositories\": { \"repo1\": { \"url\": \"https://repo.spring.io/libs-snapshot\"} } } }"

5.6. Private Docker Registry

Docker images can be pulled from a private registry on a per app basis. First a Secret must be created in the cluster. Follow the Pull an Image from a Private Registry guide to create the Secret.

Once the Secret is created, use the imagePullSecret property to set the Secret to use, for example:

deployer.<app>.kubernetes.imagePullSecret=mysecret

Replacing <app> with the name of your app and mysecret with the name of the Secret you created earlier.

The image pull secret can also be configured at the server level in the container env section of a deployment YAML, for example:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_IMAGE_PULL_SECRET
          value: mysecret

Replacing mysecret with the name of the Secret you created earlier.

5.7. Annotations

Annotations can be added to Kubernetes objects on a per app basis. The supported object types are pod Deployment, Service and Job. Annotations are defined in a key:value format allowing for multiple annotations separated by a comma. For more information and use cases on annotations see Annotations.

Applications can be configured as such:

deployer.<app>.kubernetes.podAnnotations=annotationName:annotationValue
deployer.<app>.kubernetes.serviceAnnotations=annotationName:annotationValue,annotationName2:annotationValue2
deployer.<app>.kubernetes.jobAnnotations=annotationName:annotationValue

Replacing <app> with the name of your app and the value of your annotation(s).

5.8. Entry Point Style

An Entry Point Style affects how application properties are passed to the container to be deployed. Currently there are three supported styles:

  • exec - The default Entry Point Style and passes all application properties as command line arguments

  • shell - Passes all application properties as environment variables

  • boot - Creates an environment variable SPRING_APPLICATION_JSON containing a JSON representation of all application properties

Applications can be configured as such:

deployer.<app>.kubernetes.entryPointStyle=<Entry Point Style>

Replacing <app> with the name of your app and the desired Entry Point Style.

The Entry Point Style can also be configured at the server level in the container env section of a deployment YAML, for example:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_ENTRY_POINT_STYLE
          value: entryPointStyle

Replacing entryPointStye with the desired Entry Point Style.

Choosing an Entry Point Style of either exec or shell corresponds to how the ENTRYPOINT syntax is defined in the containers Dockerfile. For more information and uses cases on exec vs shell see the ENTRYPOINT section of the Docker documentation.

Using the Entry Point Style of boot corresponds to using the exec style ENTRYPOINT. Command line arguments from the deployment request are passed to the container, with the addition of application properties mapped into the SPRING_APPLICATION_JSON environment variable rather than command line arguments.

When using the boot Entry Point Style, the deployer.<app>.kubernetes.environmentVariables property must not contain SPRING_APPLICATION_JSON.

5.9. Deployment Service Account

A custom Service Account for application deployments can be configured through properties. An existing Service Account can be used, or a new one can be created. One way to create a service account is by using kubectl, for example:

$ kubectl create serviceaccount myserviceaccountname
serviceaccount "myserviceaccountname" created

Then individual applications can be configured as such:

deployer.<app>.kubernetes.deploymentServiceAccountName=myserviceaccountname

Replacing <app> with the name of your app to be deployed using the provided Service Account name.

The Service Account Name can also be configured at the server level in the container env section of a deployment YAML, for example:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_DEPLOYMENT_SERVICE_ACCOUNT_NAME
          value: myserviceaccountname

Replacing myserviceaccountname with the Service Account Name to be applied to all deployments.

5.10. Image Pull Policy

An Image Pull Policy defines when a Docker image should be pulled to the local registry. Currently there are three supported policies:

  • IfNotPresent - The default policy, which will not pull an image if it already exists

  • Always - Always pulls the image regardless if it already exists or not

  • Never - Never pull an image, only use an image that already exists

Applications can be individually configured as such:

deployer.<app>.kubernetes.imagePullPolicy=Always

Replacing <app> with the name of your app and the desired Image Pull Policy.

An Image Pull Policy can be configured at the server level in the container env section of a deployment YAML, for example:

        env:
        - name: SPRING_CLOUD_DEPLOYER_KUBERNETES_DEPLOYMENT_IMAGE_PULL_POLICY
          value: Always

Replacing Always with the desired Image Pull Policy.

Applications

A selection of pre-built stream and task/batch starter apps for various data integration and processing scenarios to facilitate learning and experimentation. The table below includes the pre-built applications at a glance. For more details, review how to register supported applications.

6. Available Applications

Architecture

7. Introduction

Spring Cloud Data Flow simplifies the development and deployment of applications focused on data processing use cases. The major concepts of the architecture are Applications, the Data Flow Server, and the target runtime.

Applications come in two flavors:

  • Long-lived Stream applications where an unbounded amount of data is consumed or produced through messaging middleware.

  • Short-lived Task applications that process a finite set of data and then terminate.

Depending on the runtime, applications can be packaged in two ways:

  • Spring Boot uber-jar that is hosted in a maven repository, file, or HTTP(S).

  • Docker image.

The runtime is the place where applications execute. The target runtimes for applications are platforms that you may already be using for other application deployments.

The supported platforms are:

  • Cloud Foundry

  • Apache YARN

  • Kubernetes

  • Apache Mesos

  • Local Server for development

There is a deployer Service Provider Interface (SPI) that lets you extend Data Flow to deploy onto other runtimes. There are community implementations of Hashicorp’s Nomad and RedHat Openshift. We look forward to working with the community for further contributions!

There are two mutually exclusive options that determine how applications are deployed to the platform.

  1. Select a Spring Cloud Data Flow Server executable jar that targets a single platform.

  2. Enable the Spring Cloud Data Flow Server to delegate the deployment and runtime status of applications to the Spring Cloud Skipper Server, which has the capability to deploy to multiple platforms.

Selecting the Spring Cloud Skipper option also enables the ability to update and rollback applications in a Stream at runtime.

The Data Flow server is also responsible for:

  • Interpreting and executing a stream DSL that describes the logical flow of data through multiple long-lived applications.

  • Launching a long-lived task application.

  • Interpreting and executing a composed task DSL that describes the logical flow of data through multiple short-lived applications.

  • Applying a deployment manifest that describes the mapping of applications onto the runtime - for example, to set the initial number of instances, memory requirements, and data partitioning.

  • Providing the runtime status of deployed applications.

As an example, the stream DSL to describe the flow of data from an HTTP source to an Apache Cassandra sink would be written using a Unix pipes and filter syntax " http | cassandra ". Each name in the DSL is mapped to an application that can that Maven or Docker repositories. You can also register an application to an http location. Many source, processor, and sink applications for common use cases (such as JDBC, HDFS, HTTP, and router) are provided by the Spring Cloud Data Flow team. The pipe symbol represents the communication between the two applications through messaging middleware. The two messaging middleware brokers that are supported are:

  • Apache Kafka

  • RabbitMQ

In the case of Kafka, when deploying the stream, the Data Flow server is responsible for creating the topics that correspond to each pipe symbol and configure each application to produce or consume from the topics so the desired flow of data is achieved. Similarly for RabbitMQ, exchanges and queues are created as needed to achieve the desired flow.

The interaction of the main components is shown in the following image:

The Spring Cloud Data Flow High Level Architecture
Figure 1. The Spring Cloud Data High Level Architecure

In the preceding diagram, a DSL description of a stream is POSTed to the Data Flow Server. Based on the mapping of DSL application names to Maven and Docker artifacts, the http-source and cassandra-sink applications are deployed on the target runtime. Data that is posted to the HTTP application will then be stored in Cassandra. The Samples Repository shows this use case in full detail.

8. Microservice Architectural Style

The Data Flow Server deploys applications onto the target runtime that conform to the microservice architectural style. For example, a stream represents a high-level application that consists of multiple small microservice applications each running in their own process. Each microservice application can be scaled up or down independently of the other and each has its own versioning lifecycle. Using Data Flow with Skipper enables you to independently upgrade or rollback each application at runtime.

Both Streaming and Task-based microservice applications build upon Spring Boot as the foundational library. This gives all microservice applications functionality such as health checks, security, configurable logging, monitoring, and management functionality, as well as executable JAR packaging.

It is important to emphasize that these microservice applications are 'just apps' that you can run by yourself by using java -jar and passing in appropriate configuration properties. We provide many common microservice applications for common operations so you need not start from scratch when addressing common use cases that build upon the rich ecosystem of Spring Projects, such as Spring Integration, Spring Data, and Spring Batch. Creating your own microservice application is similar to creating other Spring Boot applications. You can start by using the Spring Initializr web site to create the basic scaffolding of either a Stream or Task-based microservice.

In addition to passing the appropriate application properties to each applications, the Data Flow server is responsible for preparing the target platform’s infrastructure so that the applications can be deployed. For example, in Cloud Foundry, it would bind specified services to the applications and execute the cf push command for each application. For Kubernetes, it would create the replication controller, service, and load balancer.

The Data Flow Server helps simplify the deployment of multiple, relatated, applications onto a target runtime, setting up necessary input and output topics, partitions, and metrics functionality. However, one could also opt to deploy each of the microservice applications manually and not use Data Flow at all. This approach might be more appropriate to start out with for small scale deployments, gradually adopting the convenience and consistency of Data Flow as you develop more applications. Manual deployment of Stream- and Task-based microservices is also a useful educational exercise that can help you better understand some of the automatic application configuration and platform targeting steps that the Data Flow Server provides.

8.1. Comparison to Other Platform Architectures

Spring Cloud Data Flow’s architectural style is different than other Stream and Batch processing platforms. For example in Apache Spark, Apache Flink, and Google Cloud Dataflow, applications run on a dedicated compute engine cluster. The nature of the compute engine gives these platforms a richer environment for performing complex calculations on the data as compared to Spring Cloud Data Flow, but it introduces the complexity of another execution environment that is often not needed when creating data-centric applications. That does not mean you cannot do real-time data computations when using Spring Cloud Data Flow. Refer to the section Analytics, which describes the integration of Redis to handle common counting-based use cases. Spring Cloud Stream also supports using Reactive APIs such as Project Reactor and RxJava which can be useful for creating functional style applications that contain time-sliding-window and moving-average functionality. Similarly, Spring Cloud Stream also supports the development of applications in that use the Kafka Streams API.

Apache Storm, Hortonworks DataFlow, and Spring Cloud Data Flow’s predecessor, Spring XD, use a dedicated application execution cluster, unique to each product, that determines where your code should run on the cluster and performs health checks to ensure that long-lived applications are restarted if they fail. Often, framework-specific interfaces are required in order to correctly “plug in” to the cluster’s execution framework.

As we discovered during the evolution of Spring XD, the rise of multiple container frameworks in 2015 made creating our own runtime a duplication of effort. There is no reason to build your own resource management mechanics when there are multiple runtime platforms that offer this functionality already. Taking these considerations into account is what made us shift to the current architecture, where we delegate the execution to popular runtimes, which you may already be using for other purposes. This is an advantage in that it reduces the cognitive distance for creating and managing data-centric applications as many of the same skills used for deploying other end-user/web applications are applicable.

9. Data Flow Server

The Data Flow Server provides the following functionality:

9.1. Endpoints

The Data Flow Server uses an embedded servlet container and exposes REST endpoints for creating, deploying, undeploying, and destroying streams and tasks, querying runtime state, analytics, and the like. The Data Flow Server is implemented by using Spring’s MVC framework and the Spring HATEOAS library to create REST representations that follow the HATEOAS principle, as shown in the following image:

The Spring Cloud Data Flow Server Architecture
Figure 2. The Spring Cloud Data Flow Server

9.2. Security

The Data Flow Server executable jars support basic HTTP, LDAP(S), File-based, and OAuth 2.0 authentication to access its endpoints. Refer to the security section for more information.

10. Streams

10.1. Topologies

The Stream DSL describes linear sequences of data flowing through the system. For example, in the stream definition http | transformer | cassandra, each pipe symbol connects the application on the left to the one on the right. Named channels can be used for routing and to fan in/fan out data to multiple messaging destinations.

The concept of a tap can be used to ‘listen’ to the data that is flowing across any of the pipe symbols. "Taps" are just other streams that use an input any one of the "pipes" in a target stream and have an independent life cycle from the target stream.

10.2. Concurrency

For an application that consumes events, Spring Cloud Stream exposes a concurrency setting that controls the size of a thread pool used for dispatching incoming messages. See the {spring-cloud-stream-docs}#_consumer_properties[Consumer properties] documentation for more information.

10.3. Partitioning

A common pattern in stream processing is to partition the data as it moves from one application to the next. Partitioning is a critical concept in stateful processing, for either performance or consistency reasons, to ensure that all related data is processed together. For example, in a time-windowed average calculation example, it is important that all measurements from any given sensor are processed by the same application instance. Alternatively, you may want to cache some data related to the incoming events so that it can be enriched without making a remote procedure call to retrieve the related data.

Spring Cloud Data Flow supports partitioning by configuring Spring Cloud Stream’s output and input bindings. Spring Cloud Stream provides a common abstraction for implementing partitioned processing use cases in a uniform fashion across different types of middleware. Partitioning can thus be used whether the broker itself is naturally partitioned (for example, Kafka topics) or not (RabbitMQ). The following image shows how data could be partitioned into two buckets, such that each instance of the average processor application consumes a unique set of data.

Stream Partitioning Architecture
Figure 3. Spring Cloud Stream Partitioning

To use a simple partitioning strategy in Spring Cloud Data Flow, you need only set the instance count for each application in the stream and a partitionKeyExpression producer property when deploying the stream. The partitionKeyExpression identifies what part of the message is used as the key to partition data in the underlying middleware. An ingest stream can be defined as http | averageprocessor | cassandra. (Note that the Cassandra sink is not shown in the diagram above.) Suppose the payload being sent to the HTTP source was in JSON format and had a field called sensorId. For example, consider the case of deploying the stream with the shell command stream deploy ingest --propertiesFile ingestStream.properties where the contents of the ingestStream.properties file are as follows:

deployer.http.count=3
deployer.averageprocessor.count=2
app.http.producer.partitionKeyExpression=payload.sensorId

The result is to deploy the stream such that all the input and output destinations are configured for data to flow through the applications but also ensure that a unique set of data is always delivered to each averageprocessor instance. In this case, the default algorithm is to evaluate payload.sensorId % partitionCount where the partitionCount is the application count in the case of RabbitMQ and the partition count of the topic in the case of Kafka.

Please refer to Passing Stream Partition Properties for additional strategies to partition streams during deployment and how they map onto the underlying {spring-cloud-stream-docs}#_partitioning[Spring Cloud Stream Partitioning properties].

Also note that you cannot currently scale partitioned streams. Read Scaling at Runtime for more information.

10.4. Message Delivery Guarantees

Streams are composed of applications that use the Spring Cloud Stream library as the basis for communicating with the underlying messaging middleware product. Spring Cloud Stream also provides an opinionated configuration of middleware from several vendors, in particular providing {spring-cloud-stream-docs}#_persistent_publish_subscribe_support[persistent publish-subscribe semantics].

The {spring-cloud-stream-docs}#_binders[Binder abstraction] in Spring Cloud Stream is what connects the application to the middleware. There are several configuration properties of the binder that are portable across all binder implementations and some that are specific to the middleware.

For consumer applications, there is a retry policy for exceptions generated during message handling. The retry policy is configured by using the {spring-cloud-stream-docs}#_consumer_properties[common consumer properties] maxAttempts, backOffInitialInterval, backOffMaxInterval, and backOffMultiplier. The default values of these properties retry the callback method invocation 3 times and wait one second for the first retry. A backoff multiplier of 2 is used for the second and third attempts.

When the number of retry attempts has exceeded the maxAttempts value, the exception and the failed message become the payload of a message and are sent to the application’s error channel. By default, the default message handler for this error channel logs the message. You can change the default behavior in your application by creating your own message handler that subscribes to the error channel.

Spring Cloud Stream also supports a configuration option for both Kafka and RabbitMQ binder implementations that sends the failed message and stack trace to a dead letter queue. The dead letter queue is a destination and its nature depends on the messaging middleware (for example, in the case of Kafka, it is a dedicated topic). To enable this for RabbitMQ set the republishtoDlq and autoBindDlq {spring-cloud-stream-docs}#_rabbitmq_consumer_properties[consumer properties] and the autoBindDlq {spring-cloud-stream-docs}#_rabbit_producer_properties[producer property] to true when deploying the stream. To always apply these producer and consumer properties when deploying streams, configure them as common application properties when starting the Data Flow server.

Additional messaging delivery guarantees are those provided by the underlying messaging middleware that is chosen for the application for both producing and consuming applications. Refer to the Kafka {spring-cloud-stream-docs}#_kafka_consumer_properties[Consumer] and {spring-cloud-stream-docs}#_kafka_producer_properties[Producer] and Rabbit {spring-cloud-stream-docs}#_rabbitmq_consumer_properties[Consumer] and {spring-cloud-stream-docs}#_rabbit_producer_properties[Producer] documentation for more details. You can find extensive declarative support for all the native QOS options.

11. Stream Programming Models

While Spring Boot provides the foundation for creating DevOps-friendly microservice applications, other libraries in the Spring ecosystem help create Stream-based microservice applications. The most important of these is Spring Cloud Stream.

The essence of the Spring Cloud Stream programming model is to provide an easy way to describe multiple inputs and outputs of an application that communicate over messaging middleware. These input and outputs map onto Kafka topics or Rabbit exchanges and queues. Common application configuration for a Source that generates data, a Processor that consumes and produces data, and a Sink that consumes data is provided as part of the library.

11.1. Imperative Programming Model

Spring Cloud Stream is most closely integrated with Spring Integration’s imperative "one event at a time" programming model. This means you write code that handles a single event callback, as shown in the following example,

@EnableBinding(Sink.class)
public class LoggingSink {

    @StreamListener(Sink.INPUT)
    public void log(String message) {
        System.out.println(message);
    }
}

In this case, the String payload of a message coming on the input channel is handed to the log method. The @EnableBinding annotation is used to tie the input channel to the external middleware.

11.2. Functional Programming Model

However, Spring Cloud Stream can support other programming styles, such as reactive APIs, where incoming and outgoing data is handled as continuous data flows and how each individual message should be handled is defined. With many reactive AOIs, you can also use operators that describe functional transformations from inbound to outbound data flows. Here is an example:

@EnableBinding(Processor.class)
public static class UppercaseTransformer {

  @StreamListener
  @Output(Processor.OUTPUT)
  public Flux<String> receive(@Input(Processor.INPUT) Flux<String> input) {
    return input.map(s -> s.toUpperCase());
  }
}

12. Application Versioning

Application versioning within a Stream is now supported when using Data Flow together with Skipper. You can update application and deployment properties as well as the version of the application. Rolling back to a previous application version is also supported.

13. Task Programming Model

The Spring Cloud Task programming model provides:

  • Persistence of the Task’s lifecycle events and exit code status.

  • Lifecycle hooks to execute code before or after a task execution.

  • The ability to emit task events to a stream (as a source) during the task lifecycle.

  • Integration with Spring Batch Jobs.

See the Tasks section for more information.

14. Analytics

Spring Cloud Data Flow is aware of certain Sink applications that write counter data to Redis and provides a REST endpoint to read counter data. The types of counters supported are as follows:

  • Counter: Counts the number of messages it receives, optionally storing counts in a separate store such as Redis.

  • Field Value Counter: Counts occurrences of unique values for a named field in a message payload.

  • Aggregate Counter: Stores total counts but also retains the total count values for each minute, hour, day, and month.

Note that the timestamp used in the aggregate counter can come from a field in the message itself so that out-of-order messages are properly accounted.

15. Runtime

The Data Flow Server relies on the target platform for the following runtime functionality:

15.1. Fault Tolerance

The target runtimes supported by Data Flow all have the ability to restart a long-lived application. Spring Cloud Data Flow sets up health probes are required by the runtime environment when deploying the application. You also have the ability to customize the health probes.

The collective state of all applications that make up the stream is used to determine the state of the stream. If an application fails, the state of the stream changes from ‘deployed’ to ‘partial’.

15.2. Resource Management

Each target runtime lets you control the amount of memory, disk, and CPU allocated to each application. These are passed as properties in the deployment manifest by using key names that are unique to each runtime. Refer to each platform’s server documentation for more information.

15.3. Scaling at Runtime

When deploying a stream, you can set the instance count for each individual application that makes up the stream. Once the stream is deployed, each target runtime lets you control the target number of instances for each individual application. Using the APIs, UIs, or command line tools for each runtime, you can scale up or down the number of instances as required.

Currently, scaling at runtime is not supported with the Kafka binder, as well as with partitioned streams, for which the suggested workaround is redeploying the stream with an updated number of instances. Both cases require a static consumer to be set up, based on information about the total instance count and current instance index.

Server Configuration

In this section you will learn how to configure Spring Cloud Data Flow server’s features such as the relational database to use and security.

16. Feature Toggles

Data Flow server offers specific set of features that can be enabled/disabled when launching. These features include all the lifecycle operations, REST endpoints (server, client implementations including Shell and the UI) for:

  1. Streams

  2. Tasks

  3. Analytics

You can enable or disable these features by setting the following boolean environment variables when launching the Data Flow server:

  • SPRING_CLOUD_DATAFLOW_FEATURES_STREAMS_ENABLED

  • SPRING_CLOUD_DATAFLOW_FEATURES_TASKS_ENABLED

  • SPRING_CLOUD_DATAFLOW_FEATURES_ANALYTICS_ENABLED

By default, all the features are enabled.

Since analytics feature is enabled by default, the Data Flow server is expected to have a valid Redis store available as analytic repository as we provide a default implementation of analytics based on Redis. This also means that the Data Flow server’s health depends on the redis store availability as well. If you do not want to enable HTTP endpoints to read analytics data written to Redis, then disable the analytics feature using the property mentioned above.

The REST endpoint /features provides information on the features enabled/disabled.

17. General Configuration

The Spring Cloud Data Flow server for Kubernetes uses the Fabric8 spring-cloud-kubernetes module to process both ConfigMap and Secrets settings. You just need to enable the ConfigMap support by passing in an environment variable of SPRING_CLOUD_KUBERNETES_CONFIG_NAME and setting that to the name of the ConfigMap. Same is true for the Secrets where the environment variable is SPRING_CLOUD_KUBERNETES_SECRETS_NAME. To use the Secrets you also need to set SPRING_CLOUD_KUBERNETES_SECRETS_ENABLE_API to true.

Here is an example of a snippet from a deployment that sets these environment variables.

        env:
        - name: SPRING_CLOUD_KUBERNETES_SECRETS_ENABLE_API
          value: 'true'
        - name: SPRING_CLOUD_KUBERNETES_SECRETS_NAME
          value: mysql
        - name: SPRING_CLOUD_KUBERNETES_CONFIG_NAME
          value: scdf-server

17.1. Using ConfigMap and Secrets

Configuration properties can be passed to the Data Flow Server using Kubernetes ConfigMap and Secrets.

An example configuration could look like the following where we configure RabbitMQ, MySQL and Redis as well as basic security settings for the server:

apiVersion: v1
kind: ConfigMap
metadata:
  name: scdf-server
  labels:
    app: scdf-server
data:
  application.yaml: |-
    security:
      basic:
        enabled: true
        realm: Spring Cloud Data Flow
    spring:
      cloud:
        dataflow:
          security:
            authentication:
              file:
                enabled: true
                users:
                  admin: admin, ROLE_MANAGE, ROLE_VIEW
                  user: password, ROLE_VIEW, ROLE_CREATE
        deployer:
          kubernetes:
            environmentVariables: 'SPRING_RABBITMQ_HOST=${RABBITMQ_SERVICE_HOST},SPRING_RABBITMQ_PORT=${RABBITMQ_SERVICE_PORT},SPRING_REDIS_HOST=${REDIS_SERVICE_HOST},SPRING_REDIS_PORT=${REDIS_SERVICE_PORT}'
      datasource:
        url: jdbc:mysql://${MYSQL_SERVICE_HOST}:${MYSQL_SERVICE_PORT}/mysql
        username: root
        password: ${mysql-root-password}
        driverClassName: org.mariadb.jdbc.Driver
        testOnBorrow: true
        validationQuery: "SELECT 1"
      redis:
        host: ${REDIS_SERVICE_HOST}
        port: ${REDIS_SERVICE_PORT}

We assume here that RabbitMQ is deployed using rabbitmq as the service name. For MySQL we assume the service name is mysql and for Redis we assume it is redis. Kubernetes will publish these services' host and port values as environment variables that we can use when configuring the apps we deploy.

We prefer to provide the MySQL connection password in a Secrets file:

apiVersion: v1
kind: Secret
metadata:
  name: mysql
  labels:
    app: mysql
data:
  mysql-root-password: eW91cnBhc3N3b3Jk

The password is provided as a base64 encoded value.

18. Database Configuration

Spring Cloud Data Flow provides schemas for H2, HSQLDB, MySQL, Oracle, PostgreSQL, DB2 and SQL Server that will be automatically created when the server starts.

The JDBC drivers for MySQL (via MariaDB driver), HSQLDB, PostgreSQL along with embedded H2 are available out of the box. If you are using any other database, then the corresponding JDBC driver jar needs to be on the classpath of the server.

For instance, If you are using MySQL in addition to password in the Secrets file provide the following properties in the ConfigMap:

data:
  application.yaml: |-
    spring:
      datasource:
        url: jdbc:mysql://${MYSQL_SERVICE_HOST}:${MYSQL_SERVICE_PORT}/mysql
        username: root
        password: ${mysql-root-password}
        driverClassName: org.mariadb.jdbc.Driver
        url: jdbc:mysql://${MYSQL_SERVICE_HOST}:${MYSQL_SERVICE_PORT}/test
        driverClassName: org.mariadb.jdbc.Driver

For PostgreSQL:

data:
  application.yaml: |-
    spring:
      datasource:
        url: jdbc:postgresql://${PGSQL_SERVICE_HOST}:${PGSQL_SERVICE_PORT}/database
        username: root
        password: ${postgres-password}
        driverClassName: org.postgresql.Driver

For HSQLDB:

data:
  application.yaml: |-
    spring:
      datasource:
        url: jdbc:hsqldb:hsql://${HSQLDB_SERVICE_HOST}:${HSQLDB_SERVICE_PORT}/database
        username: sa
        driverClassName: org.hsqldb.jdbc.JDBCDriver
There is a schema update to the Spring Cloud Data Flow datastore when upgrading from version 1.0.x to 1.1.x and from 1.1.x to 1.2.x. Migration scripts for specific database types can be found in the spring-cloud-task repo.

19. Security

We are now securing the server application in the sample configurations file used in the Getting Started section.

This section covers the basic configuration settings we provide in the provided sample configuration, please refer to the core security documentation for more detailed coverage of the security configuration options for the Spring Cloud Data Flow server and shell.

When using RabbitMQ as the transport, the security settings are located in the src/kubernetes/server/server-config-rabbit.yaml file and when using Kafka the settings are located in the src/kubernetes/server/server-config-kafka.yaml file:

    security:
      basic:
        enabled: true                                         (1)
        realm: Spring Cloud Data Flow                         (2)
    spring:
      cloud:
        dataflow:
          security:
            authentication:
              file:
                enabled: true
                users:
                  admin: admin, ROLE_MANAGE, ROLE_VIEW        (3)
                  user: password, ROLE_VIEW, ROLE_CREATE      (4)
1 Enable security
2 Optionally set the realm, defaults to "Spring"
3 Create an 'admin' user with password set to 'admin' that can view apps, streams and tasks and that can also view management endpoints
4 Create a 'user' user with password set to 'password' than can register apps and create streams and tasks and also view them

Feel free to change user names and passwords to suite, and also maybe move the definition of user passwords to a Kubernetes Secret.

20. Monitoring and Management

We recommend using the kubectl command for troubleshooting streams and tasks.

You can list all artifacts and resources used by using the following command:

kubectl get all,cm,secrets,pvc

You can list all resources used by a specific app or service by using a label to select resources. The following command list all resources used by the mysql service:

kubectl get all -l app=mysql

You can get the logs for a specific pod by issuing:

kubectl logs pod <pod-name>

If the pod is continuously getting restarted you can add -p as an option to see the previous log like:

kubectl logs -p <pod-name>

You can also tail or follow a log by adding an -f option:

kubectl logs -f <pod-name>

A useful command to help in troubleshooting issues, such as a container that has a fatal error starting up, is to use the describe command like:

kubectl describe pod ticktock-log-0-qnk72

20.1. Inspecting Server Logs

You can access the server logs by using the following command (just supply the name of pod for the server):

kubectl get pod -l app=scdf=server
kubectl logs <scdf-server-pod-name>

20.2. Streams

The stream apps are deployed with the stream name followed by the name of the app and for processors and sinks there is also an instance index appended.

To see all the pods that are deployed by the Spring Cloud Data Flow server you can specify the label role=spring-app:

kubectl get pod -l role=spring-app

To see details for a specific app deployment you can use (just supply the name of pod for the app):

kubectl describe pod <app-pod-name>

For the application logs use:

kubectl logs <app-pod-name>

If you would like to tail a log you can use:

kubectl logs -f <app-pod-name>

20.3. Tasks

Tasks are launched as bare pods without a replication controller. The pods remain after the tasks complete and this gives you an opportunity to review the logs.

To see all pods for a specific task use this command while providing the task name:

kubectl get pod -l task-name=<task-name>

To review the task logs use:

kubectl logs <task-pod-name>

You have two options to delete completed pods. You can delete them manually once they are no longer needed.

To delete the task pod use:

kubectl delete pod <task-pod-name>

You can also use the Data Flow shell command task execution cleanup command to remove the completed pod for a task execution.

First we need to determine the ID for the task execution:

dataflow:>task execution list
╔═════════╤══╤════════════════════════════╤════════════════════════════╤═════════╗
║Task Name│ID│         Start Time         │          End Time          │Exit Code║
╠═════════╪══╪════════════════════════════╪════════════════════════════╪═════════╣
║task1    │1 │Fri May 05 18:12:05 EDT 2017│Fri May 05 18:12:05 EDT 2017│0        ║
╚═════════╧══╧════════════════════════════╧════════════════════════════╧═════════╝

Next we issue the command to cleanup the execution artifacts (the completed pod):

dataflow:>task execution cleanup --id 1
Request to clean up resources for task execution 1 has been submitted

21. Debug Support

Debugging the Spring Cloud Data Flow Kubernetes Server and included components such as the Spring Cloud Kubernetes Deployer is supported through the Java Debug Wire Protocol (JDWP). This section will outline an approach to manually enable debugging and another that uses configuration files provided with Spring Cloud Data Flow Server Kubernetes to "patch" a running deployment.

JDWP itself does not use any authentication. This section assumes debugging is being done on a local development environment such as Minikube and guidance on securing the debug port is not provided.

21.1. Enabling Debugging Manually

To manually enable JDWP, first edit src/kubernetes/server/server-deployment.yaml and add an additional containerPort entry under spec.template.spec.containers.ports with a value of 5005. Additionally add the environment variable JAVA_TOOL_OPTIONS under spec.template.spec.containers.env as shown below:

spec:
  ...
  template:
    ...
    spec:
      containers:
      - name: scdf-server
        ...
        ports:
        ...
		- containerPort: 5005
        env:
        - name: JAVA_TOOL_OPTIONS
          value: '-agentlib:jdwp=transport=dt_socket,server=y,suspend=n,address=5005'
Port 5005 is used in this example, but it can be any number that does not conflict with another port. The chosen port number must also be the same for the added containerPort value along with the address parameter of the JAVA_TOOL_OPTIONS -agentlib flag as shown above.

The Spring Cloud Data Flow Kubernetes Server can now be started as normal. Once the server is up, changes from above can be verified on the scdf-server deployment:

$ kubectl describe deployment/scdf-server
...
...
Pod Template:
  ...
  Containers:
   scdf-server:
    ...
    Ports:       80/TCP, 5005/TCP
    ...
    Environment:
      JAVA_TOOL_OPTIONS:  -agentlib:jdwp=transport=dt_socket,server=y,suspend=n,address=5005
	  ...

With the server started and JDWP enabled, access to the port needs to be configured. In this example we will use the port-forward subcommand of kubectl. Exposing a local port to our debug target using port-forward can be done as follows:

$ kubectl get pod -l app=scdf-server
NAME                           READY     STATUS    RESTARTS   AGE
scdf-server-5b7cfd86f7-d8mj4   1/1       Running   0          10m
$ kubectl port-forward scdf-server-5b7cfd86f7-d8mj4 5005:5005
Forwarding from 127.0.0.1:5005 -> 5005
Forwarding from [::1]:5005 -> 5005

A debugger may now be attached by pointing it to 127.0.0.1 as the host and 5005 as the port. The port-forward subcommand will remain running until killed with for example, CTRL+c.

Debugging support can be removed by reverting the changes to src/kubernetes/server/server-deployment.yaml. The reverted changes will be picked up on the next deployment of the Spring Cloud Data Flow Kubernetes Server. Manually adding debug support to the configuration comes in useful when debugging should be enabled by default each time the server is deployed.

21.2. Enabling Debugging with Patching

Rather than manually changing the server-deployment.yaml, created Kubernetes Objects can be "patched" in place. For convenience patch files are included that provide the same configuration as the manual approach. To enable debugging by patching, enter the following:

$ kubectl patch deployment scdf-server -p "$(cat src/kubernetes/server/server-deployment-debug.yaml)"

Running this command will automatically add the containerPort attribute and the JAVA_TOOL_OPTIONS environment variable. Changes to the scdf-server deployment can be verified by running the following command:

$ kubectl describe deployment/scdf-server
...
...
Pod Template:
  ...
  Containers:
   scdf-server:
    ...
    Ports:       5005/TCP, 80/TCP
    ...
    Environment:
      JAVA_TOOL_OPTIONS:  -agentlib:jdwp=transport=dt_socket,server=y,suspend=n,address=5005
	  ...

To enable access to the debug port, rather than using the port-forward subcommand of kubectl, another option would be to patch the scdf-server Kubernetes Service Object. Patching this object can be done as follows:

First ensure the scdf-server Kubernetes Service Object has the proper configuration.

$ kubectl describe service/scdf-server

If the output contains the text <unset>, for example:

Port:                     <unset>  80/TCP
TargetPort:               80/TCP
NodePort:                 <unset>  30784/TCP

Patch the service to add a name for this port:

$ kubectl patch service scdf-server -p "$(cat src/kubernetes/server/server-svc.yaml)"
A port name should only be missing if the target cluster had been created prior to debug functionality being added. Since multiple ports are being added to the scdf-server Kubernetes Service Object, each needs to have their own name.

Now add the debug port:

$ kubectl patch service scdf-server -p "$(cat src/kubernetes/server/server-svc-debug.yaml)"

To verify the mapping:

$ kubectl describe service scdf-server
Name:                     scdf-server
...
...
Port:                     scdf-server-jdwp  5005/TCP
TargetPort:               5005/TCP
NodePort:                 scdf-server-jdwp  31339/TCP
...
...
Port:                     scdf-server  80/TCP
TargetPort:               80/TCP
NodePort:                 scdf-server  30883/TCP
...
...

In the output, its shown that that container port 5005 has been mapped to the NodePort of 31339. To get the IP address of the Minikube node enter the following:

$ minikube ip
192.168.99.100

With this information a debug connection can now be created using a host of 192.168.99.100 and a port of 31339.

To disable JDWP, the following commands can be used:

$ kubectl rollout undo deployment/scdf-server
$ kubectl patch service scdf-server --type json -p='[{"op": "remove", "path": "/spec/ports/0"}]'

The Kubernetes Deployment Object is rolled back to its state prior before being patched. The Kubernetes Service Object is then patched with a remove operation to remove port 5005 from the containerPorts list.

kubectl rollout undo will force the pod to restart. Patching the Kubernetes Service Object will not recreate the service and the port mapping to the scdf-server deployment will remain the same.

See Rolling Back a Deployment for more information on deployment rollbacks including managing history and Update API Objects in Place Using kubectl Patch for more information on patching.

Shell

This section covers the options for starting the shell and more advanced functionality relating to how the shell handles white spaces, quotes, and interpretation of SpEL expressions. The introductory chapters to the Stream DSL and Composed Task DSL are good places to start for the most common usage of shell commands.

22. Shell Options

The shell is built upon the Spring Shell project. There are command line options generic to Spring Shell and some specific to Data Flow. The shell takes the following command line options

unix:>java -jar spring-cloud-dataflow-shell-1.6.0.RELEASE.jar --help
Data Flow Options:
  --dataflow.uri=                              Address of the Data Flow Server [default: http://localhost:9393].
  --dataflow.username=                        Username of the Data Flow Server [no default].
  --dataflow.password=                    Password of the Data Flow Server [no default].
  --dataflow.credentials-provider-command= Executes an external command which must return an
                                                    OAuth Bearer Token (Access Token prefixed with 'Bearer '),
                                                    e.g. 'Bearer 12345'), [no default].
  --dataflow.skip-ssl-validation=       Accept any SSL certificate (even self-signed) [default: no].
  --dataflow.proxy.uri=                  Address of an optional proxy server to use [no default].
  --dataflow.proxy.username=        Username of the proxy server (if required by proxy server) [no default].
  --dataflow.proxy.password=        Password of the proxy server (if required by proxy server) [no default].
  --spring.shell.historySize=                 Default size of the shell log file [default: 3000].
  --spring.shell.commandFile=                 Data Flow Shell executes commands read from the file(s) and then exits.
  --help                                            This message.

The spring.shell.commandFile option can be used to point to an existing file that contains all the shell commands to deploy one or many related streams and tasks. This is useful when creating some scripts to help automate deployment.

Also, the following shell command helps to modularize a complex script into multiple independent files:

dataflow:>script --file <YOUR_AWESOME_SCRIPT>

23. Listing Available Commands

Typing help at the command prompt gives a listing of all available commands. Most of the commands are for Data Flow functionality, but a few are general purpose.

! - Allows execution of operating system (OS) commands
clear - Clears the console
cls - Clears the console
date - Displays the local date and time
exit - Exits the shell
http get - Make GET request to http endpoint
http post - POST data to http endpoint
quit - Exits the shell
system properties - Shows the shell's properties
version - Displays shell version

Adding the name of the command to help shows additional information on how to invoke the command.

dataflow:>help stream create
Keyword:                   stream create
Description:               Create a new stream definition
 Keyword:                  ** default **
 Keyword:                  name
   Help:                   the name to give to the stream
   Mandatory:              true
   Default if specified:   '__NULL__'
   Default if unspecified: '__NULL__'

 Keyword:                  definition
   Help:                   a stream definition, using the DSL (e.g. "http --port=9000 | hdfs")
   Mandatory:              true
   Default if specified:   '__NULL__'
   Default if unspecified: '__NULL__'

 Keyword:                  deploy
   Help:                   whether to deploy the stream immediately
   Mandatory:              false
   Default if specified:   'true'
   Default if unspecified: 'false'

24. Tab Completion

The shell command options can be completed in the shell by pressing the TAB key after the leading --. For example, pressing TAB after stream create -- results in

dataflow:>stream create --
stream create --definition    stream create --name

If you type --de and then hit tab, --definition will be expanded.

Tab completion is also available inside the stream or composed task DSL expression for application or task properties. You can also use TAB to get hints in a stream DSL expression for what available sources, processors, or sinks can be used.

25. White Space and Quoting Rules

It is only necessary to quote parameter values if they contain spaces or the | character. The following example passes a SpEL expression (which is applied to any data it encounters) to a transform processor:

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

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

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

25.1. Quotes and Escaping

There is a Spring Shell-based client that talks to the Data Flow Server and is responsible for parsing the DSL. In turn, applications may have applications properties that rely on embedded languages, such as the Spring Expression Language.

The shell, Data Flow DSL parser, and SpEL have rules about how they handle quotes and how syntax escaping works. When combined together, confusion may arise. This section explains the rules that apply and provides examples of the most complicated situations you may encounter when all three components are involved.

It’s not always that complicated

If you do not use the Data Flow shell (for example, you use the REST API directly) or if application properties are not SpEL expressions, then the escaping rules are simpler.

25.1.1. Shell rules

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 (--something) and corresponding values. There is a special, keyless mapping, though, which is described later.

  • A value cannot 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).

  • Values passed inside deployment properties (e.g. deployment <stream-name> --properties " …​") should not be quoted again.

  • 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.

  • The keyless mapping is handled in a special way such that it does not need quoting to contain spaces.

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

dataflow:>! rm something

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

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

dataflow:>stream destroy something
dataflow:>stream destroy --name something
dataflow:>stream destroy "something"
dataflow:>stream destroy --name "something"

25.1.2. Property files rules

Rules are relaxed when loading the properties from files. * The special characters used in property files (both Java and YAML) needs to be escaped. For example \ should be replaced by \\, '\t` by \\t and so forth. * For Java property files (--propertiesFile <FILE_PATH>.properties) the property values should not be surrounded by quotes! It is not needed even if they contain spaces.

filter.expression=payload > 5
  • For YAML property files (--propertiesFile <FILE_PATH>.yaml), though, the values need to be surrounded by double quotes.

app:
    filter:
        filter:
            expression: "payload > 5"

25.1.3. DSL Parsing Rules

At the parser level (that is, inside the body of a stream or task definition) the rules are as follows:

  • 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 application 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, imagine that we want to test against string messages. If we want to compare the payload to the SpEL literal string, "something", we could use the following:

filter --expression=payload=='something'           (1)
filter --expression='payload == ''something'''     (2)
filter --expression='payload == "something"'       (3)
1 This works because there are no spaces. It is not very legible, though.
2 This uses single quotes to protect the whole argument. Hence, the actual single quotes need to be doubled.
3 SpEL recognizes String literals with either single or double quotes, so this last method is arguably the most readable.

Please note that the preceding examples are to be considered outside of the shell (for example, when calling the REST API directly). When entered inside the shell, chances are that the whole stream definition is itself inside double quotes, which would need to be escaped. The whole example then becomes the following:

dataflow:>stream create something --definition "http | filter --expression=payload='something' | log"

dataflow:>stream create something --definition "http | filter --expression='payload == ''something''' | log"

dataflow:>stream create something --definition "http | filter --expression='payload == \"something\"' | log"

25.1.4. SpEL Syntax and SpEL Literals

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

  • 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 the reverse is also true.

As a last example, assume you want to use the transform processor. This processor 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 statements are equivalent:

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

However, they are different from the following (and variations upon them):

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

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

25.1.5. Putting It All Together

As a last, complete example, consider how one could force the transformation of all messages to the string literal, hello world, by creating a stream in the context of the Data Flow shell:

dataflow:>stream create something --definition "http | transform --expression='''hello world''' | log" (1)

dataflow:>stream create something --definition "http | transform --expression='\"hello world\"' | log" (2)

dataflow:>stream create something --definition "http | transform --expression=\"'hello world'\" | log" (2)
1 In the first line, there are single quotes around the string (at the Data Flow parser level), but they need to be doubled because they are inside a string literal (started by the first single quote after the equals sign).
2 The second and third lines, use single and double quotes respectively to encompass the whole string at the Data Flow parser level. Consequently, 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. Consequently, double quotes are escaped (at the shell level)

Streams

This section goes into more detail about how you can create Streams, which are collections of Spring Cloud Stream applications. It covers topics such as creating and deploying Streams.

If you are just starting out with Spring Cloud Data Flow, you should probably read the Getting Started guide before diving into this section.

26. Introduction

A Stream is are a collection of long-lived Spring Cloud Stream applications that communicate with each other over messaging middleware. A text-based DSL defines the configuration and data flow between the applications. While many applications are provided for you to implement common use-cases, you typically create a custom Spring Cloud Stream application to implement custom business logic.

The general lifecycle of a Stream is:

  1. Register applications.

  2. Create a Stream Definition.

  3. Deploy the Stream.

  4. Undeploy or Destroy the Stream.

  5. Upgrade or Rollack applications in the Stream.

If you use Skipper, you can upgrade or rollback applications in the Stream.

There are two options for deploying streams:

  • Use a Data Flow Server implementation that deploys to a single platform.

  • Configure the Data Flow Server to delegate the deployment to a new server in the Spring Cloud ecosystem named Skipper.

When using the first option, you can use the Local Data Flow Server to deploy streams to your local machine, the Data Flow Server for Cloud Foundry to deploy streams to a single org and space on Cloud Foundry. Similarly, you can use Data Flow Server for Kuberenetes to deploy a stream to a single namespace on a Kubernetes cluster. See the Spring Cloud Data Flow project page for a list of Data Flow server implementations.

When using the second option, you can configure Skipper to deploy applications to one or more Cloud Foundry orgs and spaces, one or more namespaces on a Kubernetes cluster, or to the local machine. When deploying a stream in Data Flow using Skipper, you can specify which platform to use at deployment time. Skipper also provides Data Flow with the ability to perform updates to deployed streams. There are many ways the applications in a stream can be updated, but one of the most common examples is to upgrade a processor application with new custom business logic while leaving the existing source and sink applications alone.

26.1. Stream Pipeline DSL

A stream is defined by using a unix-inspired Pipeline syntax. The syntax uses vertical bars, also known as “pipes” to connect multiple commands. The command ls -l | grep key | less in Unix takes the output of the ls -l process and pipes it to the input of the grep key process. The output of grep in turn is sent to the input of the less process. Each | symbol connects the standard output of the command on the left to the standard input of the command on the right. Data flows through the pipeline from left to right.

In Data Flow, the Unix command is replaced by a Spring Cloud Stream application and each pipe symbol represents connecting the input and output of applications over messaging middleware, such as RabbitMQ or Apache Kafka.

Each Spring Cloud Stream application is registered under a simple name. The registration process specifies where the application can be obtained (for example, in a Maven Repository or a Docker registry). You can find out more information on how to register Spring Cloud Stream applications in this section. In Data Flow, we classify the Spring Cloud Stream applications as Sources, Processors, or Sinks.

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 would be expressed as:

http | filter | transform | file

Stream definitions can be created by using the shell’s stream create command, as shown in the following example:

dataflow:> stream create --name httpIngest --definition "http | file"

The Stream DSL is passed in to the --definition command option.

The deployment of stream definitions is done through the shell’s stream deploy command.

dataflow:> stream deploy --name ticktock

The Getting Started section shows you how to start the server and how to start and use the Spring Cloud Data Flow shell.

Note that the shell calls the Data Flow Servers' REST API. For more information on making HTTP requests directly to the server, consult the REST API Guide.

26.2. Application properties

Each application takes properties to customize its behavior. As an example, the http source module exposes a port setting that allows the data ingestion port to be changed from the default value.

dataflow:> stream create --definition "http --port=8090 | log" --name myhttpstream

This port property is actually the same as the standard Spring Boot server.port property. Data Flow adds the ability to use the shorthand form port instead of server.port. One may also specify the longhand version as well, as shown in the following example:

dataflow:> stream create --definition "http --server.port=8000 | log" --name myhttpstream

This shorthand behavior is discussed more in the section on Whitelisting application properties. If you have registered application property metadata you can use tab completion in the shell after typing -- to get a list of candidate property names.

The shell provides tab completion for application properties. The shell command app info --name <appName> --type <appType> provides additional documentation for all the supported properties.

Supported Stream <appType> possibilities are: source, processor, and sink.

27. Stream Lifecycle

The lifecycle of a stream, in "classic" mode, goes through the following stages:

27.1. Register a Stream App

You can register a Stream App with the App Registry by using the Spring Cloud Data Flow Shell app register command. You must provide a unique name, an application type, and a URI that can be resolved to the app artifact. For the type, specify source, processor, or sink. Here are a few examples:

dataflow:>app register --name mysource --type source --uri maven://com.example:mysource:0.0.1-SNAPSHOT

dataflow:>app register --name myprocessor --type processor --uri file:///Users/example/myprocessor-1.2.3.jar

dataflow:>app register --name mysink --type sink --uri http://example.com/mysink-2.0.1.jar

When providing a URI with the maven scheme, the format should conform to the following:

maven://<groupId>:<artifactId>[:<extension>[:<classifier>]]:<version>

For example, if you would like to register the snapshot versions of the http and log applications built with the RabbitMQ binder, you could do the following:

dataflow:>app register --name http --type source --uri maven://org.springframework.cloud.stream.app:http-source-rabbit:1.2.1.BUILD-SNAPSHOT
dataflow:>app register --name log --type sink --uri maven://org.springframework.cloud.stream.app:log-sink-rabbit:1.2.1.BUILD-SNAPSHOT

If you would like to register multiple apps at one time, you can store them in a properties file where the keys are formatted as <type>.<name> and the values are the URIs.

For example, if you would like to register the snapshot versions of the http and log applications built with the RabbitMQ binder, you could have the following in a properties file (for example, stream-apps.properties):

source.http=maven://org.springframework.cloud.stream.app:http-source-rabbit:1.2.1.BUILD-SNAPSHOT
sink.log=maven://org.springframework.cloud.stream.app:log-sink-rabbit:1.2.1.BUILD-SNAPSHOT

Then to import the apps in bulk, use the app import command and provide the location of the properties file with the --uri switch, as follows:

dataflow:>app import --uri file:///<YOUR_FILE_LOCATION>/stream-apps.properties

27.1.1. Register Supported Applications and Tasks

For convenience, we have the static files with application-URIs (for both maven and docker) available for all the out-of-the-box stream and task/batch app-starters. You can point to this file and import all the application-URIs in bulk. Otherwise, as explained previously, you can register them individually or have your own custom property file with only the required application-URIs in it. It is recommended, however, to have a “focused” list of desired application-URIs in a custom property file.

The following table lists the bit.ly links to the available Stream Application Starters based on Spring Boot 1.5.x:

Artifact Type Stable Release SNAPSHOT Release

RabbitMQ + Maven

bit.ly/Celsius-SR3-stream-applications-rabbit-maven

bit.ly/Celsius-BUILD-SNAPSHOT-stream-applications-rabbit-maven

RabbitMQ + Docker

bit.ly/Celsius-SR3-stream-applications-rabbit-docker

bit.ly/Celsius-BUILD-SNAPSHOT-stream-applications-rabbit-docker

Kafka 0.10 + Maven

bit.ly/Celsius-SR3-stream-applications-kafka-10-maven

bit.ly/Celsius-BUILD-SNAPSHOT-stream-applications-kafka-10-maven

Kafka 0.10 + Docker

bit.ly/Celsius-SR3-stream-applications-kafka-10-docker

bit.ly/Celsius-BUILD-SNAPSHOT-stream-applications-kafka-10-docker

The following table lists the bit.ly links to the available Stream Application Starters based on Spring Boot 2.0.x:

App Starter actuator endpoints are secured by default. You can disable security by deploying streams with the property app.*.spring.autoconfigure.exclude=org.springframework.boot.autoconfigure.security.servlet.SecurityAutoConfiguration. On Kubernetes refer to the section Liveness and readiness probes to configure security for actuator endpoints.
Artifact Type Stable Release SNAPSHOT Release

RabbitMQ + Maven

bit.ly/Darwin-GA-stream-applications-rabbit-maven

bit.ly/Darwin-BUILD-SNAPSHOT-stream-applications-rabbit-maven

RabbitMQ + Docker

bit.ly/Darwin-GA-stream-applications-rabbit-docker

bit.ly/Darwin-BUILD-SNAPSHOT-stream-applications-rabbit-docker

Kafka 0.11 and above + Maven

bit.ly/Darwin-GA-stream-applications-kafka-10-maven

bit.ly/Darwin-BUILD-SNAPSHOT-stream-applications-kafka-10-maven

Kafka 0.11 and above + Docker

bit.ly/Darwin-GA-stream-applications-kafka-10-docker

bit.ly/Darwin-BUILD-SNAPSHOT-stream-applications-kafka-10-docker

The following table lists the available Task Application Starters:

Artifact Type Stable Release SNAPSHOT Release

Maven

bit.ly/Clark-GA-task-applications-maven

bit.ly/Clark-BUILD-SNAPSHOT-task-applications-maven

Docker

bit.ly/Clark-GA-task-applications-docker

bit.ly/Clark-BUILD-SNAPSHOT-task-applications-docker

You can find more information about the available task starters in the Task App Starters Project Page and related reference documentation. For more information about the available stream starters, look at the Stream App Starters Project Page and related reference documentation.

As an example, if you would like to register all out-of-the-box stream applications built with the Kafka binder in bulk, you can use the following command:

$ dataflow:>app import --uri http://bit.ly/Darwin-GA-stream-applications-kafka-10-maven

Alternatively you can register all the stream applications with the Rabbit binder, as follows:

$ dataflow:>app import --uri http://bit.ly/Darwin-GA-stream-applications-rabbit-maven

You can also pass the --local option (which is true by default) to indicate whether the properties file location should be resolved within the shell process itself. If the location should be resolved from the Data Flow Server process, specify --local false.

When using either app register or app import, if an app is already registered with the provided name and type, it is not overridden by default. If you would like to override the pre-existing app coordinates, then include the --force option.

Note, however, that, once downloaded, applications may be cached locally on the Data Flow server, based on the resource location. If the resource location does not change (even though the actual resource bytes may be different), then it is not re-downloaded. When using maven:// resources on the other hand, using a constant location may still circumvent caching (if using -SNAPSHOT versions).

Moreover, if a stream is already deployed and using some version of a registered app, then (forcibly) re-registering a different app has no effect until the stream is deployed again.

In some cases, the Resource is resolved on the server side. In others, the URI is passed to a runtime container instance where it is resolved. Consult the specific documentation of each Data Flow Server for more detail.

27.1.2. Whitelisting application properties

Stream and Task applications are Spring Boot applications that are aware of many Common Application Properties, such as server.port but also families of properties such as those with the prefix spring.jmx and logging. When creating your own application, you should whitelist properties so that the shell and the UI can display them first as primary properties when presenting options through TAB completion or in drop-down boxes.

To whitelist application properties, create a file named spring-configuration-metadata-whitelist.properties in the META-INF resource directory. There are two property keys that can be used inside this file. The first key is named configuration-properties.classes. The value is a comma separated list of fully qualified @ConfigurationProperty class names. The second key is configuration-properties.names, whose value is a comma-separated list of property names. This can contain the full name of the property, such as server.port, or a partial name to whitelist a category of property names, such as spring.jmx.

The Spring Cloud Stream application starters are a good place to look for examples of usage. The following example comes from the file sink’s spring-configuration-metadata-whitelist.properties file:

configuration-properties.classes=org.springframework.cloud.stream.app.file.sink.FileSinkProperties

If we also want to add server.port to be white listed, it would become the following line:

configuration-properties.classes=org.springframework.cloud.stream.app.file.sink.FileSinkProperties
configuration-properties.names=server.port

Make sure to add 'spring-boot-configuration-processor' as an optional dependency to generate configuration metadata file for the properties.

<dependency>
    <groupId>org.springframework.boot</groupId>
    <artifactId>spring-boot-configuration-processor</artifactId>
    <optional>true</optional>
</dependency>

27.1.3. Creating and Using a Dedicated Metadata Artifact

You can go a step further in the process of describing the main properties that your stream or task app supports by creating a metadata companion artifact. This jar file contains only the Spring boot JSON file about configuration properties metadata and the whitelisting file described in the previous section.

The following example shows the contents of such an artifact, for the canonical log sink:

$ jar tvf log-sink-rabbit-1.2.1.BUILD-SNAPSHOT-metadata.jar
373848 META-INF/spring-configuration-metadata.json
   174 META-INF/spring-configuration-metadata-whitelist.properties

Note that the spring-configuration-metadata.json file is quite large. This is because it contains the concatenation of all the properties that are available at runtime to the log sink (some of them come from spring-boot-actuator.jar, some of them come from spring-boot-autoconfigure.jar, some more from spring-cloud-starter-stream-sink-log.jar, and so on). Data Flow always relies on all those properties, even when a companion artifact is not available, but here all have been merged into a single file.

To help with that (you do not want to try to craft this giant JSON file by hand), you can use the following plugin in your build:

<plugin>
 	<groupId>org.springframework.cloud</groupId>
 	<artifactId>spring-cloud-app-starter-metadata-maven-plugin</artifactId>
 	<executions>
 		<execution>
 			<id>aggregate-metadata</id>
 			<phase>compile</phase>
 			<goals>
 				<goal>aggregate-metadata</goal>
 			</goals>
 		</execution>
 	</executions>
 </plugin>
This plugin comes in addition to the spring-boot-configuration-processor that creates the individual JSON files. Be sure to configure both.

The benefits of a companion artifact include:

  • Being much lighter. (The companion artifact is usually a few kilobytes, as opposed to megabytes for the actual app.) Consequently, they are quicker to download, allowing quicker feedback when using, for example, app info or the Dashboard UI.

  • As a consequence of being lighter, they can be used in resource constrained environments (such as PaaS) when metadata is the only piece of information needed.

  • For environments that do not deal with Spring Boot uber jars directly (for example, Docker-based runtimes such as Kubernetes or Mesos), this is the only way to provide metadata about the properties supported by the app.

Remember, though, that this is entirely optional when dealing with uber jars. The uber jar itself also includes the metadata in it already.

27.1.4. Using the Companion Artifact

Once you have a companion artifact at hand, you need to make the system aware of it so that it can be used.

When registering a single app with app register, you can use the optional --metadata-uri option in the shell, as follows:

dataflow:>app register --name log --type sink
    --uri maven://org.springframework.cloud.stream.app:log-sink-kafka-10:1.2.1.BUILD-SNAPSHOT
    --metadata-uri=maven://org.springframework.cloud.stream.app:log-sink-kafka-10:jar:metadata:1.2.1.BUILD-SNAPSHOT

When registering several files by using the app import command, the file should contain a <type>.<name>.metadata line in addition to each <type>.<name> line. Strictly speaking, doing so is optional (if some apps have it but some others do not, it works), but it is best practice.

The following example shows a Dockerized app, where the metadata artifact is being hosted in a Maven repository (retrieving it through http:// or file:// would be equally possible).

...
source.http=docker:springcloudstream/http-source-rabbit:latest
source.http.metadata=maven://org.springframework.cloud.stream.app:http-source-rabbit:jar:metadata:1.2.1.BUILD-SNAPSHOT
...

27.1.5. Creating Custom Applications

While there are out-of-the-box source, processor, sink applications available, you can extend these applications or write a custom Spring Cloud Stream application.

The process of creating Spring Cloud Stream applications with Spring Initializr is detailed in the Spring Cloud Stream {spring-cloud-stream-docs}#_getting_started[documentation]. It is possible to include multiple binders to an application. If doing so, see the instructions in Passing Spring Cloud Stream properties for how to configure them.

For supporting property whitelisting, Spring Cloud Stream applications running in Spring Cloud Data Flow may include the Spring Boot configuration-processor as an optional dependency, as shown in the following example:

<dependencies>
  <!-- other dependencies -->
  <dependency>
    <groupId>org.springframework.boot</groupId>
    <artifactId>spring-boot-configuration-processor</artifactId>
    <optional>true</optional>
  </dependency>
</dependencies>

Make sure that the spring-boot-maven-plugin is included in the POM. The plugin is necessary for creating the executable jar that is registered with Spring Cloud Data Flow. Spring Initialzr includes the plugin in the generated POM.

Once a custom application has been created, it can be registered as described in Register a Stream App.

27.2. Creating a Stream

The Spring Cloud Data Flow Server exposes a full RESTful API for managing the lifecycle of stream definitions, but the easiest way to use is it is through the Spring Cloud Data Flow shell. Start the shell as described in the Getting Started section.

New streams are created with the help of stream definitions. The definitions are built from a simple DSL. For example, consider what happens if we execute the following shell command:

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

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

27.2.1. Application Properties

Application properties are the properties associated with each application in the stream. When the application is deployed, the application properties are applied to the application through command line arguments or environment variables, depending on the underlying deployment implementation.

The following stream can have application properties defined at the time of stream creation:

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

The shell command app info --name <appName> --type <appType> displays the white-listed application properties for the application. For more info on the property white listing, refer to Whitelisting application properties

The following listing shows the white_listed properties for the time app:

dataflow:> app info --name time --type source
╔══════════════════════════════╤══════════════════════════════╤══════════════════════════════╤══════════════════════════════╗
║         Option Name          │         Description          │           Default            │             Type             ║
╠══════════════════════════════╪══════════════════════════════╪══════════════════════════════╪══════════════════════════════╣
║trigger.time-unit             │The TimeUnit to apply to delay│<none>                        │java.util.concurrent.TimeUnit ║
║                              │values.                       │                              │                              ║
║trigger.fixed-delay           │Fixed delay for periodic      │1                             │java.lang.Integer             ║
║                              │triggers.                     │                              │                              ║
║trigger.cron                  │Cron expression value for the │<none>                        │java.lang.String              ║
║                              │Cron Trigger.                 │                              │                              ║
║trigger.initial-delay         │Initial delay for periodic    │0                             │java.lang.Integer             ║
║                              │triggers.                     │                              │                              ║
║trigger.max-messages          │Maximum messages per poll, -1 │1                             │java.lang.Long                ║
║                              │means infinity.               │                              │                              ║
║trigger.date-format           │Format for the date value.    │<none>                        │java.lang.String              ║
╚══════════════════════════════╧══════════════════════════════╧══════════════════════════════╧══════════════════════════════╝

The following listing shows the white-listed properties for the log app:

dataflow:> app info --name log --type sink
╔══════════════════════════════╤══════════════════════════════╤══════════════════════════════╤══════════════════════════════╗
║         Option Name          │         Description          │           Default            │             Type             ║
╠══════════════════════════════╪══════════════════════════════╪══════════════════════════════╪══════════════════════════════╣
║log.name                      │The name of the logger to use.│<none>                        │java.lang.String              ║
║log.level                     │The level at which to log     │<none>                        │org.springframework.integratio║
║                              │messages.                     │                              │n.handler.LoggingHandler$Level║
║log.expression                │A SpEL expression (against the│payload                       │java.lang.String              ║
║                              │incoming message) to evaluate │                              │                              ║
║                              │as the logged message.        │                              │                              ║
╚══════════════════════════════╧══════════════════════════════╧══════════════════════════════╧══════════════════════════════╝

The application properties for the time and log apps can be specified at the time of stream creation as follows:

dataflow:> stream create --definition "time --fixed-delay=5 | log --level=WARN" --name ticktock

Note that, in the preceding example, the fixed-delay and level properties defined for the apps time and log are the "'short-form'" property names provided by the shell completion. These "'short-form'" property names are applicable only for the white-listed properties. In all other cases, only fully qualified property names should be used.

27.2.2. Common Application Properties

In addition to configuration through DSL, Spring Cloud Data Flow provides a mechanism for setting common properties to all the streaming applications that are launched by it. This can be done by adding properties prefixed with spring.cloud.dataflow.applicationProperties.stream when starting the server. When doing so, the server passes all the properties, without the prefix, to the instances it launches.

For example, all the launched applications can be configured to use a specific Kafka broker by launching the Data Flow server with the following options:

--spring.cloud.dataflow.applicationProperties.stream.spring.cloud.stream.kafka.binder.brokers=192.168.1.100:9092
--spring.cloud.dataflow.applicationProperties.stream.spring.cloud.stream.kafka.binder.zkNodes=192.168.1.100:2181

Doing so causes the properties spring.cloud.stream.kafka.binder.brokers and spring.cloud.stream.kafka.binder.zkNodes to be passed to all the launched applications.

Properties configured with this mechanism have lower precedence than stream deployment properties. They are overridden if a property with the same key is specified at stream deployment time (for example, app.http.spring.cloud.stream.kafka.binder.brokers overrides the common property).

27.3. Deploying a Stream

This section describes how to deploy a Stream when the Spring Cloud Data Flow server is responsible for deploying the stream. The following section, Stream Lifecycle with Skipper, covers the new deployment and upgrade features when the Spring Cloud Data Flow server delegates to Skipper for stream deployment. The description of how deployment properties applies to both approaches of Stream deployment.

Give the ticktock stream definition:

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

To deploy the stream, use the following shell command:

dataflow:> stream deploy --name ticktock

The Data Flow Server resolves time and log to maven coordinates and uses those to launch the time and log applications of the stream, as shown in the following listing:

2016-06-01 09:41:21.728  INFO 79016 --- [nio-9393-exec-6] o.s.c.d.spi.local.LocalAppDeployer       : deploying app ticktock.log instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/ticktock-1464788481708/ticktock.log
2016-06-01 09:41:21.914  INFO 79016 --- [nio-9393-exec-6] o.s.c.d.spi.local.LocalAppDeployer       : deploying app ticktock.time instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/ticktock-1464788481910/ticktock.time

In the preceding example, the time source sends the current time as a message each second, and the log sink outputs it by using the logging framework. You can tail the stdout log (which has an <instance> suffix). The log files are located within the directory displayed in the Data Flow Server’s log output, as shown in the following listing:

$ tail -f /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/ticktock-1464788481708/ticktock.log/stdout_0.log
2016-06-01 09:45:11.250  INFO 79194 --- [  kafka-binder-] log.sink    : 06/01/16 09:45:11
2016-06-01 09:45:12.250  INFO 79194 --- [  kafka-binder-] log.sink    : 06/01/16 09:45:12
2016-06-01 09:45:13.251  INFO 79194 --- [  kafka-binder-] log.sink    : 06/01/16 09:45:13

You can also create and deploy the stream in one step by passing the --deploy flag when creating the stream, as follows:

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

However, it is not very common in real-world use cases to create and deploy the stream in one step. The reason is that when you use the stream deploy command, you can pass in properties that define how to map the applications onto the platform (for example, what is the memory size of the container to use, the number of each application to run, and whether to enable data partitioning features). Properties can also override application properties that were set when creating the stream. The next sections cover this feature in detail.

27.3.1. Deployment Properties

When deploying a stream, you can specify properties that fall into two groups:

  • Properties that control how the apps are deployed to the target platform. These properties use a deployer prefix and are referred to as deployer properties.

  • Properties that set application properties or override application properties set during stream creation and are referred to as application properties.

The syntax for deployer properties is deployer.<app-name>.<short-property-name>=<value>, and the syntax for application properties app.<app-name>.<property-name>=<value>. This syntax is used when passing deployment properties through the shell. You may also specify them in a YAML file, which is discussed later in this chapter.

The following table shows the difference in behavior between setting deployer and application properties when deploying an application.

Application Properties Deployer Properties

Example Syntax

app.filter.expression=something

deployer.filter.count=3

What the application "sees"

expression=something or, if expression is one of the whitelisted properties, <some-prefix>.expression=something

Nothing

What the deployer "sees"

Nothing

spring.cloud.deployer.count=3. The spring.cloud.deployer prefix is automatically and always prepended to the property name.

Typical usage

Passing/Overriding application properties, passing Spring Cloud Stream binder or partitioning properties

Setting the number of instances, memory, disk, and others

Passing Instance Count

If you would like to have multiple instances of an application in the stream, you can include a deployer property called count with the deploy command:

dataflow:> stream deploy --name ticktock --properties "deployer.time.count=3"

Note that count is the reserved property name used by the underlying deployer. Consequently, if the application also has a custom property named count, it is not supported when specified in 'short-form' form during stream deployment as it could conflict with the instance count deployer property. Instead, the count as a custom application property can be specified in its fully qualified form (for example, app.something.somethingelse.count) during stream deployment or it can be specified by using the 'short-form' or the fully qualified form during the stream creation, where it is processed as an app property.

Inline Versus File-based Properties

When using the Spring Cloud Data Flow Shell, there are two ways to provide deployment properties: either inline or through a file reference. Those two ways are exclusive.

Inline properties use the --properties shell option and list properties as a comma separated list of key=value pairs, as shown in the following example:

stream deploy foo
    --properties "deployer.transform.count=2,app.transform.producer.partitionKeyExpression=payload"

File references use the --propertiesFile option and point it to a local .properties, .yaml or .yml file (that is, a file that resides 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, and others), although we recommend using = as a key-value pair delimiter, for consistency. The following example shows a stream deploy command that uses the --propertiesFile option:

stream deploy something --propertiesFile myprops.properties

Assume that myprops.properties contains the following properties:

deployer.transform.count=2
app.transform.producer.partitionKeyExpression=payload

Both of the properties are passed as deployment properties for the something stream.

If you use YAML as the format for the deployment properties, use the .yaml or .yml file extention when deploying the stream, as shown in the following example:

stream deploy foo --propertiesFile myprops.yaml

In that case, the myprops.yaml file might contain the following content:

deployer:
  transform:
    count: 2
app:
  transform:
    producer:
      partitionKeyExpression: payload
Passing application properties

The application properties can also be specified when deploying a stream. When specified during deployment, these application properties can either be specified as 'short-form' property names (applicable for white-listed properties) or as fully qualified property names. The application properties should have the prefix app.<appName/label>.

For example, consider the following stream command:

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

The stream in the precedig example can also be deployed with application properties by using the 'short-form' property names, as shown in the following example:

dataflow:>stream deploy ticktock --properties "app.time.fixed-delay=5,app.log.level=ERROR"

Consider the following example:

stream create ticktock --definition "a: time | b: log"

When using the app label, the application properties can be defined as follows:

stream deploy ticktock --properties "app.a.fixed-delay=4,app.b.level=ERROR"
Passing Spring Cloud Stream properties

Spring Cloud Data Flow sets the required Spring Cloud Stream properties for the applications inside the stream. Most importantly, the spring.cloud.stream.bindings.<input/output>.destination is set internally for the apps to bind.

If you want to override any of the Spring Cloud Stream properties, they can be set with deployment properties.

For example, consider the following stream definition:

dataflow:> stream create --definition "http | transform --expression=payload.getValue('hello').toUpperCase() | log" --name ticktock

If there are multiple binders available in the classpath for each of the applications and the binder is chosen for each deployment, then the stream can be deployed with the specific Spring Cloud Stream properties, as follows:

dataflow:>stream deploy ticktock --properties "app.time.spring.cloud.stream.bindings.output.binder=kafka,app.transform.spring.cloud.stream.bindings.input.binder=kafka,app.transform.spring.cloud.stream.bindings.output.binder=rabbit,app.log.spring.cloud.stream.bindings.input.binder=rabbit"
Overriding the destination names is not recommended, because Spring Cloud Data Flow internally takes care of setting this property.
Passing Per-binding Producer and Consumer Properties

A Spring Cloud Stream application can have producer and consumer properties set on a per-binding basis. While Spring Cloud Data Flow supports specifying short-hand notation for per-binding producer properties such as partitionKeyExpression and partitionKeyExtractorClass (as described in Passing Stream Partition Properties), all the supported Spring Cloud Stream producer/consumer properties can be set as Spring Cloud Stream properties for the app directly as well.

The consumer properties can be set for the inbound channel name with the prefix app.[app/label name].spring.cloud.stream.bindings.<channelName>.consumer.. The producer properties can be set for the outbound channel name with the prefix app.[app/label name].spring.cloud.stream.bindings.<channelName>.producer.. Consider the following example:

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

The stream can be deployed with producer and consumer properties, as follows:

dataflow:>stream deploy ticktock --properties "app.time.spring.cloud.stream.bindings.output.producer.requiredGroups=myGroup,app.time.spring.cloud.stream.bindings.output.producer.headerMode=raw,app.log.spring.cloud.stream.bindings.input.consumer.concurrency=3,app.log.spring.cloud.stream.bindings.input.consumer.maxAttempts=5"

The binder-specific producer and consumer properties can also be specified in a similar way, as shown in the following example:

dataflow:>stream deploy ticktock --properties "app.time.spring.cloud.stream.rabbit.bindings.output.producer.autoBindDlq=true,app.log.spring.cloud.stream.rabbit.bindings.input.consumer.transacted=true"
Passing Stream Partition Properties

A common pattern in stream processing is to partition the data as it is streamed. This entails deploying multiple instances of a message-consuming app and using content-based routing so that messages with a given key (as determined at runtime) are always routed to the same app instance. You can pass the partition properties during stream deployment to declaratively configure a partitioning strategy to route each message to a specific consumer instance.

The following list shows variations of deploying partitioned streams:

  • app.[app/label name].producer.partitionKeyExtractorClass: The class name of a PartitionKeyExtractorStrategy (default: null)

  • app.[app/label name].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 app is not partitioned (default: null)

  • app.[app/label name].producer.partitionSelectorClass: The class name of a PartitionSelectorStrategy (default: null)

  • app.[app/label name].producer.partitionSelectorExpression: A SpEL expression, evaluated against the partition key, to determine the partition index to which the message is routed. The final partition index is the return value (an integer) modulo [nextModule].count. If both the class and expression are null, the underlying binder’s default PartitionSelectorStrategy is applied to the key (default: null)

In summary, an app is partitioned if its count is > 1 and the previous app has a partitionKeyExtractorClass or partitionKeyExpression (partitionKeyExtractorClass takes precedence). When a partition key is extracted, the partitioned app instance is determined by invoking the partitionSelectorClass, if present, or the partitionSelectorExpression % partitionCount. partitionCount is application count, in the case of RabbitMQ, or the underlying partition count of the topic, in the case of Kafka.

If neither a partitionSelectorClass nor a partitionSelectorExpression is present, the result is key.hashCode() % partitionCount.

Passing application content type properties

In a stream definition, you can specify that the input or the output of an application must be converted to a different type. You can use the inputType and outputType properties to specify the content type for the incoming data and outgoing data, respectively.

For example, consider the following stream:

dataflow:>stream create tuple --definition "http | filter --inputType=application/x-spring-tuple
 --expression=payload.hasFieldName('hello') | transform --expression=payload.getValue('hello').toUpperCase()
 | log" --deploy

The http app is expected to send the data in JSON and the filter app receives the JSON data and processes it as a Spring Tuple. In order to do so, we use the inputType property on the filter app to convert the data into the expected Spring Tuple format. The transform application processes the Tuple data and sends the processed data to the downstream log application.

Consider the following example of sending some data to the http application:

dataflow:>http post --data {"hello":"world","something":"somethingelse"} --contentType application/json --target localhost:<http-port>;

At the log application, you see the content as follows:

INFO 18745 --- [transform.tuple-1] log.sink : WORLD

Depending on how applications are chained, the content type conversion can be specified either as an --outputType in the upstream app or as an --inputType in the downstream app. For instance, in the above stream, instead of specifying the --inputType on the 'transform' application to convert, the option --outputType=application/x-spring-tuple can also be specified on the 'http' application.

For the complete list of message conversion and message converters, please refer to Spring Cloud Stream {spring-cloud-stream-docs}#contenttypemanagement[documentation].

Overriding Application Properties During Stream Deployment

Application properties that are defined during deployment override the same properties defined during the stream creation.

For example, the following stream has application properties defined during stream creation:

dataflow:> stream create --definition "time --fixed-delay=5 | log --level=WARN" --name ticktock

To override these application properties, you can specify the new property values during deployment, as follows:

dataflow:>stream deploy ticktock --properties "app.time.fixed-delay=4,app.log.level=ERROR"

27.4. Destroying a Stream

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

dataflow:> stream destroy --name ticktock

If the stream was deployed, it is undeployed before the stream definition is deleted.

27.5. Undeploying a Stream

Often you want to stop a stream but retain the name and definition for future use. In that case, you can undeploy the stream by name.

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

You can issue the deploy command at a later time to restart it.

dataflow:> stream deploy --name ticktock

28. Stream Lifecycle with Skipper

An additional lifecycle stage of Stream is available if you run in "skipper" mode.

  1. Upgrade or Rollback applications in the Stream. (Skipper mode)

Skipper is a server that you discover Spring Boot applications and manage their lifecycle on multiple Cloud Platforms.

Applications in Skipper are bundled as packages that contain the application’s resource location, application properties and deployment properites. You can think Skipper packages as analogous to packages found in tools such as apt-get or brew.

When Data Flow deploys a Stream, it will generate and upload a package to Skipper that represents the applications in the Stream. Subsequent commands to upgrade or rollback the applications within the Stream are passed through to Skipper. In addition, the Stream definition is reverse engineered from the package and the status of the Stream is also delegated to Skipper.

28.1. Register a Versioned Stream App

Skipper extends the Register a Stream App lifecycle with support of multi-versioned stream applications. This allows to upgrade or rollback those applications at runtime using the deployment properties.

Register a versioned stream application using the app register command. You must provide a unique name, application type, and a URI that can be resolved to the app artifact. For the type, specify "source", "processor", or "sink". The version is resolved from the URI. Here are a few examples:

dataflow:>app register --name mysource --type source --uri maven://com.example:mysource:0.0.1
dataflow:>app register --name mysource --type source --uri maven://com.example:mysource:0.0.2
dataflow:>app register --name mysource --type source --uri maven://com.example:mysource:0.0.3

dataflow:>app list --id source:mysource
╔══════════════════╤═════════╤════╤════╗
║     source       │processor│sink│task║
╠══════════════════╪═════════╪════╪════╣
║> mysource-0.0.1 <│         │    │    ║
║mysource-0.0.2    │         │    │    ║
║mysource-0.0.3    │         │    │    ║
╚══════════════════╧═════════╧════╧════╝

The application URI should conform to one the following schema formats:

  • maven schema

maven://<groupId>:<artifactId>[:<extension>[:<classifier>]]:<version>
  • http schema

http://<web-path>/<artifactName>-<version>.jar
  • file schema

file:///<local-path>/<artifactName>-<version>.jar
  • docker schema

docker:<docker-image-path>/<imageName>:<version>
The URI <version> part is compulsory for the versioned stream applications

Multiple versions can be registered for the same applications (e.g. same name and type) but only one can be set as default. The default version is used for deploying Streams.

The first time an application is registered it will be marked as default. The default application version can be altered with the app default command:

dataflow:>app default --id source:mysource --version 0.0.2
dataflow:>app list --id source:mysource
╔══════════════════╤═════════╤════╤════╗
║     source       │processor│sink│task║
╠══════════════════╪═════════╪════╪════╣
║mysource-0.0.1    │         │    │    ║
║> mysource-0.0.2 <│         │    │    ║
║mysource-0.0.3    │         │    │    ║
╚══════════════════╧═════════╧════╧════╝

The app list --id <type:name> command lists all versions for a given stream application.

The app unregister command has an optional --version parameter to specify the app version to unregister.

dataflow:>app unregister --name mysource --type source --version 0.0.1
dataflow:>app list --id source:mysource
╔══════════════════╤═════════╤════╤════╗
║     source       │processor│sink│task║
╠══════════════════╪═════════╪════╪════╣
║> mysource-0.0.2 <│         │    │    ║
║mysource-0.0.3    │         │    │    ║
╚══════════════════╧═════════╧════╧════╝

If a --version is not specified, the default version is unregistered.

All applications in a stream should have a default version set for the stream to be deployed. Otherwise they will be treated as unregistered application during the deployment. Use the app default to set the defaults.

app default --id source:mysource --version 0.0.3
dataflow:>app list --id source:mysource
╔══════════════════╤═════════╤════╤════╗
║     source       │processor│sink│task║
╠══════════════════╪═════════╪════╪════╣
║mysource-0.0.2    │         │    │    ║
║> mysource-0.0.3 <│         │    │    ║
╚══════════════════╧═════════╧════╧════╝

The stream deploy necessitates default app versions to be set. The stream update and stream rollback commands though can use all (default and non-default) registered app versions.

dataflow:>stream create foo --definition "mysource | log"

This will create stream using the default mysource version (0.0.3). Then we can update the version to 0.0.2 like this:

dataflow:>stream update foo --properties version.mysource=0.0.2

Only pre-registered applications can be used to deploy, update or rollback a Stream.

An attempt to update the mysource to version 0.0.1 (not registered) will fail!

28.2. Creating and Deploying a Stream

You create and deploy a stream by using Skipper in two steps:

  1. Creating the stream definition.

  2. Deploying the stream.

The following example shows the two steps in action:

dataflow:> stream create --name httptest --definition "http --server.port=9000 | log"
dataflow:> stream deploy --name httptest

The stream info command shows useful information about the stream, including the deployment properties, as shown (with its output) in the following example:

dataflow:>stream info httptest
╔══════════════════════════════╤══════════════════════════════╤════════════════════════════╗
║             Name             │             DSL              │          Status            ║
╠══════════════════════════════╪══════════════════════════════╪════════════════════════════╣
║httptest                      │http --server.port=9000 | log │deploying                   ║
╚══════════════════════════════╧══════════════════════════════╧════════════════════════════╝

Stream Deployment properties: {
  "log" : {
    "spring.cloud.deployer.indexed" : "true",
    "spring.cloud.deployer.group" : "httptest",
    "maven://org.springframework.cloud.stream.app:log-sink-rabbit" : "1.1.0.RELEASE"
  },
  "http" : {
    "spring.cloud.deployer.group" : "httptest",
    "maven://org.springframework.cloud.stream.app:http-source-rabbit" : "1.1.0.RELEASE"
  }
}

There is an important optional command argument (called --platformName) to the stream deploy command. Skipper can be configured to deploy to multiple platforms. Skipper is pre-configured with a platform named default, which deploys applications to the local machine where Skipper is running. The default value of the command line argument --platformName is default. If you commonly deploy to one platform, when installing Skipper, you can override the configuration of the default platform. Otherwise, specify the platformName to one of the values returned by the stream platform-list command.

28.3. Updating a Stream

To update the stream, use the command stream update which takes as a command argument either --properties or --propertiesFile. You can pass in values to these command arguments in the same format as when deploy the stream with or without Skipper. There is an important new top level prefix available when using Skipper, which is version. If the Stream http | log was deployed, and the version of log which registered at the time of deployment was 1.1.0.RELEASE, the following command will update the Stream to use the 1.2.0.RELEASE of the log application. Before updating the stream with the specific version of the app, we need to make sure that the app is registered with that version.

dataflow:>app register --name log --type sink --uri maven://org.springframework.cloud.stream.app:log-sink-rabbit:1.2.0.RELEASE
Successfully registered application 'sink:log'
dataflow:>stream update --name httptest --properties version.log=1.2.0.RELEASE

Only pre-registered application versions can be used to deploy, update, or rollback a stream.

To verify the deployment properties and the updated version, we can use stream info, as shown (with its output) in the following example:

dataflow:>stream info httptest
╔══════════════════════════════╤══════════════════════════════╤════════════════════════════╗
║             Name             │             DSL              │          Status            ║
╠══════════════════════════════╪══════════════════════════════╪════════════════════════════╣
║httptest                      │http --server.port=9000 | log │deploying                   ║
╚══════════════════════════════╧══════════════════════════════╧════════════════════════════╝

Stream Deployment properties: {
  "log" : {
    "spring.cloud.deployer.indexed" : "true",
    "spring.cloud.deployer.count" : "1",
    "spring.cloud.deployer.group" : "httptest",
    "maven://org.springframework.cloud.stream.app:log-sink-rabbit" : "1.2.0.RELEASE"
  },
  "http" : {
    "spring.cloud.deployer.group" : "httptest",
    "maven://org.springframework.cloud.stream.app:http-source-rabbit" : "1.1.0.RELEASE"
  }
}

28.4. Stream versions

Skipper keeps a history of the streams that were deployed. After updating a Stream, there will be a second version of the stream. You can query for the history of the versions using the command stream history --name <name-of-stream>.

dataflow:>stream history --name httptest
╔═══════╤════════════════════════════╤════════╤════════════╤═══════════════╤════════════════╗
║Version│        Last updated        │ Status │Package Name│Package Version│  Description   ║
╠═══════╪════════════════════════════╪════════╪════════════╪═══════════════╪════════════════╣
║2      │Mon Nov 27 22:41:16 EST 2017│DEPLOYED│httptest    │1.0.0          │Upgrade complete║
║1      │Mon Nov 27 22:40:41 EST 2017│DELETED │httptest    │1.0.0          │Delete complete ║
╚═══════╧════════════════════════════╧════════╧════════════╧═══════════════╧════════════════╝

28.5. Stream Manifests

Skipper keeps a “manifest” of the all the applications, their application properties, and their deployment properties after all values have been substituted. This represents the final state of what was deployed to the platform. You can view the manifest for any of the versions of a Stream by using the following command:

stream manifest --name <name-of-stream> --releaseVersion <optional-version>

If the --releaseVersion is not specified, the manifest for the last version is returned.

The following example shows the use of the manifest:

dataflow:>stream manifest --name httptest

Using the command results in the following output:

# Source: log.yml
apiVersion: skipper.spring.io/v1
kind: SpringCloudDeployerApplication
metadata:
  name: log
spec:
  resource: maven://org.springframework.cloud.stream.app:log-sink-rabbit
  version: 1.2.0.RELEASE
  applicationProperties:
    spring.metrics.export.triggers.application.includes: integration**
    spring.cloud.dataflow.stream.app.label: log
    spring.cloud.stream.metrics.key: httptest.log.${spring.cloud.application.guid}
    spring.cloud.stream.bindings.input.group: httptest
    spring.cloud.stream.metrics.properties: spring.application.name,spring.application.index,spring.cloud.application.*,spring.cloud.dataflow.*
    spring.cloud.dataflow.stream.name: httptest
    spring.cloud.dataflow.stream.app.type: sink
    spring.cloud.stream.bindings.input.destination: httptest.http
  deploymentProperties:
    spring.cloud.deployer.indexed: true
    spring.cloud.deployer.group: httptest
    spring.cloud.deployer.count: 1

---
# Source: http.yml
apiVersion: skipper.spring.io/v1
kind: SpringCloudDeployerApplication
metadata:
  name: http
spec:
  resource: maven://org.springframework.cloud.stream.app:http-source-rabbit
  version: 1.2.0.RELEASE
  applicationProperties:
    spring.metrics.export.triggers.application.includes: integration**
    spring.cloud.dataflow.stream.app.label: http
    spring.cloud.stream.metrics.key: httptest.http.${spring.cloud.application.guid}
    spring.cloud.stream.bindings.output.producer.requiredGroups: httptest
    spring.cloud.stream.metrics.properties: spring.application.name,spring.application.index,spring.cloud.application.*,spring.cloud.dataflow.*
    server.port: 9000
    spring.cloud.stream.bindings.output.destination: httptest.http
    spring.cloud.dataflow.stream.name: httptest
    spring.cloud.dataflow.stream.app.type: source
  deploymentProperties:
    spring.cloud.deployer.group: httptest

The majority of the deployment and application properties were set by Data Flow to enable the applications to talk to each other and to send application metrics with identifying labels.

28.6. Rollback a Stream

You can rollback to a previous version of the stream using the command stream rollback.

dataflow:>stream rollback --name httptest

The optional --releaseVersion command argument adds the version of the stream. If not specified, the rollback goes to the previous stream version.

28.7. Application Count

The application count is a dynamic property of the system. If, due to scaling at runtime, the application to be upgraded has 5 instances running, then 5 instances of the upgraded application are deployed.

28.8. Skipper’s Upgrade Strategy

Skipper has a simple 'red/black' upgrade strategy. It deploys the new version of the applications, using as many instances as the currently running version, and checks the /health endpoint of the application. If the health of the new application is good, then the previous application is undeployed. If the health of the new application is bad, then all new applications are undeployed and the upgrade is considered to be not successful.

The upgrade strategy is not a rolling upgrade, so if five applications of the application are running, then in a sunny-day scenario, five of the new applications are also running before the older version is undeployed.

29. Stream DSL

This section covers additional features of the Stream DSL not covered in the Stream DSL introduction.

29.1. Tap a Stream

Taps can be created at various producer endpoints in a stream. For a stream such as that defined in the following example, taps can be created at the output of http, step1 and step2:

stream create --definition "http | step1: transform --expression=payload.toUpperCase() | step2: transform --expression=payload+'!' | log" --name mainstream --deploy

To create a stream that acts as a 'tap' on another stream requires specifying the source destination name for the tap stream. The syntax for the source destination name is as follows:

:<streamName>.<label/appName>

To create a tap at the output of http in the preceding stream, the source destination name is mainstream.http To create a tap at the output of the first transform app in the stream above, the source destination name is mainstream.step1

The tap stream DSL resembles the following:

stream create --definition ":mainstream.http > counter" --name tap_at_http --deploy

stream create --definition ":mainstream.step1 > jdbc" --name tap_at_step1_transformer --deploy

Note the colon (:) prefix before the destination names. The colon lets the parser recognize this as a destination name instead of an app name.

29.2. Using Labels in a Stream

When a stream is made up of multiple apps with the same name, they must be qualified with labels: stream create --definition "http | firstLabel: transform --expression=payload.toUpperCase() | secondLabel: transform --expression=payload+'!' | log" --name myStreamWithLabels --deploy

29.3. Named Destinations

Instead of referencing a source or sink application, you can use a named destination. A named destination corresponds to a specific destination name in the middleware broker (Rabbit, Kafka, and others). When using the | symbol, applications are connected to each other with messaging middleware destination names created by the Data Flow server. In keeping with the Unix analogy, one can redirect standard input and output using the less-than (<) and greater-than (>) characters. To specify the name of the destination, prefix it with a colon (:). For example, the following stream has the destination name in the source position:

dataflow:>stream create --definition ":myDestination > log" --name ingest_from_broker --deploy

This stream receives messages from the destination called myDestination, located at the broker, and connects it to the log app. You can also create additional streams that consume data from the same named destination.

The following stream has the destination name in the sink position:

dataflow:>stream create --definition "http > :myDestination" --name ingest_to_broker --deploy

It is also possible to connect two different destinations (source and sink positions) at the broker in a stream, as shown in the following example:

dataflow:>stream create --definition ":destination1 > :destination2" --name bridge_destinations --deploy

In the precding stream, both the destinations (destination1 and destination2) are located in the broker. The messages flow from the source destination to the sink destination over a bridge app that connects them.

29.4. Fan-in and Fan-out

By using named destinations, you can support fan-in and fan-out use cases. Fan-in use cases are when multiple sources all send data to the same named destination, as shown in the following example:

s3 > :data
ftp > :data
http > :data

The preceding example directs the data payloads from the Amazon S3, FTP, and HTTP sources to the same named destination called data. Then an additional stream created with the following DSL expression would have all the data from those three sources sent to the file sink:

:data > file

The fan-out use case is when you determine the destination of a stream based on some information that is only known at runtime. In this case, the Router Application can be used to specify how to direct the incoming message to one of N named destinations.

A nice Video showing Fan-in and Fan-out behavior is also available.

30. Stream Java DSL

Instead of using the shell to create and deploy streams, you can use the Java-based DSL provided by the spring-cloud-dataflow-rest-client module. The Java DSL is a convenient wrapper around the DataFlowTemplate class that enables creating and deploying streams programmatically.

To get started, you need to add the following dependency to your project, as follows:

<dependency>
	<groupId>org.springframework.cloud</groupId>
	<artifactId>spring-cloud-dataflow-rest-client</artifactId>
	<version>1.6.0.RELEASE</version>
</dependency>
A complete sample can be found in the Spring Cloud Data Flow Samples Repository.

30.1. Overview

The classes at the heart of the Java DSL are StreamBuilder, StreamDefinition, Stream, StreamApplication, and DataFlowTemplate. The entry point is a builder method on Stream that takes an instance of a DataFlowTemplate. To create an instance of a DataFlowTemplate, you need to provide a URI location of the Data Flow Server.

Spring Boot auto-configuration for StreamBuilder and DataFlowTemplate is also available. The properties in DataFlowClientProperties can be used to configure the connection to the Data Flow server. The common property to start using is spring.cloud.dataflow.client.uri

Consider the following example, using the definition style.

URI dataFlowUri = URI.create("http://localhost:9393");
DataFlowOperations dataFlowOperations = new DataFlowTemplate(dataFlowUri);
dataFlowOperations.appRegistryOperations().importFromResource(
                     "http://bit.ly/Darwin-GA-stream-applications-rabbit-maven", true);
StreamDefinition streamDefinition = Stream.builder(dataFlowOperations)
                                      .name("ticktock")
                                      .definition("time | log")
                                      .create();

The create method returns an instance of a StreamDefinition representing a Stream that has been created but not deployed. This is called the definition style since it takes a single string for the stream definition, same as in the shell. If applications have not yet been registered in the Data Flow server, you can use the DataFlowOperations class to register them. With the StreamDefinition instance, you have methods available to deploy or destory the stream.

Stream stream = streamDefinition.deploy();

The Stream instance provides getStatus, destroy and undeploy methods to control and query the stream. If you are going to immediately deploy the stream, there is no need to create a separate local variable of the type StreamDefinition. You can just chain the calls together, as follows:

Stream stream = Stream.builder(dataFlowOperations)
                  .name("ticktock")
                  .definition("time | log")
                  .create()
                  .deploy();

The deploy method is overloaded to take a java.util.Map of deployment properties.

The StreamApplication class is used in the 'fluent' Java DSL style and is discussed in the next section. The StreamBuilder class is returned from the method Stream.builder(dataFlowOperations). In larger applications, it is common to create a single instance of the StreamBuilder as a Spring @Bean and share it across the application.

30.2. Java DSL styles

The Java DSL offers two styles to create Streams.

  • The definition style keeps the feel of using the pipes and filters textual DSL in the shell. This style is selected by using the definition method after setting the stream name - for example, Stream.builder(dataFlowOperations).name("ticktock").definition(<definition goes here>).

  • The fluent style lets you chain together sources, processors, and sinks by passing in an instance of a StreamApplication. This style is selected by using the source method after setting the stream name - for example, Stream.builder(dataFlowOperations).name("ticktock").source(<stream application instance goes here>). You then chain together processor() and sink() methods to create a stream definition.

To demonstrate both styles, we include a simple stream that uses both approaches. A complete sample for you to get started can be found in the Spring Cloud Data Flow Samples Repository.

The following example demonstrates the definition approach:

public void definitionStyle() throws Exception{

  DataFlowOperations dataFlowOperations = createDataFlowOperations();
  Map<String, String> deploymentProperties = createDeploymentProperties();

  Stream woodchuck = Stream.builder(dataFlowOperations)
          .name("woodchuck")
          .definition("http --server.port=9900 | splitter --expression=payload.split(' ') | log")
          .create()
          .deploy(deploymentProperties);

  waitAndDestroy(woodchuck)
}

The following example demonstrates the fluent approach:

public void fluentStyle() throws Exception {

  DataFlowOperations dataFlowOperations = createDataFlowOperations();

  StreamApplication source = new StreamApplication("http").addProperty("server.port", 9900);

  StreamApplication processor = new StreamApplication("splitter")
                                 .addProperty("producer.partitionKeyExpression", "payload");

  StreamApplication sink = new StreamApplication("log")
                            .addDeploymentProperty("count", 2);

  Stream woodchuck = Stream.builder(dataFlowOperations).name("woodchuck")
          .source(source)
          .processor(processor)
          .sink(sink)
          .create()
          .deploy(deploymentProperties);

  waitAndDestroy(woodchuck)

}

The waitAndDestroy method uses the getStatus method to poll for the stream’s status, as shown in the following example:

private void waitAndDestroy(Stream stream) throws InterruptedException {

  while(!stream.getStatus().equals("deployed")){
    System.out.println("Wating for deployment of stream.");
    Thread.sleep(5000);
  }

  System.out.println("Letting the stream run for 2 minutes.");
  // Let the stream run for 2 minutes
  Thread.sleep(120000);

  System.out.println("Destroying stream");
  stream.destroy();
}

When using the definition style, the deployment properties are specified as a java.util.Map in the same manner as using the shell. The createDeploymentProperties method is defined as follows:

private Map<String, String> createDeploymentProperties() {
  Map<String, String> deploymentProperties = new HashMap<>();
  deploymentProperties.put("app.splitter.producer.partitionKeyExpression", "payload");
  deploymentProperties.put("deployer.log.memory","512");
  deploymentProperties.put("deployer.log.count", "2");
  return deploymentProperties;
}

Is this case, application properties are also overridden at deployment time in addition to setting the deployer property count for the log application. When using the fluent style, the deployment properties are added by using the method addDeploymentProperty (for example, new StreamApplication("log").addDeploymentProperty("count", 2)), and you do not need to prefix the property with deployer.<app_name>.

In order to create and deploy your streams, you need to make sure that the corresponding apps have been registered in the DataFlow server first. Attempting to create or deploy a stream that contains an unknown app throws an exception. You can register your application by using the DataFlowTemplate, as follows:
dataFlowOperations.appRegistryOperations().importFromResource(
            "http://bit.ly/Darwin-GA-stream-applications-rabbit-maven", true);

The Stream applications can also be beans within your application that are injected in other classes to create Streams. There are many ways to structure Spring applications, but one way is to have an @Configuration class define the StreamBuilder and StreamApplications, as shown in the following example:

@Configuration
public StreamConfiguration {

  @Bean
  public StreamBuilder builder() {
    return Stream.builder(new DataFlowTemplate(URI.create("http://localhost:9393")));
  }

  @Bean
  public StreamApplication httpSource(){
    return new StreamApplication("http");
  }

  @Bean
  public StreamApplication logSink(){
    return new StreamApplication("log");
  }
}

Then in another class you can @Autowire these classes and deploy a stream.

@Component
public MyStreamApps {

  @Autowired
  private StreamBuilder streamBuilder;

  @Autowired
  private StreamApplication httpSource;

  @Autowired
  private StreamApplication logSink;


  public void deploySimpleStream() {
    Stream simpleStream = streamBuilder.name("simpleStream")
                            .source(httpSource);
                            .sink(logSink)
                            .create()
                            .deploy();
  }
}

This style lets you share StreamApplications across multiple Streams.

30.3. Using the DeploymentPropertiesBuilder

Regardless of style you choose, the deploy(Map<String, String> deploymentProperties) method allows customization of how your streams will be deployed. We made it a easier to create a map with properties by using a builder style, as well as creating static methods for some properties so you don’t need to remember the name of such properties. If you take the previous example of createDeploymentProperties it could be rewritten as:

private Map<String, String> createDeploymentProperties() {
	return new DeploymentPropertiesBuilder()
		.count("log", 2)
		.memory("log", 512)
		.put("app.splitter.producer.partitionKeyExpression", "payload")
		.build();
}

This utility class is meant to help with the creation of a Map and adds a few methods to assist with defining pre-defined properties.

31. Deploying using Skipper

If you desire to deploy your streams using Skipper, you need to pass certain properties to the server specific to a Skipper based deployment, for example selecting the target platfrom. The SkipperDeploymentPropertiesBuilder provides you all the properties in DeploymentPropertiesBuilder and adds those needed for Skipper.

private Map<String, String> createDeploymentProperties() {
	return new SkipperDeploymentPropertiesBuilder()
		.count("log", 2)
		.memory("log", 512)
		.put("app.splitter.producer.partitionKeyExpression", "payload")
		.platformName("pcf")
		.build();
}

32. Stream Applications with Multiple Binder Configurations

In some cases, a stream can have its applications bound to multiple spring cloud stream binders when they are required to connect to different messaging middleware configurations. In those cases, it is important to make sure the applications are configured appropriately with their binder configurations. For example, consider the following stream:

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

In this stream, each application connects to messaging middleware in the following way:

  1. The HTTP source sends events to RabbitMQ (rabbit1).

  2. The Transform processor receives events from RabbitMQ (rabbit1) and sends the processed events into Kafka (kafka1).

  3. The log sink receives events from Kafka (kafka1).

Here, rabbit1 and kafka1 are the binder names given in the spring cloud stream application properties. Based on this setup, the applications have the following binder(s) in their classpath with the appropriate configuration:

  • HTTP: Rabbit binder

  • Transform: Both Kafka and Rabbit binders

  • Log: Kafka binder

The spring-cloud-stream binder configuration properties can be set within the applications themselves. If not, they can be passed through deployment properties when the stream is deployed as shown in the following example:

dataflow:>stream create --definition "http | transform --expression=payload.toUpperCase() | log" --name mystream

dataflow:>stream deploy mystream --properties "app.http.spring.cloud.stream.bindings.output.binder=rabbit1,app.transform.spring.cloud.stream.bindings.input.binder=rabbit1,
app.transform.spring.cloud.stream.bindings.output.binder=kafka1,app.log.spring.cloud.stream.bindings.input.binder=kafka1"

One can override any of the binder configuration properties by specifying them through deployment properties.

33. Examples

This chapter includes the following examples:

You can find links to more samples in the “[dataflow-samples]” chapter.

33.1. Simple Stream Processing

As an example of a simple processing step, we can transform the payload of the HTTP posted data to upper case by using the following stream definition: http | transform --expression=payload.toUpperCase() | log

To create this stream enter the following command in the shell dataflow:> stream create --definition "http | transform --expression=payload.toUpperCase() | log" --name mystream --deploy

The following example uses a shell command to post some data:

dataflow:> http post --target localhost:1234 --data "hello"

The preceding example results in an upper-case 'HELLO' in the log, as follows:

2016-06-01 09:54:37.749 INFO 80083 --- [ kafka-binder-] log.sink : HELLO

33.2. Stateful Stream Processing

To demonstrate the data partitioning functionality, the following listing deploys a stream with Kafka as the binder:

dataflow:>stream create --name words --definition "http --server.port=9900 | splitter --expression=payload.split(' ') | log"
Created new stream 'words'

dataflow:>stream deploy words --properties "app.splitter.producer.partitionKeyExpression=payload,deployer.log.count=2"
Deployed stream 'words'

dataflow:>http post --target http://localhost:9900 --data "How much wood would a woodchuck chuck if a woodchuck could chuck wood"
> POST (text/plain;Charset=UTF-8) http://localhost:9900 How much wood would a woodchuck chuck if a woodchuck could chuck wood
> 202 ACCEPTED

You should then see the following in the server logs:

2016-06-05 18:33:24.982  INFO 58039 --- [nio-9393-exec-9] o.s.c.d.spi.local.LocalAppDeployer       : deploying app words.log instance 0
   Logs will be in /var/folders/c3/ctx7_rns6x30tq7rb76wzqwr0000gp/T/spring-cloud-dataflow-694182453710731989/words-1465176804970/words.log
2016-06-05 18:33:24.988  INFO 58039 --- [nio-9393-exec-9] o.s.c.d.spi.local.LocalAppDeployer       : deploying app words.log instance 1
   Logs will be in /var/folders/c3/ctx7_rns6x30tq7rb76wzqwr0000gp/T/spring-cloud-dataflow-694182453710731989/words-1465176804970/words.log

When you review the words.log instance 0 logs, you should see the following:

2016-06-05 18:35:47.047  INFO 58638 --- [  kafka-binder-] log.sink                                 : How
2016-06-05 18:35:47.066  INFO 58638 --- [  kafka-binder-] log.sink                                 : chuck
2016-06-05 18:35:47.066  INFO 58638 --- [  kafka-binder-] log.sink                                 : chuck

When you review the words.log instance 1 logs, you shoul see the following:

2016-06-05 18:35:47.047  INFO 58639 --- [  kafka-binder-] log.sink                                 : much
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : wood
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : would
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : a
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : woodchuck
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : if
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : a
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : woodchuck
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : could
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : wood

This example has shown that payload splits that contain the same word are routed to the same application instance.

33.3. Other Source and Sink Application Types

This example shows something a bit more complicated: swapping 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 from the Data Flow Server (default 8080). By default, the port is randomly assigned.

To create a stream using an http source but still using the same log sink, we would change the original command in the Simple Stream Processing example to the following:

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

The preceding command produces the following output from the server:

2016-06-01 09:47:58.920  INFO 79016 --- [io-9393-exec-10] o.s.c.d.spi.local.LocalAppDeployer       : deploying app myhttpstream.log instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/myhttpstream-1464788878747/myhttpstream.log
2016-06-01 09:48:06.396  INFO 79016 --- [io-9393-exec-10] o.s.c.d.spi.local.LocalAppDeployer       : deploying app myhttpstream.http instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/myhttpstream-1464788886383/myhttpstream.http

Note that we do not see any other output this time until we actually post some data (by using a shell command). In order to see the randomly assigned port on which the http source is listening, run the following command:

dataflow:> runtime apps

You should see that the corresponding http source has a url property containing the host and port information on which it is listening. You are now ready to post to that url, as shown in the following example:

dataflow:> http post --target http://localhost:1234 --data "hello"
dataflow:> http post --target http://localhost:1234 --data "goodbye"

The stream then funnels the data from the http source to the output log implemented by the log sink, yielding output similar to the following:

2016-06-01 09:50:22.121  INFO 79654 --- [  kafka-binder-] log.sink    : hello
2016-06-01 09:50:26.810  INFO 79654 --- [  kafka-binder-] log.sink    : goodbye

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 applications that are available. You can also define your own applications.

Streams deployed using Skipper

We will proceed with the assumption that Spring Cloud Data Flow, Spring Cloud Skipper, RDBMS, and desired messaging middleware is up and running in minikube. In this section we will use RabbitMQ as the messaging middleware. The same concepts and steps apply for Kafka as well, except Kafka based binder Docker images would be used as described in Create Streams without Skipper.

$ kubectl get all
NAME                              READY     STATUS    RESTARTS   AGE
po/mysql-777890292-z0dsw          1/1       Running   0          38m
po/rabbitmq-317767540-2qzrr       1/1       Running   0          38m
po/redis-4054078334-37m0l         1/1       Running   0          38m
po/scdf-server-2734071167-bjd3g   1/1       Running   0          48s
po/skipper-2408247821-50z31       1/1       Running   0          3m

...
...

To start the Data Flow Shell for the Data Flow server running in skipper mode:

$ java -jar spring-cloud-dataflow-shell-1.6.0.RELEASE.jar --dataflow.mode=skipper

If the Data Flow Server and shell are not running on the same host, point the shell to the Data Flow server URL.

Use the kubectl get svc scdf-server command to locate the EXTERNAL_IP address assigned to scdf-server, we will use that to connect from the shell.

$ kubectl get svc scdf-server
NAME         CLUSTER-IP       EXTERNAL-IP       PORT(S)    AGE
scdf-server  10.103.246.82    130.211.203.246   80/TCP     4m

So the URL you need to use is in this case 130.211.203.246

If you are using Minikube then you don’t have an external load balancer and the EXTERNAL-IP will show as <pending>. You need to use the NodePort assigned for the skipper service. Use this command to look up the URL to use:

$ minikube service --url scdf-server
http://192.168.99.100:31991

Configure the Data Flow server URI with the following command using the default user and password settings:

server-unknown:>dataflow config server --username user --password password --uri http://130.211.203.246
Successfully targeted http://130.211.203.246
dataflow:>

Alternatively, pass in the command line option --dataflow.uri. The shell’s command line option --help shows what is available.

Verify the registered platforms in Skipper.

dataflow:>stream platform-list
╔════════╤══════════╤════════════════════════════════════════════════════════════════════════════════════════════════╗
║  Name  │   Type   │                                      Description                                               ║
╠════════╪══════════╪════════════════════════════════════════════════════════════════════════════════════════════════╣
║minikube│kubernetes│master url = [https://kubernetes.default.svc/], namespace = [default], api version = [v1]       ║
╚════════╧══════════╧════════════════════════════════════════════════════════════════════════════════════════════════╝

Let’s start with deploying a stream with the time-source pointing to 1.3.0.RELEASE and log-sink pointing to 1.2.0.RELEASE. The goal is to rolling upgrade the log-sink application to 1.3.0.RELEASE.

dataflow:>app register --type source --name time --uri docker://springcloudstream/time-source-rabbit:1.3.0.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:time-source-rabbit:jar:metadata:1.3.0.RELEASE
Successfully registered application 'source:time'

dataflow:>app register --type sink --name log --uri docker://springcloudstream/log-sink-rabbit:1.2.0.RELEASE --metadata-uri maven://org.springframework.cloud.stream.app:log-sink-rabbit:jar:metadata:1.2.0.RELEASE
Successfully registered application 'sink:log'

dataflow:>app info time --type source
Information about source application 'time':
Version: '1.3.0.RELEASE':
Default application version: 'true':
Resource URI: docker://springcloudstream/time-source-rabbit:1.3.0.RELEASE
╔══════════════════════════════╤══════════════════════════════╤══════════════════════════════╤══════════════════════════════╗
║         Option Name          │         Description          │           Default            │             Type             ║
╠══════════════════════════════╪══════════════════════════════╪══════════════════════════════╪══════════════════════════════╣
║trigger.time-unit             │The TimeUnit to apply to delay│<none>                        │java.util.concurrent.TimeUnit ║
║                              │values.                       │                              │                              ║
║trigger.fixed-delay           │Fixed delay for periodic      │1                             │java.lang.Integer             ║
║                              │triggers.                     │                              │                              ║
║trigger.cron                  │Cron expression value for the │<none>                        │java.lang.String              ║
║                              │Cron Trigger.                 │                              │                              ║
║trigger.initial-delay         │Initial delay for periodic    │0                             │java.lang.Integer             ║
║                              │triggers.                     │                              │                              ║
║trigger.max-messages          │Maximum messages per poll, -1 │1                             │java.lang.Long                ║
║                              │means infinity.               │                              │                              ║
║trigger.date-format           │Format for the date value.    │<none>                        │java.lang.String              ║
╚══════════════════════════════╧══════════════════════════════╧══════════════════════════════╧══════════════════════════════╝

dataflow:>app info log --type sink
Information about sink application 'log':
Version: '1.2.0.RELEASE':
Default application version: 'true':
Resource URI: docker://springcloudstream/log-sink-rabbit:1.2.0.RELEASE
╔══════════════════════════════╤══════════════════════════════╤══════════════════════════════╤══════════════════════════════╗
║         Option Name          │         Description          │           Default            │             Type             ║
╠══════════════════════════════╪══════════════════════════════╪══════════════════════════════╪══════════════════════════════╣
║log.name                      │The name of the logger to use.│<none>                        │java.lang.String              ║
║log.level                     │The level at which to log     │<none>                        │org.springframework.integratio║
║                              │messages.                     │                              │n.handler.LoggingHandler$Level║
║log.expression                │A SpEL expression (against the│payload                       │java.lang.String              ║
║                              │incoming message) to evaluate │                              │                              ║
║                              │as the logged message.        │                              │                              ║
╚══════════════════════════════╧══════════════════════════════╧══════════════════════════════╧══════════════════════════════╝
  1. Create stream.

    dataflow:>stream create foo --definition "time | log"
    Created new stream 'foo'
  2. Deploy stream.

    dataflow:>stream deploy foo --platformName minikube
    Deployment request has been sent for stream 'foo'

    While deploying the stream, we are supplying --platformName and that indicates the platform repository (i.e., minikube) to use when deploying the stream applications via Skipper.

  3. List pods.

    $ kubectl get pods
    NAME                              READY     STATUS    RESTARTS   AGE
    foo-log-v1-0-2k4r8             1/1       Running   0          2m
    foo-time-v1-qhdqq              1/1       Running   0          2m
    mysql-777890292-z0dsw          1/1       Running   0          49m
    rabbitmq-317767540-2qzrr       1/1       Running   0          49m
    redis-4054078334-37m0l         1/1       Running   0          49m
    scdf-server-2734071167-bjd3g   1/1       Running   0          12m
    skipper-2408247821-50z31       1/1       Running   0          15m
    
    ...
    ...
  4. Verify logs.

    $ kubectl logs -f foo-log-v1-0-2k4r8
    ...
    ...
    2017-10-30 22:59:04.966  INFO 1 --- [ foo.time.foo-1] log-sink                                 : 10/30/17 22:59:04
    2017-10-30 22:59:05.968  INFO 1 --- [ foo.time.foo-1] log-sink                                 : 10/30/17 22:59:05
    2017-10-30 22:59:07.000  INFO 1 --- [ foo.time.foo-1] log-sink                                 : 10/30/17 22:59:06
  5. Verify the stream history.

    dataflow:>stream history --name foo
    ╔═══════╤════════════════════════════╤════════╤════════════╤═══════════════╤════════════════╗
    ║Version│        Last updated        │ Status │Package Name│Package Version│  Description   ║
    ╠═══════╪════════════════════════════╪════════╪════════════╪═══════════════╪════════════════╣
    ║1      │Mon Oct 30 16:18:28 PDT 2017│DEPLOYED│foo         │1.0.0          │Install complete║
    ╚═══════╧════════════════════════════╧════════╧════════════╧═══════════════╧════════════════╝
  6. Verify the package manifest. The log-sink should be at 1.2.0.RELEASE.

    dataflow:>stream manifest --name foo
    
    ---
    # Source: log.yml
    apiVersion: skipper.spring.io/v1
    kind: SpringCloudDeployerApplication
    metadata:
      "name": "log"
    spec:
      resource: "docker:springcloudstream/log-sink-rabbit"
      resourceMetadata: "docker:springcloudstream/log-sink-rabbit:jar:metadata:1.2.0.RELEASE"
      version: "1.2.0.RELEASE"
      applicationProperties:
        "spring.metrics.export.triggers.application.includes": "integration**"
        "spring.cloud.dataflow.stream.app.label": "log"
        "spring.cloud.stream.metrics.key": "foo.log.${spring.cloud.application.guid}"
        "spring.cloud.stream.bindings.input.group": "foo"
        "spring.cloud.stream.metrics.properties": "spring.application.name,spring.application.index,spring.cloud.application.*,spring.cloud.dataflow.*"
        "spring.cloud.stream.bindings.applicationMetrics.destination": "metrics"
        "spring.cloud.dataflow.stream.name": "foo"
        "spring.cloud.dataflow.stream.app.type": "sink"
        "spring.cloud.stream.bindings.input.destination": "foo.time"
      deploymentProperties:
        "spring.cloud.deployer.group": "foo"
    
    ---
    # Source: time.yml
    apiVersion: skipper.spring.io/v1
    kind: SpringCloudDeployerApplication
    metadata:
      "name": "time"
    spec:
      resource: "docker:springcloudstream/time-source-rabbit"
      resourceMetadata: "docker:springcloudstream/time-source-rabbit:jar:metadata:1.3.0.RELEASE"
      version: "1.3.0.RELEASE"
      applicationProperties:
        "spring.metrics.export.triggers.application.includes": "integration**"
        "spring.cloud.dataflow.stream.app.label": "time"
        "spring.cloud.stream.metrics.key": "foo.time.${spring.cloud.application.guid}"
        "spring.cloud.stream.bindings.output.producer.requiredGroups": "foo"
        "spring.cloud.stream.metrics.properties": "spring.application.name,spring.application.index,spring.cloud.application.*,spring.cloud.dataflow.*"
        "spring.cloud.stream.bindings.applicationMetrics.destination": "metrics"
        "spring.cloud.stream.bindings.output.destination": "foo.time"
        "spring.cloud.dataflow.stream.name": "foo"
        "spring.cloud.dataflow.stream.app.type": "source"
      deploymentProperties:
        "spring.cloud.deployer.group": "foo"
  7. Let’s register log-sink application version 1.3.0.RELEASE and update our stream to use it

    dataflow:>app register --name log --type sink --uri docker:springcloudstream/log-sink-rabbit:1.3.0.RELEASE --force
    Successfully registered application 'sink:log'
    
    dataflow:>stream update --name foo --properties version.log=1.3.0.RELEASE
    Update request has been sent for stream 'foo'
  8. List pods.

    $ kubectl get pods
    NAME                              READY     STATUS        RESTARTS   AGE
    foo-log-v1-0-2k4r8             1/1       Terminating   0          3m
    foo-log-v2-0-fjnlt             0/1       Running       0          9s
    foo-time-v1-qhdqq              1/1       Running       0          3m
    mysql-777890292-z0dsw          1/1       Running       0          51m
    rabbitmq-317767540-2qzrr       1/1       Running       0          51m
    redis-4054078334-37m0l         1/1       Running       0          51m
    scdf-server-2734071167-bjd3g   1/1       Running       0          14m
    skipper-2408247821-50z31       1/1       Running       0          16m
    
    ...
    ...

    Notice that there are two versions of the log-sink applications. The foo-log-v1-0-2k4r8 pod is going down and the newly spawned foo-log-v2-0-fjnlt pod is bootstrapping. The version number is incremented and the version-number (v2) is included in the new application name.

  9. Once the new pod is up and running, let’s verify the logs.

    $ kubectl logs -f foo-log-v2-0-fjnlt
    ...
    ...
    2017-10-30 23:24:30.016  INFO 1 --- [ foo.time.foo-1] log-sink                                 : 10/30/17 23:24:30
    2017-10-30 23:24:31.017  INFO 1 --- [ foo.time.foo-1] log-sink                                 : 10/30/17 23:24:31
    2017-10-30 23:24:32.018  INFO 1 --- [ foo.time.foo-1] log-sink                                 : 10/30/17 23:24:32
  10. Let’s look at the updated package manifest persisted in Skipper. We should now be seeing log-sink at 1.3.0.RELEASE.

    dataflow:>stream manifest --name foo
    
    ---
    # Source: log.yml
    apiVersion: skipper.spring.io/v1
    kind: SpringCloudDeployerApplication
    metadata:
      "name": "log"
    spec:
      resource: "docker:springcloudstream/log-sink-rabbit"
      resourceMetadata: "docker:springcloudstream/log-sink-rabbit:jar:metadata:1.3.0.RELEASE"
      version: "1.3.0.RELEASE"
      applicationProperties:
        "spring.metrics.export.triggers.application.includes": "integration**"
        "spring.cloud.dataflow.stream.app.label": "log"
        "spring.cloud.stream.metrics.key": "foo.log.${spring.cloud.application.guid}"
        "spring.cloud.stream.bindings.input.group": "foo"
        "spring.cloud.stream.metrics.properties": "spring.application.name,spring.application.index,spring.cloud.application.*,spring.cloud.dataflow.*"
        "spring.cloud.stream.bindings.applicationMetrics.destination": "metrics"
        "spring.cloud.dataflow.stream.name": "foo"
        "spring.cloud.dataflow.stream.app.type": "sink"
        "spring.cloud.stream.bindings.input.destination": "foo.time"
      deploymentProperties:
        "spring.cloud.deployer.group": "foo"
        "spring.cloud.deployer.count": "1"
    
    ---
    # Source: time.yml
    apiVersion: skipper.spring.io/v1
    kind: SpringCloudDeployerApplication
    metadata:
      "name": "time"
    spec:
      resource: "docker:springcloudstream/time-source-rabbit"
      resourceMetadata: "docker:springcloudstream/time-source-rabbit:jar:metadata:1.3.0.RELEASE"
      version: "1.3.0.RELEASE"
      applicationProperties:
        "spring.metrics.export.triggers.application.includes": "integration**"
        "spring.cloud.dataflow.stream.app.label": "time"
        "spring.cloud.stream.metrics.key": "foo.time.${spring.cloud.application.guid}"
        "spring.cloud.stream.bindings.output.producer.requiredGroups": "foo"
        "spring.cloud.stream.metrics.properties": "spring.application.name,spring.application.index,spring.cloud.application.*,spring.cloud.dataflow.*"
        "spring.cloud.stream.bindings.applicationMetrics.destination": "metrics"
        "spring.cloud.stream.bindings.output.destination": "foo.time"
        "spring.cloud.dataflow.stream.name": "foo"
        "spring.cloud.dataflow.stream.app.type": "source"
      deploymentProperties:
        "spring.cloud.deployer.group": "foo"
  11. Verify stream history for the latest updates.

    dataflow:>stream history --name foo
    ╔═══════╤════════════════════════════╤════════╤════════════╤═══════════════╤════════════════╗
    ║Version│        Last updated        │ Status │Package Name│Package Version│  Description   ║
    ╠═══════╪════════════════════════════╪════════╪════════════╪═══════════════╪════════════════╣
    ║2      │Mon Oct 30 16:21:55 PDT 2017│DEPLOYED│foo         │1.0.0          │Upgrade complete║
    ║1      │Mon Oct 30 16:18:28 PDT 2017│DELETED │foo         │1.0.0          │Delete complete ║
    ╚═══════╧════════════════════════════╧════════╧════════════╧═══════════════╧════════════════╝
  12. Rolling-back to the previous version is just a command away.

    dataflow:>stream rollback --name foo
    Rollback request has been sent for the stream 'foo'
    
    ...
    ...
    
    dataflow:>stream history --name foo
    ╔═══════╤════════════════════════════╤════════╤════════════╤═══════════════╤════════════════╗
    ║Version│        Last updated        │ Status │Package Name│Package Version│  Description   ║
    ╠═══════╪════════════════════════════╪════════╪════════════╪═══════════════╪════════════════╣
    ║3      │Mon Oct 30 16:22:51 PDT 2017│DEPLOYED│foo         │1.0.0          │Upgrade complete║
    ║2      │Mon Oct 30 16:21:55 PDT 2017│DELETED │foo         │1.0.0          │Delete complete ║
    ║1      │Mon Oct 30 16:18:28 PDT 2017│DELETED │foo         │1.0.0          │Delete complete ║
    ╚═══════╧════════════════════════════╧════════╧════════════╧═══════════════╧════════════════╝

Tasks

This section goes into more detail about how you can work with Spring Cloud Task. It covers topics such as creating and running task applications.

If you are just starting out with Spring Cloud Data Flow, you should probably read the “Getting Started” guide before diving into this section.

34. Introduction

A task executes a process on demand. In the case of Spring Cloud Task, a task is a Spring Boot application that is annotated with @EnableTask. A user launches a task that performs a certain process, and, once complete, the task ends. Unlike a stream where a stream definition can have at most one deployment a single task definition can be launched multiple times simultaneously. An example of a task would be a Spring Boot application that exports data from a JDBC repository to an HDFS instance. Tasks record the start time and the end time as well as the boot exit code in a relational database. The task implementation is based on the Spring Cloud Task project.

34.1. Application properties

Each application takes properties to customize its behavior. As an example, the timestamp task format setting establishes a output format that is different from the default value.

dataflow:> task create --definition "timestamp --format=\"yyyy\"" --name printTimeStamp

This timestamp property is actually the same as the timestamp.format property specified by the timestamp application. Data Flow adds the ability to use the shorthand form format instead of timestamp.format. One may also specify the longhand version as well, as shown in the following example:

dataflow:> task create --definition "timestamp --timestamp.format=\"yyyy\"" --name printTimeStamp

This shorthand behavior is discussed more in the section on Whitelisting application properties. If you have registered application property metadata you can use tab completion in the shell after typing -- to get a list of candidate property names.

The shell provides tab completion for application properties. The shell command app info --name <appName> --type <appType> provides additional documentation for all the supported properties.

The supported Task <appType> is task.

35. The Lifecycle of a Task

Before we dive deeper into the details of creating Tasks, we need to understand the typical lifecycle for tasks in the context of Spring Cloud Data Flow:

35.1. Creating a Task Application

While Spring Cloud Task does provide a number of out-of-the-box applications (at spring-cloud-task-app-starters), most task applications require custom development. To create a custom task application:

  1. Use the Spring Initializer to create a new project, making sure to select the following starters:

    1. Cloud Task: This dependency is the spring-cloud-starter-task.

    2. JDBC: This dependency is the spring-jdbc starter.

  2. Within your new project, create a new class to serve as your main class, as follows:

    @EnableTask
    @SpringBootApplication
    public class MyTask {
    
        public static void main(String[] args) {
    		SpringApplication.run(MyTask.class, args);
    	}
    }
  3. With this class, you need one or more CommandLineRunner or ApplicationRunner implementations within your application. You can either implement your own or use the ones provided by Spring Boot (there is one for running batch jobs, for example).

  4. Packaging your application with Spring Boot into an über jar is done through the standard {spring-boot-docs-reference}/html/getting-started-first-application.html#getting-started-first-application-executable-jar[Spring Boot conventions]. The packaged application can be registered and deployed as noted below.

35.1.1. Task Database Configuration

When launching a task application, be sure that the database driver that is being used by Spring Cloud Data Flow is also a dependency on the task application. For example, if your Spring Cloud Data Flow is set to use Postgresql, be sure that the task application also has Postgresql as a dependency.
When you run tasks externally (that is, from the command line) and you want Spring Cloud Data Flow to show the TaskExecutions in its UI, be sure that common datasource settings are shared among the both. By default, Spring Cloud Task uses a local H2 instance, and the execution is recorded to the database used by Spring Cloud Data Flow.

35.2. Registering a Task Application

You can register a Task App with the App Registry by using the Spring Cloud Data Flow Shell app register command. You must provide a unique name and a URI that can be resolved to the app artifact. For the type, specify "task". The following listing shows three examples:

dataflow:>app register --name task1 --type task --uri maven://com.example:mytask:1.0.2

dataflow:>app register --name task2 --type task --uri file:///Users/example/mytask-1.0.2.jar

dataflow:>app register --name task3 --type task --uri http://example.com/mytask-1.0.2.jar

When providing a URI with the maven scheme, the format should conform to the following:

maven://<groupId>:<artifactId>[:<extension>[:<classifier>]]:<version>

If you would like to register multiple apps at one time, you can store them in a properties file where the keys are formatted as <type>.<name> and the values are the URIs. For example, the followinng listing would be a valid properties file:

task.foo=file:///tmp/foo.jar
task.bar=file:///tmp/bar.jar

Then you can use the app import command and provide the location of the properties file by using the --uri option, as follows:

app import --uri /tmp/task-apps.properties

For convenience, we have the static files with application-URIs (for both maven and docker) available for all the out-of-the-box Task app-starters. You can point to this file and import all the application-URIs in bulk. Otherwise, as explained earlier in this chapter, you can register them individually or have your own custom property file with only the required application-URIs in it. It is recommended, however, to have a “focused” list of desired application-URIs in a custom property file.

The following table lists the available static property files:

Artifact Type Stable Release SNAPSHOT Release

Maven

bit.ly/Clark-GA-task-applications-maven

bit.ly/Clark-BUILD-SNAPSHOT-task-applications-maven

Docker

bit.ly/Clark-GA-task-applications-docker

bit.ly/Clark-BUILD-SNAPSHOT-task-applications-docker

For example, if you would like to register all out-of-the-box task applications in bulk, you can do so with the following command:

dataflow:>app import --uri bit.ly/Clark-GA-task-applications-maven

You can also pass the --local option (which is TRUE by default) to indicate whether the properties file location should be resolved within the shell process itself. If the location should be resolved from the Data Flow Server process, specify --local false.

When using either app register or app import, if a task app is already registered with the provided name, it is not overridden by default. If you would like to override the pre-existing task app, then include the --force option.

In some cases, the Resource is resolved on the server side. In other cases, the URI is passed to a runtime container instance where it is resolved. Consult the specific documentation of each Data Flow Server for more detail.

35.3. Creating a Task Definition

You can create a task Definition from a task app by providing a definition name as well as properties that apply to the task execution. Creating a task definition can be done through the RESTful API or the shell. To create a task definition by using the shell, use the task create command to create the task definition, as shown in the following example:

dataflow:>task create mytask --definition "timestamp --format=\"yyyy\""
Created new task 'mytask'

A listing of the current task definitions can be obtained through the RESTful API or the shell. To get the task definition list by using the shell, use the task list command.

35.4. Launching a Task

An adhoc task can be launched through the RESTful API or the shell. To launch an ad-hoc task through the shell, use the task launch command, as shown in the following example:

dataflow:>task launch mytask
 Launched task 'mytask'

When a task is launched, any properties that need to be passed as command line arguments to the task application can be set when launching the task, as follows:

dataflow:>task launch mytask --arguments "--server.port=8080 --custom=value"
The arguments need to be passed as space delimited values.

Additional properties meant for a TaskLauncher itself can be passed in by using a --properties option. The format of this option is a comma-separated string of properties prefixed with app.<task definition name>.<property>. Properties are passed to TaskLauncher as application properties. It is up to an implementation to choose how those are passed into an actual task application. If the property is prefixed with deployer instead of app, it is passed to TaskLauncher as a deployment property and its meaning may be TaskLauncher implementation specific.

dataflow:>task launch mytask --properties "deployer.timestamp.custom1=value1,app.timestamp.custom2=value2"

35.4.1. Common application properties

In addition to configuration through DSL, Spring Cloud Data Flow provides a mechanism for setting common properties to all the task applications that are launched by it. This can be done by adding properties prefixed with spring.cloud.dataflow.applicationProperties.task when starting the server. When doing so, the server passes all the properties, without the prefix, to the instances it launches.

For example, all the launched applications can be configured to use the properties prop1 and prop2 by launching the Data Flow server with the following options:

--spring.cloud.dataflow.applicationProperties.task.prop1=value1
--spring.cloud.dataflow.applicationProperties.task.prop2=value2

This causes the properties, prop1=value1 and prop2=value2, to be passed to all the launched applications.

Properties configured by using this mechanism have lower precedence than task deployment properties. They are overridden if a property with the same key is specified at task launch time (for example, app.trigger.prop2 overrides the common property).

35.5. Reviewing Task Executions

Once the task is launched, the state of the task is stored in a relational DB. The state includes:

  • Task Name

  • Start Time

  • End Time

  • Exit Code

  • Exit Message

  • Last Updated Time

  • Parameters

A user can check the status of their task executions through the RESTful API or the shell. To display the latest task executions through the shell, use the task execution list command.

To get a list of task executions for just one task definition, add --name and the task definition name, for example task execution list --name foo. To retrieve full details for a task execution use the task execution status command with the id of the task execution, for example task execution status --id 549.

35.6. Destroying a Task Definition

Destroying a Task Definition removes the definition from the definition repository. This can be done through the RESTful API or the shell. To destroy a task through the shell, use the task destroy command, as shown in the following example:

dataflow:>task destroy mytask
 Destroyed task 'mytask'

The task execution information for previously launched tasks for the definition remains in the task repository.

This does not stop any currently executing tasks for this definition. Instead, it removes the task definition from the database.

36. Subscribing to Task/Batch Events

You can also tap into various task and batch events when the task is launched. If the task is enabled to generate task or batch events (with the additional dependencies spring-cloud-task-stream and, in the case of Kafka as the binder, spring-cloud-stream-binder-kafka), those events are published during the task lifecycle. By default, the destination names for those published events on the broker (Rabbit, Kafka, and others) are the event names themselves (for instance: task-events, job-execution-events, and so on).

dataflow:>task create myTask --definition “myBatchJob"
dataflow:>task launch myTask
dataflow:>stream create task-event-subscriber1 --definition ":task-events > log" --deploy

You can control the destination name for those events by specifying explicit names when launching the task, as follows:

dataflow:>task launch myTask --properties "spring.cloud.stream.bindings.task-events.destination=myTaskEvents"
dataflow:>stream create task-event-subscriber2 --definition ":myTaskEvents > log" --deploy

The following table lists the default task and batch event and destination names on the broker:

Table 1. Task and Batch Event Destinations

Event

Destination

Task events

task-events

Job Execution events

job-execution-events

Step Execution events

step-execution-events

Item Read events

item-read-events

Item Process events

item-process-events

Item Write events

item-write-events

Skip events

skip-events

37. Composed Tasks

Spring Cloud Data Flow lets a user create a directed graph where each node of the graph is a task application. This is done by using the DSL for composed tasks. A composed task can be created through the RESTful API, the Spring Cloud Data Flow Shell, or the Spring Cloud Data Flow UI.

37.1. Configuring the Composed Task Runner

Composed tasks are executed through a task application called the Composed Task Runner.

37.1.1. Registering the Composed Task Runner

By default, the Composed Task Runner application is not registered with Spring Cloud Data Flow. Consequently, to launch composed tasks, we must first register the Composed Task Runner as an application with Spring Cloud Data Flow, as follows:

app register --name composed-task-runner --type task --uri maven://org.springframework.cloud.task.app:composedtaskrunner-task:<DESIRED_VERSION>

You can also configure Spring Cloud Data Flow to use a different task definition name for the composed task runner. This can be done by setting the spring.cloud.dataflow.task.composedTaskRunnerName property to the name of your choice. You can then register the composed task runner application with the name you set by using that property.

37.1.2. Configuring the Composed Task Runner

The Composed Task Runner application has a dataflow.server.uri property that is used for validation and for launching child tasks. This defaults to localhost:9393. If you run a distributed Spring Cloud Data Flow server, as you would if you deploy the server on Cloud Foundry, YARN, or Kubernetes, you need to provide the URI that can be used to access the server. You can either provide this dataflow.server.uri property for the Composed Task Runner application when launching a composed task or you can provide a spring.cloud.dataflow.server.uri property for the Spring Cloud Data Flow server when it is started. For the latter case, the dataflow.server.uri Composed Task Runner application property is automatically set when a composed task is launched.

In some cases, you may wish to execute an instance of the Composed Task Runner through the Task Launcher sink. In that case, you must configure the Composed Task Runner to use the same datasource that the Spring Cloud Data Flow instance is using. The datasource properties are set with the TaskLaunchRequest through the use of the commandlineArguments or the environmentProperties switches. This is because the Composed Task Runner monitors the task_executions table to check the status of the tasks that it is running. Using information from the table, it determines how it should navigate the graph.

Configuration Options

The ComposedTaskRunner task has the following options:

  • increment-instance-enabled Allows a single ComposedTaskRunner instance to be re-executed without changing the parameters. Default is false which means a ComposedTaskRunner instance can only be executed once with a given set of parameters, if true it can be re-executed. (Boolean, default: false). ComposedTaskRunner is built using Spring Batch and thus upon a successful execution the batch job is considered complete. To launch the same ComposedTaskRunner definition multiple times you must set the increment-instance-enabled property to true or change the parameters for the definition for each launch.

  • interval-time-between-checks The amount of time in millis that the ComposedTaskRunner will wait between checks of the database to see if a task has completed. (Integer, default: 10000). ComposedTaskRunner uses the datastore to determine the status of each child tasks. This interval indicates to ComposedTaskRunner how often it should check the status its child tasks.

  • max-wait-time The maximum amount of time in millis that a individual step can run before the execution of the Composed task is failed (Integer, default: 0). Determines the maximum time each child task is allowed to run before the CTR will terminate with a failure. The default of 0 indicates no timeout.

  • split-thread-allow-core-thread-timeout Specifies whether to allow split core threads to timeout. Default is false; (Boolean, default: false) Sets the policy governing whether core threads may timeout and terminate if no tasks arrive within the keep-alive time, being replaced if needed when new tasks arrive.

  • split-thread-core-pool-size Split’s core pool size. Default is 1; (Integer, default: 1) Each child task contained in a split requires a thread in order to execute. So for example a definition like: <AAA || BBB || CCC> && <DDD || EEE> would require a split-thread-core-pool-size of 3. This is because the largest split contains 3 child tasks. A count of 2 would mean that AAA and BBB would run in parallel but CCC would wait until either AAA or BBB to finish in order to run. Then DDD and EEE would run in parallel.

  • split-thread-keep-alive-seconds Split’s thread keep alive seconds. Default is 60. (Integer, default: 60) If the pool currently has more than corePoolSize threads, excess threads will be terminated if they have been idle for more than the keepAliveTime.

  • split-thread-max-pool-size Split’s maximum pool size. Default is {@code Integer.MAX_VALUE} (Integer, default: <none>). Establish the maximum number of threads allowed for the thread pool.

  • split-thread-queue-capacity Capacity for Split’s BlockingQueue. Default is {@code Integer.MAX_VALUE}. (Integer, default: <none>)

    • If fewer than corePoolSize threads are running, the Executor always prefers adding a new thread rather than queuing.

    • If corePoolSize or more threads are running, the Executor always prefers queuing a request rather than adding a new thread.

    • If a request cannot be queued, a new thread is created unless this would exceed maximumPoolSize, in which case, the task will be rejected.

  • split-thread-wait-for-tasks-to-complete-on-shutdown Whether to wait for scheduled tasks to complete on shutdown, not interrupting running tasks and executing all tasks in the queue. Default is false; (Boolean, default: false)

Note when using the options above as environment variables, convert to uppercase, remove the dash character and replace with the underscore character. For example: increment-instance-enabled would be INCREMENT_INSTANCE_ENABLED.

37.2. The Lifecycle of a Composed Task

The lifecycle of a composed task has three parts:

37.2.1. Creating a Composed Task

The DSL for the composed tasks is used when creating a task definition through the task create command, as shown in the following example:

dataflow:> app register --name timestamp --type task --uri maven://org.springframework.cloud.task.app:timestamp-task:<DESIRED_VERSION>
dataflow:> app register --name mytaskapp --type task --uri file:///home/tasks/mytask.jar
dataflow:> task create my-composed-task --definition "mytaskapp && timestamp"
dataflow:> task launch my-composed-task

In the preceding example, we assume that the applications to be used by our composed task have not been registered yet. Consequently, in the first two steps, we register two task applications. We then create our composed task definition by using the task create command. The composed task DSL in the preceding example, when launched, runs mytaskapp and then runs the timestamp application.

But before we launch the my-composed-task definition, we can view what Spring Cloud Data Flow generated for us. This can be done by executing the task list command, as shown (including its output) in the following example:

dataflow:>task list
╔══════════════════════════╤══════════════════════╤═══════════╗
║        Task Name         │   Task Definition    │Task Status║
╠══════════════════════════╪══════════════════════╪═══════════╣
║my-composed-task          │mytaskapp && timestamp│unknown    ║
║my-composed-task-mytaskapp│mytaskapp             │unknown    ║
║my-composed-task-timestamp│timestamp             │unknown    ║
╚══════════════════════════╧══════════════════════╧═══════════╝

In the example, Spring Cloud Data Flow created three task definitions, one for each of the applications that makes up our composed task (my-composed-task-mytaskapp and my-composed-task-timestamp) as well as the composed task (my-composed-task) definition. We also see that each of the generated names for the child tasks is made up of the name of the composed task and the name of the application, separated by a dash - (as in my-composed-task - mytaskapp).

Task Application Parameters

The task applications that make up the composed task definition can also contain parameters, as shown in the following example:

dataflow:> task create my-composed-task --definition "mytaskapp --displayMessage=hello && timestamp --format=YYYY"

37.2.2. Launching a Composed Task

Launching a composed task is done the same way as launching a stand-alone task, as follows:

task launch my-composed-task

Once the task is launched, and assuming all the tasks complete successfully, you can see three task executions when executing a task execution list, as shown in the following example:

dataflow:>task execution list
╔══════════════════════════╤═══╤════════════════════════════╤════════════════════════════╤═════════╗
║        Task Name         │ID │         Start Time         │          End Time          │Exit Code║
╠══════════════════════════╪═══╪════════════════════════════╪════════════════════════════╪═════════╣
║my-composed-task-timestamp│713│Wed Apr 12 16:43:07 EDT 2017│Wed Apr 12 16:43:07 EDT 2017│0        ║
║my-composed-task-mytaskapp│712│Wed Apr 12 16:42:57 EDT 2017│Wed Apr 12 16:42:57 EDT 2017│0        ║
║my-composed-task          │711│Wed Apr 12 16:42:55 EDT 2017│Wed Apr 12 16:43:15 EDT 2017│0        ║
╚══════════════════════════╧═══╧════════════════════════════╧════════════════════════════╧═════════╝

In the preceding example, we see that my-compose-task launched and that it also launched the other tasks in sequential order. All of them executed successfully with Exit Code as 0.

Passing properties to the child tasks

To set the properties for child tasks in a composed task graph at task launch time, you would use the following format of app.<composed task definition name>.<child task app name>.<property>. Using the following Composed Task definition as an example:

dataflow:> task create my-composed-task --definition "mytaskapp  && mytimestamp"

To have mytaskapp display 'HELLO' and set the mytimestamp timestamp format to 'YYYY' for the Composed Task definition, you would use the following task launch format:

task launch my-composed-task --properties "app.my-composed-task.mytaskapp.displayMessage=HELLO,app.my-composed-task.mytimestamp.timestamp.format=YYYY"

Similar to application properties, the deployer properties can also be set for child tasks using the format format of deployer.<composed task definition name>.<child task app name>.<deployer-property>.

task launch my-composed-task --properties "deployer.my-composed-task.mytaskapp.memory=2048m,app.my-composed-task.mytimestamp.timestamp.format=HH:mm:ss"
Launched task 'a1'
Passing arguments to the composed task runner

Command line arguments for the composed task runner can be passed using --arguments option.

For example:

dataflow:>task create my-composed-task --definition "<aaa: timestamp || bbb: timestamp>"
Created new task 'my-composed-task'

dataflow:>task launch my-composed-task --arguments "--increment-instance-enabled=true --max-wait-time=50000 --split-thread-core-pool-size=4" --properties "app.my-composed-task.bbb.timestamp.format=dd/MM/yyyy HH:mm:ss"
Launched task 'my-composed-task'
Exit Statuses

The following list shows how the Exit Status is set for each step (task) contained in the composed task following each step execution:

  • If the TaskExecution has an ExitMessage, that is used as the ExitStatus.

  • If no ExitMessage is present and the ExitCode is set to zero, then the ExitStatus for the step is COMPLETED.

  • If no ExitMessage is present and the ExitCode is set to any non-zero number, the ExitStatus for the step is FAILED.

37.2.3. Destroying a Composed Task

The command used to destroy a stand-alone task is the same as the command used to destroy a composed task. The only difference is that destroying a composed task also destroys the child tasks associated with it. The following example shows the task list before and after using the destroy command:

dataflow:>task list
╔══════════════════════════╤══════════════════════╤═══════════╗
║        Task Name         │   Task Definition    │Task Status║
╠══════════════════════════╪══════════════════════╪═══════════╣
║my-composed-task          │mytaskapp && timestamp│COMPLETED  ║
║my-composed-task-mytaskapp│mytaskapp             │COMPLETED  ║
║my-composed-task-timestamp│timestamp             │COMPLETED  ║
╚══════════════════════════╧══════════════════════╧═══════════╝
...
dataflow:>task destroy my-composed-task
dataflow:>task list
╔═════════╤═══════════════╤═══════════╗
║Task Name│Task Definition│Task Status║
╚═════════╧═══════════════╧═══════════╝

37.2.4. Stopping a Composed Task

In cases where a composed task execution needs to be stopped, you can do so through the:

  • RESTful API

  • Spring Cloud Data Flow Dashboard

To stop a composed task through the dashboard, select the Jobs tab and click the Stop button next to the job execution that you want to stop.

The composed task run is stopped when the currently running child task completes. The step associated with the child task that was running at the time that the composed task was stopped is marked as STOPPED as well as the composed task job execution.

37.2.5. Restarting a Composed Task

In cases where a composed task fails during execution and the status of the composed task is FAILED, the task can be restarted. You can do so through the:

  • RESTful API

  • The shell

  • Spring Cloud Data Flow Dashboard

To restart a composed task through the shell, launch the task with the same parameters. To restart a composed task through the dashboard, select the Jobs tab and click the Restart button next to the job execution that you want to restart.

Restarting a Composed Task job that has been stopped (through the Spring Cloud Data Flow Dashboard or RESTful API) relaunches the STOPPED child task and then launches the remaining (unlaunched) child tasks in the specified order.

38. Composed Tasks DSL

Composed tasks can be run in three ways:

38.1. Conditional Execution

Conditional execution is expressed by using a double ampersand symbol (&&). This lets each task in the sequence be launched only if the previous task successfully completed, as shown in the following example:

task create my-composed-task --definition "task1 && task2"

When the composed task called my-composed-task is launched, it launches the task called task1 and, if it completes successfully, then the task called task2 is launched. If task1 fails, then task2 does not launch.

You can also use the Spring Cloud Data Flow Dashboard to create your conditional execution, by using the designer to drag and drop applications that are required and connecting them together to create your directed graph, as shown in the following image:

Composed Task Conditional Execution
Figure 4. Conditional Execution

The preceding diagram is a screen capture of the directed graph as it being created by using the Spring Cloud Data Flow Dashboard. You can see that are four components in the diagram that comprise a conditional execution:

  • Start icon: All directed graphs start from this symbol. There is only one.

  • Task icon: Represents each task in the directed graph.

  • End icon: Represents the termination of a directed graph.

  • Solid line arrow: Represents the flow conditional execution flow between:

    • Two applications.

    • The start control node and an application.

    • An application and the end control node.

  • End icon: All directed graphs end at this symbol.

You can view a diagram of your directed graph by clicking the Detail button next to the composed task definition on the Definitions tab.

38.2. Transitional Execution

The DSL supports fine- grained control over the transitions taken during the execution of the directed graph. Transitions are specified by providing a condition for equality based on the exit status of the previous task. A task transition is represented by the following symbol ->.

38.2.1. Basic Transition

A basic transition would look like the following:

task create my-transition-composed-task --definition "foo 'FAILED' → bar 'COMPLETED' → baz"

In the preceding example, foo would launch, and, if it had an exit status of FAILED, the bar task would launch. If the exit status of foo was COMPLETED, baz would launch. All other statuses returned by foo have no effect, and the task would terminate normally.

Using the Spring Cloud Data Flow Dashboard to create the same " basic transition " would resemble the following image:

Composed Task Basic Transition
Figure 5. Basic Transition

The preceding diagram is a screen capture of the directed graph as it being created in the Spring Cloud Data Flow Dashboard. Notice that there are two different types of connectors:

  • Dashed line: Represents transitions from the application to one of the possible destination applications.

  • Solid line: Connects applications in a conditional execution or a connection between the application and a control node (start or end).

To create a transitional connector:

  1. When creating a transition, link the application to each possible destination by using the connector.

  2. Once complete, go to each connection and select it by clicking it.

  3. A bolt icon appears.

  4. Click that icon.

  5. Enter the exit status required for that connector.

  6. The solid line for that connector turns to a dashed line.

38.2.2. Transition With a Wildcard

Wildcards are supported for transitions by the DSL, as shown in the following:

task create my-transition-composed-task --definition "foo 'FAILED' → bar '*' → baz"

In the preceding example, foo would launch, and, if it had an exit status of FAILED, the bar task would launch. For any exit status of foo other than FAILED, baz would launch.

Using the Spring Cloud Data Flow Dashboard to create the same “transition with wildcard” would resemble the following image:

Composed Task Basic Transition with Wildcard
Figure 6. Basic Transition With Wildcard

38.2.3. Transition With a Following Conditional Execution

A transition can be followed by a conditional execution so long as the wildcard is not used, as shown in the following example:

task create my-transition-conditional-execution-task --definition "foo 'FAILED' → bar 'UNKNOWN' → baz && qux && quux"

In the preceding example, foo would launch, and, if it had an exit status of FAILED, the bar task would launch. If foo had an exit status of UNKNOWN, baz would launch. For any exit status of foo other than FAILED or UNKNOWN, qux would launch and, upon successful completion, quux would launch.

Using the Spring Cloud Data Flow Dashboard to create the same “transition with conditional execution” would resemble the following image:

Composed Task Transition with Conditional Execution
Figure 7. Transition With Conditional Execution
In this diagram we see the dashed line (transition) connecting the foo application to the target applications, but a solid line connecting the conditional executions between foo, qux, and quux.

38.3. Split Execution

Splits allow multiple tasks within a composed task to be run in parallel. It is denoted by using angle brackets (<>) to group tasks and flows that are to be run in parallel. These tasks and flows are separated by the double pipe || symbol, as shown in the following example:

task create my-split-task --definition "<foo || bar || baz>"

The preceding example above launches tasks foo, bar and baz in parallel.

Using the Spring Cloud Data Flow Dashboard to create the same “split execution” would resemble the following image:

Composed Task Split
Figure 8. Split

With the task DSL, a user may also execute multiple split groups in succession, as shown in the following example:

`task create my-split-task --definition "<foo || bar || baz> && <qux || quux>"'

In the preceding example, tasks foo, bar, and baz are launched in parallel. Once they all complete, then tasks qux and quux are launched in parallel. Once they complete, the composed task ends. However, if foo, bar, or baz fails, the split containing qux and quux does not launch.

Using the Spring Cloud Data Flow Dashboard to create the same “split with multiple groups” would resemble the following image:

Composed Task Split
Figure 9. Split as a part of a conditional execution

Notice that there is a SYNC control node that is inserted by the designer when connecting two consecutive splits.

38.3.1. Split Containing Conditional Execution

A split can also have a conditional execution within the angle brackets, as shown in the following example:

task create my-split-task --definition "<foo && bar || baz>"

In the preceding example, we see that foo and baz are launched in parallel. However, bar does not launch until foo completes successfully.

Using the Spring Cloud Data Flow Dashboard to create the same " split containing conditional execution " resembles the following image:

Composed Task Split With Conditional Execution
Figure 10. Split with conditional execution

38.3.2. Establishing the proper thread count for splits

Each child task contained in a split requires a thread in order to execute. To set this properly you want to look at your graph and count the split that has the largest number of child tasks, this will be the number of threads you will need to utilize. To set the thread count use the split-thread-core-pool-size property (defaults to 1). So for example a definition like: <AAA || BBB || CCC> && <DDD || EEE> would require a split-thread-core-pool-size of 3. This is because the largest split contains 3 child tasks. A count of 2 would mean that AAA and BBB would run in parallel but CCC would wait until either AAA or BBB to finish in order to run. Then DDD and EEE would run in parallel.

39. Launching Tasks from a Stream

You can launch a task from a stream by using one of the available task-launcher sinks. Currently the platforms supported by the task-launcher sinks are:

task-launcher-local is meant for development purposes only.

A task-launcher sink expects a message containing a TaskLaunchRequest object in its payload. From the TaskLaunchRequest object, the task-launcher obtains the URI of the artifact to be launched, as well as the environment properties, command line arguments, deployment properties, and application name to be used by the task.

The task-launcher-local can be added to the available sinks by executing the app register command, as follows (for the Rabbit Binder, in this case):

app register --name task-launcher-local --type sink --uri maven://org.springframework.cloud.stream.app:task-launcher-local-sink-rabbit:jar:1.2.0.RELEASE

In the case of a Maven-based task that is to be launched, the task-launcher application is responsible for downloading the artifact. You must configure the task-launcher with the appropriate configuration of Maven Properties, such as --maven.remote-repositories.repo1.url=http://repo.spring.io/libs-milestone" to resolve artifacts (in this case against a milestone repo). Note that this repostory can be different than the one used to register the task-launcher application itself.

39.1. TriggerTask

One way to launch a task with the task-launcher is to use the triggertask source. The triggertask source emits a message with a TaskLaunchRequest object that contains the required launch information. The triggertask can be added to the available sources by running the app register command, as follows (for the Rabbit Binder, in this case):

app register --type source --name triggertask --uri maven://org.springframework.cloud.stream.app:triggertask-source-rabbit:1.2.0.RELEASE

For example, to launch the timestamp task once every 60 seconds, the stream implementation would be as follows:

stream create foo --definition "triggertask --triggertask.uri=maven://org.springframework.cloud.task.app:timestamp-task:jar:1.2.0.RELEASE --trigger.fixed-delay=60 --triggertask.environment-properties=spring.datasource.url=jdbc:h2:tcp://localhost:19092/mem:dataflow,spring.datasource.username=sa | task-launcher-local --maven.remote-repositories.repo1.url=http://repo.spring.io/libs-release" --deploy

If you run runtime apps, you can find the log file for the task launcher sink. By using the tail command on that file, you can find the log file for the launched tasks. Setting of triggertask.environment-properties establishes the Data Flow Server’s H2 Database as the database where the task executions will be recorded. You can then see the list of task executions by using the shell command task execution list, as shown (with its output) in the following example:

dataflow:>task execution list
╔════════════════════╤══╤════════════════════════════╤════════════════════════════╤═════════╗
║     Task Name      │ID│         Start Time         │          End Time          │Exit Code║
╠════════════════════╪══╪════════════════════════════╪════════════════════════════╪═════════╣
║timestamp-task_26176│4 │Tue May 02 12:13:49 EDT 2017│Tue May 02 12:13:49 EDT 2017│0        ║
║timestamp-task_32996│3 │Tue May 02 12:12:49 EDT 2017│Tue May 02 12:12:49 EDT 2017│0        ║
║timestamp-task_58971│2 │Tue May 02 12:11:50 EDT 2017│Tue May 02 12:11:50 EDT 2017│0        ║
║timestamp-task_13467│1 │Tue May 02 12:10:50 EDT 2017│Tue May 02 12:10:50 EDT 2017│0        ║
╚════════════════════╧══╧════════════════════════════╧════════════════════════════╧═════════╝

39.2. TaskLaunchRequest-transform

Another way to start a task with the task-launcher would be to create a stream by using the Tasklaunchrequest-transform processor to translate a message payload to a TaskLaunchRequest.

The tasklaunchrequest-transform can be added to the available processors by executing the app register command, as follows (for the Rabbit Binder, in this case):

app register --type processor --name tasklaunchrequest-transform --uri maven://org.springframework.cloud.stream.app:tasklaunchrequest-transform-processor-rabbit:1.2.0.RELEASE

The following example shows the creation of a task that includes the tasklaunchrequest-transform:

stream create task-stream --definition "http --port=9000 | tasklaunchrequest-transform --uri=maven://org.springframework.cloud.task.app:timestamp-task:jar:1.2.0.RELEASE | task-launcher-local --maven.remote-repositories.repo1.url=http://repo.spring.io/libs-release"

39.3. Launching a Composed Task From a Stream

A composed task can be launched with one of the task-launcher sinks as discussed here. Since we use the ComposedTaskRunner directly, we need to set up the task definitions it uses prior to the creation of the composed task launching stream. Suppose we wanted to create the following composed task definition: AAA && BBB. The first step would be to create the task definitions, as shown in the following example:

task create AAA --definition "timestamp"
task create BBB --definition "timestamp"

Now that the task definitions we need for composed task definition are ready, we need to create a stream that launches ComposedTaskRunner. So, in this case, we create a stream with

  • A trigger that emits a message once every 30 seconds

  • A transformer that creates a TaskLaunchRequest for each message received

  • A task-launcher-local sink that launches a the ComposedTaskRunner on our local machine

The stream should resemble the following:

stream create ctr-stream --definition "time --fixed-delay=30 | tasklaunchrequest-transform --uri=maven://org.springframework.cloud.task.app:composedtaskrunner-task:<current release> --command-line-arguments='--graph=AAA&&BBB --increment-instance-enabled=true --spring.datasource.url=...' | task-launcher-local"

In the preceding example, we see that the tasklaunchrequest-transform is establishing two primary components:

  • uri: The URI of the ComposedTaskRunner that is used

  • command-line-arguments: To configure the ComposedTaskRunner

For now, we focus on the configuration that is required to launch the ComposedTaskRunner:

  • graph: this is the graph that is to be executed by the ComposedTaskRunner. In this case it is AAA&&BBB.

  • increment-instance-enabled: This lets each execution of ComposedTaskRunner be unique. ComposedTaskRunner is built by using Spring Batch. Thus, we want a new Job Instance for each launch of the ComposedTaskRunner. To do this, we set increment-instance-enabled to be true.

  • spring.datasource.*: The datasource that is used by Spring Cloud Data Flow, which lets the user track the tasks launched by the ComposedTaskRunner and the state of the job execution. Also, this is so that the ComposedTaskRunner can track the state of the tasks it launched and update its state.

Releases of ComposedTaskRunner can be found here.

40. Sharing Spring Cloud Data Flow’s Datastore with Tasks

As discussed in the Tasks documentation Spring Cloud Data Flow allows a user to view Spring Cloud Task App executions. So in this section we will discuss what is required by a Task Application and Spring Cloud Data Flow to share the task execution information.

40.1. A Common DataStore Dependency

Spring Cloud Data Flow supports many database types out-of-the-box, so all the user typically has to do is declare the spring_datasource_* environment variables to establish what data store Spring Cloud Data Flow will need. So whatever database you decide to use for Spring Cloud Data Flow make sure that the your task also includes that database dependency in its pom.xml or gradle.build file. If the database dependency that is used by Spring Cloud Data Flow is not present in the Task Application, the task will fail and the task execution will not be recorded.

40.2. A Common Data Store

Spring Cloud Data Flow and your task application must access the same datastore instance. This is so that the task executions recorded by the task application can be read by Spring Cloud Data Flow to list them in the Shell and Dashboard views. Also the task app must have read & write privileges to the task data tables that are used by Spring Cloud Data Flow.

Given the understanding of Datasource dependency between Task apps and Spring Cloud Data Flow, let’s review how to apply them in various Task orchestration scenarios.

40.2.1. Simple Task Launch

When launching a task from Spring Cloud Data Flow, Data Flow adds its datasource properties (spring.datasource.url, spring.datasource.driverClassName, spring.datasource.username, spring.datasource.password) to the app properties of the task being launched. Thus a task application will record its task execution information to the Spring Cloud Data Flow repository.

40.2.2. Task Launcher Sink

The Task Launcher Sink allows tasks to be launched via a stream as discussed here. Since tasks launched by the Task Launcher Sink may not want their task executions recorded to the same datastore as Spring Cloud Data Flow, each TaskLaunchRequest received by the Task Launcher Sink must have the required datasource information established as app properties or command line arguments. Both TaskLaunchRequest-Transform and TriggerTask Source are examples of how a source and a processor allow a user to set the datasource properties via the app properties or command line arguments.

40.2.3. Composed Task Runner

Spring Cloud Data Flow allows a user to create a directed graph where each node of the graph is a task application and this is done via the Composed Task Runner. In this case the rules that applied to a Simple Task Launch or Task Launcher Sink apply to the composed task runner as well. All child apps must also have access to the datastore that is being used by the composed task runner Also, All child apps must have the same database dependency as the composed task runner enumerated in their pom.xml or gradle.build file.

40.2.4. Launching a task externally from Spring Cloud Data Flow

Users may wish to launch Spring Cloud Task applications via another method (scheduler for example) but still track the task execution via Spring Cloud Data Flow. This can be done so long as the task applications observe the rules specified here and here.

If a user wishes to use Spring Cloud Data Flow to view their Spring Batch jobs, the user must make sure that their batch application use the @EnableTask annotation and follow the rules enumerated here and here. More information is available here.

41. Scheduling Tasks

Spring Cloud Data Flow lets a user schedule the execution of tasks via a cron expression. A schedule can be created through the RESTful API or the Spring Cloud Data Flow UI.

41.1. The Scheduler

Spring Cloud Data Flow will schedule the execution of its tasks via a scheduling agent that is available on the cloud platform. For example: on the Cloud Foundry platform Spring Cloud Data Flow will use the PCF Scheduler.

Scheduler Architecture Overview
Figure 11. Architectural Overview

41.2. Enabling Scheduling

By default the Spring Cloud Data Flow leaves the scheduling feature disabled. To enable the scheduling feature the following feature properties must be set to true:

  • spring.cloud.dataflow.features.schedules-enabled

  • spring.cloud.dataflow.features.tasks-enabled

41.3. The Lifecycle of a Schedule

The lifecycle of a schedule has 2 parts:

41.3.1. Scheduling a Task Execution

You can schedule a task execution via the:

  • RESTful API

  • Spring Cloud Data Flow Dashboard

To schedule a task from the UI click the Tasks tab at the top of the screen, this will take you to the Task Definitions screen. Then from the Task Definition that you wish to schedule click the "clock" icon associated with task definition you wish to schedule. This will lead you to a Create Schedule(s) screen, where you will create a unique name for the schedule and enter the associated cron expression. Keep in mind you can always create multiple schedules for a single task definition.

41.3.2. Deleting a Schedule

You can delete a schedule via the:

  • RESTful API

  • Spring Cloud Data Flow Dashboard

To delete a schedule through the dashboard, select the Schedule tab under Tasks tab and click the garbage can icon next to the schedule you wish to delete.

Any currently running tasks that were run by the scheduling agent will not be stopped if the schedule is deleted. It only prevents future executions.

Dashboard

This section describes how to use the dashboard of Spring Cloud Data Flow.

42. Introduction

Spring Cloud Data Flow provides a browser-based GUI called the dashboard to manage the following information:

  • Apps: The Apps tab lists all available applications and provides the controls to register/unregister them.

  • Runtime: The Runtime tab provides the list of all running applications.

  • Streams: The Streams tab lets you list, design, create, deploy, and destroy Stream Definitions.

  • Tasks: The Tasks tab lets you list, create, launch, schedule and, destroy Task Definitions.

  • Jobs: The Jobs tab lets you perform batch job related functions.

  • Analytics: The Analytics tab lets you create data visualizations for the various analytics applications.

Upon starting Spring Cloud Data Flow, the dashboard is available at:

For example, if Spring Cloud Data Flow is running locally, the dashboard is available at http://localhost:9393/dashboard.

If you have enabled https, then the dashboard will be located at https://localhost:9393/dashboard. If you have enabled security, a login form is available at http://localhost:9393/dashboard/#/login.

The default Dashboard server port is 9393.

The following image shows the opening page of the Spring Cloud Data Flow dashboard:

The Spring Cloud Data Flow Dashboard
Figure 12. The Spring Cloud Data Flow Dashboard

43. Apps

The Apps section of the dashboard lists all the available applications and provides the controls to register and unregister them (if applicable). It is possible to import a number of applications at once by using the Bulk Import Applications action.

The following image shows a typical list of available apps within the dashboard:

List of available applications
Figure 13. List of Available Applications

43.1. Bulk Import of Applications

The Bulk Import Applications page provides numerous options for defining and importing a set of applications all at once. For bulk import, the application definitions are expected to be expressed in a properties style, as follows:

<type>.<name> = <coordinates>

The following examples show a typical application definitions:

task.timestamp=maven://org.springframework.cloud.task.app:timestamp-task:1.2.0.RELEASE
processor.transform=maven://org.springframework.cloud.stream.app:transform-processor-rabbit:1.2.0.RELEASE

At the top of the bulk import page, a URI can be specified that points to a properties file stored elsewhere, it should contain properties formatted as shown in the previous example. Alternatively, by using the textbox labeled “Apps as Properties”, you can directly list each property string. Finally, if the properties are stored in a local file, the “Select Properties File” option opens a local file browser to select the file. After setting your definitions through one of these routes, click Import.

The following image shows the Bulk Import Applications page:

Bulk Import Applications
Figure 14. Bulk Import Applications

44. Runtime

The Runtime section of the Dashboard application shows the list of all running applications. For each runtime app, the state of the deployment and the number of deployed instances is shown. A list of the used deployment properties is available by clicking on the App Id.

The following image shows an example of the Runtime tab in use:

List of running applications
Figure 15. List of Running Applications

45. Streams

The Streams tab has two child tabs: Definitions and Create Stream. The following topics describe how to work with each one:

45.1. Working with Stream Definitions

The Streams section of the Dashboard includes the Definitions tab that provides a listing of Stream definitions. There you have the option to deploy or undeploy those stream definitions. Additionally, you can remove the definition by clicking on Destroy. Each row includes an arrow on the left, which you can click to see a visual representation of the definition. Hovering over the boxes in the visual representation shows more details about the apps, including any options passed to them.

In the following screenshot, the timer stream has been expanded to show the visual representation:

List of Stream Definitions
Figure 16. List of Stream Definitions

If you click the details button, the view changes to show a visual representation of that stream and any related streams. In the preceding example, if you click details for the timer stream, the view changes to the following view, which clearly shows the relationship between the three streams (two of them are tapping into the timer stream):

Stream Details Page
Figure 17. Stream Details Page

45.2. Creating a Stream

The Streams section of the Dashboard includes the Create Stream tab, which makes available the Spring Flo designer: a canvas application that offers an interactive graphical interface for creating data pipelines.

In this tab, you can:

  • Create, manage, and visualize stream pipelines using DSL, a graphical canvas, or both

  • Write pipelines via DSL with content-assist and auto-complete

  • Use auto-adjustment and grid-layout capabilities in the GUI for simpler and interactive organization of pipelines

You should watch this screencast that highlights some of the "Flo for Spring Cloud Data Flow" capabilities. The Spring Flo wiki includes more detailed content on core Flo capabilities.

The following image shows the Flo designer in use:

Flo for Spring Cloud Data Flo
Figure 18. Flo for Spring Cloud Data Flow

45.3. Deploying a Stream

The stream deploy page includes tabs that provide different ways to setup the deployment properties and deploy the stream. The following screenshots show the stream deploy page for foobar (time | log).

You can define deployments properties using:

  • Form builder tab: a builder which help you to define deployment properties (deployer, application properties…​)

  • Free text tab: a free textarea (key/value pairs)

You can switch between the both views, the form builder provides a more stronger validation of the inputs.

Form builder
Figure 19. The following image shows the form builder
Free text
Figure 20. The following image shows the same properties in the free text

45.4. Creating Fan-In/Fan-Out Streams

In chapter Fan-in and Fan-out you learned how we can support fan-in and fan-out use cases using named destinations. The UI provides dedicated support for named destinations as well:

Fan-in and Fan-out example
Figure 21. Flo for Spring Cloud Data Flow

In this example we have data from an HTTP Source and a JDBC Source that is being sent to the sharedData channel which represents a Fan-in use case. On the other end we have a Cassandra Sink and a File Sink subscribed to the sharedData channel which represents a Fan-out use case.

45.5. Creating a Tap Stream

Creating Taps using the Dashboard is straightforward. Let’s say you have stream consisting of an HTTP Source and a File Sink and you would like to tap into the stream to also send data to a JDBC Sink. In order to create the tap stream simply connect the output connector of the HTTP Source to the JDBC Sink. The connection will be displayed as a dotted line, indicating that you created a tap stream.

Tap stream example
Figure 22. Creating a Tap Stream

The primary stream (HTTP Source to File Sink) will be automatically named, in case you did not provide a name for the stream, yet. When creating tap streams, the primary stream must always be explicitly named. In the picture above, the primary stream was named HTTP_INGEST.

Using the Dashboard, you can also switch the primary stream to become the secondary tap stream.

Switch tap stream to primary stream
Figure 23. Change Primary Stream to Secondary Tap Stream

Simply hover over the existing primary stream, the line between HTTP Source and File Sink. Several control icons will appear, and by clicking on the icon labeled Switch to/from tap, you change the primary stream into a tap stream. Do the same for the tap stream and switch it to a primary stream.

End result of switching the tap stream to a primary stream
Figure 24. End Result of Switching the Primary Stream
When interacting directly with named destinations, there can be "n" combinations (Inputs/Outputs). This allows you to create complex topologies involving a wide variety of data sources and destinations.

46. Tasks

The Tasks section of the Dashboard currently has three tabs:

46.1. Apps

Each app encapsulates a unit of work into a reusable component. Within the Data Flow runtime environment, apps let users create definitions for streams as well as tasks. Consequently, the Apps tab within the Tasks section lets users create task definitions.

You can also use this tab to create Batch Jobs.

The following image shows a typical list of task apps:

List of Task Apps
Figure 25. List of Task Apps

On this screen, you can perform the following actions:

  • View details, such as the task app options.

  • Create a task definition from the respective app.

46.1.1. View Task App Details

On this page you can view the details of a selected task app, including the list of available options (properties) for that app.

46.1.2. Create a Task Definition

List of Task Apps

At a minimum, you must provide a name for the new definition. You also have the option to specify various properties that are used during the deployment of the app.

Each parameter is included only if the Include checkbox is selected.

46.2. Definitions

This page lists the Data Flow task definitions and provides actions to launch or destroy those tasks. It also provides a shortcut operation to define one or more tasks with simple textual input, indicated by the Bulk Define Tasks button.

The following image shows the Definitions page:

List of Task Definitions
Figure 26. List of Task Definitions

46.2.1. Creating Composed Task Definitions

The dashboard includes the Create Composed Task tab, which provides an interactive graphical interface for creating composed tasks.

In this tab, you can:

  • Create and visualize composed tasks using DSL, a graphical canvas, or both.

  • Use auto-adjustment and grid-layout capabilities in the GUI for simpler and interactive organization of the composed task.

On the Create Composed Task screen, you can define one or more task parameters by entering both the parameter key and the parameter value.

Task parameters are not typed.

The following image shows the composed task designer:

Composed Task Designer
Figure 27. Composed Task Designer

46.2.2. Launching Tasks

Once the task definition has been created, the tasks can be launched through the dashboard. To do so, click the Definitions tab and select the task you want to launch by pressing Launch.

46.3. Executions

The Executions tab shows the current running and completed tasks.

The following image shows the Executions tab:

List of Task Executions
Figure 28. List of Task Executions

47. Jobs

The Jobs section of the Dashboard lets you inspect batch jobs. The main section of the screen provides a list of job executions. Batch jobs are tasks that each execute one or more batch jobs. Each job execution has a reference to the task execution ID (in the Task Id column).

The list of Job Executions also shows the state of the underlying Job Definition. Thus, if the underlying definition has been deleted, “No definition found” appears in the Status column.

You can take the following actions for each job:

  • Restart (for failed jobs).

  • Stop (for running jobs).

  • View execution details.

Note: Clicking the stop button actually sends a stop request to the running job, which may not immediately stop.

The following image shows the Jobs page:

List of Job Executions
Figure 29. List of Job Executions

47.1. Job Execution Details

After having launched a batch job, the Job Execution Details page will show information about the job.

The following image shows the Job Execution Details page:

Job Execution Details
Figure 30. Job Execution Details

The Job Execution Details page contains a list of the executed steps. You can further drill into the details of each step’s execution by clicking the magnifying glass icon.

47.2. Step Execution Details

The Step Execution Details page provides information about an individual step within a job.

The following image shows the Step Execution Details page:

Step Execution History
Figure 31. Step Execution Details

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

The Step Execution Details screen provides a complete list of all Step Execution Context key/value pairs.

For exceptions, the Exit Description field contains additional error information. However, this field can have a maximum of 2500 characters. Therefore, in the case of long exception stack traces, trimming of error messages may occur. When that happens, refer to the server log files for further details.

47.3. Step Execution Progress

On this screen, you can see a progress bar indicator in regards to the execution of the current step. Under the Step Execution History, you can also view various metrics associated with the selected step, such as duration, read counts, write counts, and others.

48. Scheduling

48.1. Creating or deleting a Schedule from the Task Definition’s page

From the Task Definitions page a user can create or delete a schedule for a specific task definition.

Task Definitions with Schedule Controls
Figure 32. Task Definitions with Schedule controls

On this screen you can perform the following actions:

  • The user can click the clock icon and this will take you to the Schedule Creation screen.

  • The user can click the clock icon with the x to the upper right to delete the schedule(s) associated with the task definition.

48.2. Creating a Schedule

Create Schedule for Task Execution
Figure 33. Create Schedule for Task Execution

Once the user clicks the clock icon on the Task Definition screen, Spring Cloud Data Flow will take the user to the Schedule Creation screen. On this screen a user can establish the schedule name, the cron expression as well as establish the properties and arguments to be used when the task is launched by this schedule.

48.3. Listing Available Schedules

List Available Schedules
Figure 34. List Available Schedules

On this screen you can perform the following actions:

  • Delete a schedule

  • Get details for a schedule

49. Analytics

The Analytics page of the Dashboard provides the following data visualization capabilities for the various analytics applications available in Spring Cloud Data Flow:

  • Counters

  • Field-Value Counters

  • Aggregate Counters

For example, if you create a stream with a Counter application, you can create the corresponding graph from within the Dashboard tab. To do so:

  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.

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.

REST API Guide

You can find the documentation about the Data Flow REST API in the core documentation.

Appendices

Having trouble with Spring Cloud Data Flow, We’d like to help!

Appendix A: ‘How-to’ guides

A.1. Logging

Spring Cloud Data Flow is built upon several Spring projects, but ultimately the dataflow-server is a Spring Boot app, so the logging techniques that apply to any Spring Boot application are applicable here as well.

While troubleshooting, following are the two primary areas where enabling the DEBUG logs could be useful.

A.1.1. Deployment Logs

Spring Cloud Data Flow builds upon Spring Cloud Deployer SPI and the platform specific dataflow-server uses the respective SPI implementations. Specifically, if we were to troubleshoot deployment specific issues; such as the network errors, it’d be useful to enable the DEBUG logs at the underlying deployer and the libraries used by it.

  1. For instance, if you’d like to enable DEBUG logs for the kubernetes-deployer, you’d be starting the server with following environment variable set.

    LOGGING_LEVEL_ORG_SPRINGFRAMEWORK_CLOUD_DEPLOYER_SPI_KUBERNETES=DEBUG

A.1.2. Application Logs

The streaming applications in Spring Cloud Data Flow are Spring Boot applications and they can be independently setup with logging configurations.

For instance, if you’d have to troubleshoot the header and payload specifics that are being passed around source, processor and sink channels, you’d be deploying the stream with the following options.

dataflow:>stream create foo --definition "http --logging.level.org.springframework.integration=DEBUG | transform --logging.level.org.springframework.integration=DEBUG | log --logging.level.org.springframework.integration=DEBUG" --deploy

(where, org.springframework.integration is the global package for everything Spring Integration related, which is responsible for messaging channels)

These properties can also be specified via deployment properties when deploying the stream.

dataflow:>stream deploy foo --properties "app.*.logging.level.org.springframework.integration=DEBUG"

Appendix B: Data Flow Template

As described in the previous chapter, Spring Cloud Data Flow’s functionality is completely exposed through REST endpoints. While you can use those endpoints directly, Spring Cloud Data Flow also provides a Java-based API, which makes using those REST endpoints even easier.

The central entry point is the DataFlowTemplate class in the org.springframework.cloud.dataflow.rest.client package.

This class implements the DataFlowOperations interface and delegates to the following sub-templates that provide the specific functionality for each feature-set:

Interface Description

StreamOperations

REST client for stream operations

CounterOperations

REST client for counter operations

FieldValueCounterOperations

REST client for field value counter operations

AggregateCounterOperations

REST client for aggregate counter operations

TaskOperations

REST client for task operations

JobOperations

REST client for job operations

AppRegistryOperations

REST client for app registry operations

CompletionOperations

REST client for completion operations

RuntimeOperations

REST Client for runtime operations

When the DataFlowTemplate is being initialized, the sub-templates can be discovered through the REST relations, which are provided by HATEOAS.[1]

If a resource cannot be resolved, the respective sub-template results in NULL. A common cause is that Spring Cloud Data Flow offers for specific sets of features to be enabled/disabled when launching. For more information, see “Feature Toggles”.

B.1. Using the Data Flow Template

When you use the Data Flow Template, the only needed Data Flow dependency is the Spring Cloud Data Flow Rest Client, as shown in the following Maven snippet:

<dependency>
  <groupId>org.springframework.cloud</groupId>
  <artifactId>spring-cloud-dataflow-rest-client</artifactId>
  <version>1.6.0.RELEASE</version>
</dependency>

With that dependency, you get the DataFlowTemplate class as well as all the dependencies needed to make calls to a Spring Cloud Data Flow server.

When instantiating the DataFlowTemplate, you also pass in a RestTemplate. Please be aware that the needed RestTemplate requires some additional configuration to be valid in the context of the DataFlowTemplate. When declaring a RestTemplate as a bean, the following configuration suffices:

  @Bean
  public static RestTemplate restTemplate() {
    RestTemplate restTemplate = new RestTemplate();
    restTemplate.setErrorHandler(new VndErrorResponseErrorHandler(restTemplate.getMessageConverters()));
    for(HttpMessageConverter<?> converter : restTemplate.getMessageConverters()) {
      if (converter instanceof MappingJackson2HttpMessageConverter) {
        final MappingJackson2HttpMessageConverter jacksonConverter =
            (MappingJackson2HttpMessageConverter) converter;
        jacksonConverter.getObjectMapper()
            .registerModule(new Jackson2HalModule())
            .addMixIn(JobExecution.class, JobExecutionJacksonMixIn.class)
            .addMixIn(JobParameters.class, JobParametersJacksonMixIn.class)
            .addMixIn(JobParameter.class, JobParameterJacksonMixIn.class)
            .addMixIn(JobInstance.class, JobInstanceJacksonMixIn.class)
            .addMixIn(ExitStatus.class, ExitStatusJacksonMixIn.class)
            .addMixIn(StepExecution.class, StepExecutionJacksonMixIn.class)
            .addMixIn(ExecutionContext.class, ExecutionContextJacksonMixIn.class)
            .addMixIn(StepExecutionHistory.class, StepExecutionHistoryJacksonMixIn.class);
      }
    }
    return restTemplate;
  }

Now you can instantiate the DataFlowTemplate with the following code:

DataFlowTemplate dataFlowTemplate = new DataFlowTemplate(
    new URI("http://localhost:9393/"), restTemplate);         (1)
1 The URI points to the ROOT of your Spring Cloud Data Flow Server.

Depending on your requirements, you can now make calls to the server. For instance, if you want to get a list of currently available applications you can run the following code:

PagedResources<AppRegistrationResource> apps = dataFlowTemplate.appRegistryOperations().list();

System.out.println(String.format("Retrieved %s application(s)",
    apps.getContent().size()));

for (AppRegistrationResource app : apps.getContent()) {
  System.out.println(String.format("App Name: %s, App Type: %s, App URI: %s",
    app.getName(),
    app.getType(),
    app.getUri()));
}

Appendix C: Spring XD to SCDF

This appendix describes how to migrate from Spring XD to Spring Cloud Data Flow, along with some tips and tricks that may be helpful.

C.1. Terminology Changes

The following table describes the changes in terminology from Spring XD to Spring Cloud Data Flow:

Old New

XD-Admin

Server (implementations: local, Cloud Foundry, Apache Yarn, Kubernetes, and Apache Mesos)

XD-Container

N/A

Modules

Applications

Admin UI

Dashboard

Message Bus

Binders

Batch / Job

Task

C.2. Modules to Applications

If you have custom Spring XD modules, you need refactor them to use Spring Cloud Stream and Spring Cloud Task annotations. As part of that work, you need to update dependencies and build them as normal Spring Boot applications.

C.2.1. Custom Applications

As you convert custom applications, keep the following information in mind:

C.2.2. Application Registration

As you register your applications, keep the following information in mind:

  • Custom Stream/Task applications require being installed to a Maven repository for local, Yarn, and Cloud Foundry implementations or, as docker images, when deploying to Kubernetes or Mesos. Other than Maven and docker resolution, you can also resolve application artifacts from http, file, or as hdfs coordinates.

  • Unlike Spring XD, you do not have to upload the application bits while registering custom applications anymore. Instead, you need to register the application coordinates that are hosted in the Maven repository or by other means as discussed in the previous bullet.

  • By default, none of the out-of-the-box applications are preloaded. It is intentionally designed to provide the flexibility to register apps as you find appropriate for the given use-case requirement.

  • Depending on the binder choice, you can manually add the appropriate binder dependency to build applications specific to that binder-type. Alternatively, you can follow the Spring Initializr procedure to create an application with binder embedded in it.

C.2.3. Application Properties

As you modify your applications' properties, keep the following information in mind:

  • Counter-sink:

    • The peripheral redis is not required in Spring Cloud Data Flow. If you intend to use the counter-sink, then redis is required, and you need to have your own running redis cluster.

  • field-value-counter-sink:

    • The peripheral redis is not required in Spring Cloud Data Flow. If you intend to use the field-value-counter-sink, then redis becomes required, and you need to have your own running redis cluster.

  • Aggregate-counter-sink:

    • The peripheral redis is not required in Spring Cloud Data Flow. If you intend to use the aggregate-counter-sink, then redis becomes required, and you need to have your own running redis cluster.

C.3. Message Bus to Binders

Terminology wise, in Spring Cloud Data Flow, the message bus implementation is commonly referred to as binders.

C.3.1. Message Bus

Similar to Spring XD, Spring Cloud Data Flow includes an abstraction that you can use to extend the binder interface. By default, we take the opinionated view of Apache Kafka and RabbitMQ as the production-ready binders. They are available as GA releases.

C.3.2. Binders

Selecting a binder requires providing the right binder dependency in the classpath. If you choose Kafka as the binder, you need to register stream applications that are pre-built with Kafka binder in it. If you to create a custom application with Kafka binder, you need add the following dependency in the classpath:

<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>spring-cloud-stream-binder-kafka</artifactId>
    <version>1.0.2.RELEASE</version>
</dependency>
  • Spring Cloud Stream supports Apache Kafka and RabbitMQ. All binder implementations are maintained and managed in their individual repositories.

  • Every Stream/Task application can be built with the binder implementation of your choice. All the out-of-the-box applications are pre-built for both Kafka and Rabbit and are readily available for use as Maven artifacts (Spring Cloud Stream or Spring Cloud Task) or as Docker images (Spring Cloud Stream or Spring Cloud Task). Changing the binder requires selecting the right binder dependency. Alternatively, you can download the pre-built application from this version of Spring Initializr with the desired “binder-starter” dependency.

C.3.3. Named Channels

Fundamentally, all the messaging channels are backed by pub/sub semantics. Unlike Spring XD, the messaging channels are backed only by topics or topic-exchange and there is no representation of queues in the new architecture.

  • ${xd.module.index} is no longer supported. Instead, you can directly interact with named destinations.

  • stream.index changes to :<stream-name>.<label/app-name>

    • For example, ticktock.0 changes to :ticktock.time.

  • “topic/queue” prefixes are not required to interact with named-channels.

    • For example, topic:mytopic changes to :mytopic.

    • For example, stream create stream1 --definition ":foo > log".

C.3.4. Directed Graphs

If you build non-linear streams, you can take advantage of named destinations to build directed graphs.

Consider the following example from Spring XD:

stream create f --definition "queue:foo > transform --expression=payload+'-sample1' | log" --deploy
stream create b --definition "queue:bar > transform --expression=payload+'-sample2' | log" --deploy
stream create r --definition "http | router --expression=payload.contains('a')?'queue:sample1':'queue:sample2'" --deploy

You can do the following in Spring Cloud Data Flow:

stream create f --definition ":foo > transform --expression=payload+'-sample1' | log" --deploy
stream create b --definition ":bar > transform --expression=payload+'-sample2' | log" --deploy
stream create r --definition "http | router --expression=payload.contains('a')?'sample1':'sample2'" --deploy

C.4. Batch to Tasks

A Task, by definition, is any application that does not run forever, and they end at some point. Tasks include Spring Batch jobs. Task applications can be used for on-demand use cases, such as database migration, machine learning, scheduled operations, and others. With Spring Cloud Task, you can build Spring Batch jobs as microservice applications.

  • Spring Batch jobs from Spring XD are being refactored to Spring Boot applications, also known as Spring Cloud Task applications.

  • Unlike Spring XD, these tasks do not require explicit deployment. Instead, a task is ready to be launched directly once the definition is declared.

C.5. Shell and DSL Command Changes

The following table shows the changes to shell and DSL commands:

Old Command New Command

module upload

app register / app import

module list

app list

module info

app info

admin config server

dataflow config server

job create

task create

job launch

task launch

job list

task list

job status

task status

job display

task display

job destroy

task destroy

job execution list

task execution list

runtime modules

runtime apps

C.6. REST API Changes

The following table shows the changes to the REST API:

Old API New API

/modules

/apps

/runtime/modules

/runtime/apps

/runtime/modules/{moduleId}

/runtime/apps/{appId}

/jobs/definitions

/task/definitions

/jobs/deployments

/task/deployments

C.7. UI (including Flo)

The Admin-UI is now named Dashboard. The URI for accessing the Dashboard is changed from localhost:9393/admin-ui to localhost:9393/dashboard.

  • Apps (a new view): Lists all the registered applications that are available for use. This view includes details such as the URI and the properties supported by each application. You can also register/unregister applications from this view.

  • Runtime (was Container): Container changes to Runtime. The notion of xd-container is gone, replaced by out-of-the-box applications running as autonomous Spring Boot applications. The Runtime tab displays the applications running in the runtime platforms (implementations: Cloud Foundry, Apache Yarn, Apache Mesos, or Kubernetes). You can click on each application to review relevant details, such as where it is running, what resources it uses, and other details.

  • Spring Flo is now an OSS product. Flo for Spring Cloud Data Flow’s “Create Stream” is now the designer-tab in the Dashboard.

  • Tasks (a new view):

    • The “Modules” sub-tab is renamed to “Apps”.

    • The “Definitions” sub-tab lists all the task definitions, including Spring Batch jobs that are orchestrated as tasks.

    • The “Executions” sub-tab lists all the task execution details in a fashion similar to the listing of Spring XD’s Job executions.

C.8. Architecture Components

Spring Cloud Data Flow comes with a significantly simplified architecture. In fact, when compared with Spring XD, you need fewer peripherals to use Spring Cloud Data Flow.

C.8.1. ZooKeeper

ZooKeeper is not used in the new architecture.

C.8.2. RDBMS

Spring Cloud Data Flow uses an RDBMS instead of Redis for stream/task definitions, application registration, and for job repositories. The default configuration uses an embedded H2 instance, but Oracle, DB2, SqlServer, MySQL/MariaDB, PostgreSQL, H2, and HSQLDB databases are supported. To use Oracle, DB2, and SqlServer, you need to create your own Data Flow Server by using Spring Initializr and add the appropriate JDBC driver dependency.

C.8.3. Redis

Running a Redis cluster is only required for analytics functionality. Specifically, when you use the counter-sink, field-value-counter-sink, or aggregate-counter-sink applications, you also need to have a running instance of Redis cluster.

C.8.4. Cluster Topology

Spring XD’s xd-admin and xd-container server components are replaced by stream and task applications that are themselves running as autonomous Spring Boot applications. The applications run natively on various platforms, including Cloud Foundry, Apache YARN, Apache Mesos, and Kubernetes. You can develop, test, deploy, scale up or down, and interact with (Spring Boot) applications individually, and they can evolve in isolation.

C.9. Central Configuration

To support centralized and consistent management of an application’s configuration properties, Spring Cloud Config client libraries have been included in the Spring Cloud Data Flow server as well as the Spring Cloud Stream applications provided by the Spring Cloud Stream App Starters. You can also pass common application properties to all streams when the Data Flow Server starts.

C.10. Distribution

Spring Cloud Data Flow is a Spring Boot application. Depending on the platform of your choice, you can download the respective release uber jar and deploy or push it to the runtime platform (Cloud Foundry, Apache Yarn, Kubernetes, or Apache Mesos). For example, if you run Spring Cloud Data Flow on Cloud Foundry, you can download the Cloud Foundry server implementation and do a cf push, as explained in the Cloud Foundry Reference Guide.

C.11. Hadoop Distribution Compatibility

The hdfs-sink application builds upon Spring Hadoop 2.4.0 release, so this application is compatible with the following Hadoop distributions:

  • Cloudera: cdh5

  • Pivotal Hadoop: phd30

  • Hortonworks Hadoop: hdp24

  • Hortonworks Hadoop: hdp23

  • Vanilla Hadoop: hadoop26

  • Vanilla Hadoop: 2.7.x (default)

C.12. YARN Deployment

Spring Cloud Data Flow can be deployed and used with Apche YARN in two different ways:

  • Deploy the server directly in a YARN cluster.

  • Use the Apache Ambari plugin to provision Spring Cloud Data Flow as a service.

C.13. Use Case Comparison

The remainder of this appendix reviews some use cases to show the differences between Spring XD and Spring Cloud Data Flow.

C.13.1. Use Case #1: Ticktock

This use case assumes that you have already downloaded both the XD and the SCDF distributions.

Description: Simple ticktock example using local/singlenode.

The following table describes the differences:

Spring XD Spring Cloud Data Flow

Start an xd-singlenode server from CLI

→ xd-singlenode

Start a binder of your choice

Start a local-server implementation of SCDF from the CLI

→ java -jar spring-cloud-dataflow-server-local-1.0.0.BUILD-SNAPSHOT.jar

Start an xd-shell server from the CLI

→ xd-shell

Start dataflow-shell server from the CLI

→ java -jar spring-cloud-dataflow-shell-1.0.0.BUILD-SNAPSHOT.jar

Create ticktock stream

xd:>stream create ticktock --definition “time | log” --deploy

Create ticktock stream

dataflow:>stream create ticktock --definition “time | log” --deploy

Review ticktock results in the xd-singlenode server console

Review ticktock results by using the tail commang to view the ticktock.log/stdout_log application logs

C.13.2. Use Case #2: Stream with Custom Module or Application

This use case assumes that you have already downloaded both the XD and the SCDF distributions.

Description: Stream with custom module or application.

The following table describes the differences:

Spring XD Spring Cloud Data Flow

Start an xd-singlenode server from CLI

→ xd-singlenode

Start a binder of your choice

Start a local-server implementation of SCDF from the CLI

→ java -jar spring-cloud-dataflow-server-local-1.0.0.BUILD-SNAPSHOT.jar

Start an xd-shell server from the CLI

→ xd-shell

Start dataflow-shell server from the CLI

→ java -jar spring-cloud-dataflow-shell-1.0.0.BUILD-SNAPSHOT.jar

Register a custom “processor” module to transform the payload to the desired format

xd:>module upload --name toupper --type processor --file <CUSTOM_JAR_FILE_LOCATION>

Register custom “processor” application to transform payload to a desired format

dataflow:>app register --name toupper --type processor --uri <MAVEN_URI_COORDINATES>

Create a stream with a custom module

xd:>stream create testupper --definition “http | toupper | log” --deploy

Create a stream with custom application

dataflow:>stream create testupper --definition “http | toupper | log” --deploy

Review results in the xd-singlenode server console

Review results by using the tail command to view the testupper.log/stdout_log application logs

C.13.3. Use Case #3: Batch Job

This use case assumes that you have already downloaded both the XD and the SCDF distributions.

Description: batch-job.

Spring XD Spring Cloud Data Flow

Start an xd-singlenode server from CLI

→ xd-singlenode

Start a local-server implementation of SCDF from the CLI

→ java -jar spring-cloud-dataflow-server-local-1.0.0.BUILD-SNAPSHOT.jar

Start an xd-shell server from the CLI

→ xd-shell

Start dataflow-shell server from the CLI

→ java -jar spring-cloud-dataflow-shell-1.0.0.BUILD-SNAPSHOT.jar

Register a custom “batch job” module

xd:>module upload --name simple-batch --type job --file <CUSTOM_JAR_FILE_LOCATION>

Register a custom “batch-job” as task application

dataflow:>app register --name simple-batch --type task --uri <MAVEN_URI_COORDINATES>

Create a job with custom batch-job module

xd:>job create batchtest --definition “simple-batch”

Create a task with a custom batch-job application

dataflow:>task create batchtest --definition “simple-batch”

Deploy job

xd:>job deploy batchtest

NA

Launch job

xd:>job launch batchtest

Launch task

dataflow:>task launch batchtest

Review results in the xd-singlenode server console as well as the Jobs tab in the UI (executions sub-tab should include all step details)

Review results by using the tail command to view the batchtest/stdout_log application logs as well as the Task tab in UI (the executions sub-tab should include all step details)

Appendix D: Building

To build the source, you need to install JDK 1.8.

The build uses the Maven wrapper so that you do not have to install a specific version of Maven. To enable the tests for Redis, run the server before building. More information on how to run Redis appears later in this appendix.

The main build command is as follows:

$ ./mvnw clean install

If you like, you can add '-DskipTests' to avoid running the tests.

You can also install Maven (>=3.3.3) yourself and run the mvn command in place of ./mvnw in the examples below. If you do that, you also might need to add -P spring if your local Maven settings do not contain repository declarations for Spring pre-release artifacts.
You might need to increase the amount of memory available to Maven by setting a MAVEN_OPTS environment variable with a value similar to -Xmx512m -XX:MaxPermSize=128m. We try to cover this in the .mvn configuration, so, if you find you have to do it to make a build succeed, please raise a ticket to get the settings added to source control.

D.1. Documentation

There is a full profile that generates documentation. You can build only the documentation by using the following command:

$ ./mvnw clean package -DskipTests -P full -pl spring-cloud-dataflow-server-kubernetes-docs -am

D.2. Working with the Code

If you do not have an IDE preference, we recommend that you use Spring Tools Suite or Eclipse when working with the code. We use the m2eclipse Eclipse plugin for Maven support. Other IDEs and tools generally also work without issue.

D.2.1. Importing into Eclipse with m2eclipse

We recommend the m2eclipe eclipse plugin when working with Eclipse. If you do not already have m2eclipse installed, it is available from the Eclipse marketplace.

Unfortunately, m2e does not yet support Maven 3.3. Consequently, once the projects are imported into Eclipse, you also need to tell m2eclipse to use the .settings.xml file for the projects. If you do not do this, you may see many different errors related to the POMs in the projects. To do so:

  1. Open your Eclipse preferences.

  2. Expand the Maven preferences.

  3. Select User Settings.

  4. In the User Settings field, click Browse and navigate to the Spring Cloud project you imported.

  5. Select the .settings.xml file in that project.

  6. Click Apply.

  7. Click OK.

Alternatively, you can copy the repository settings from Spring Cloud’s .settings.xml file into your own ~/.m2/settings.xml.

D.2.2. Importing into Eclipse without m2eclipse

If you prefer not to use m2eclipse, you can generate Eclipse project metadata by using the following command:

$ ./mvnw eclipse:eclipse

The generated Eclipse projects can be imported by selecting Import existing projects from the File menu.

Appendix E: Contributing

Spring Cloud is released under the non-restrictive Apache 2.0 license and follows a very standard Github development process, using Github tracker for issues and merging pull requests into master. If you want to contribute even something trivial, please do not hesitate, but follow the guidelines below.

E.1. Sign the Contributor License Agreement

Before we accept a non-trivial patch or pull request, we need you to sign the contributor’s agreement. Signing the contributor’s agreement does not grant anyone commit rights to the main repository, but it does mean that we can accept your contributions, and you will get an author credit if we do. Active contributors might be asked to join the core team and be given the ability to merge pull requests.

E.2. Code Conventions and Housekeeping

None of the following guidelines is essential for a pull request, but they all help your fellow developers understand and work with your code. They can also be added after the original pull request but before a merge.

  • Use the Spring Framework code format conventions. If you use Eclipse, you can import formatter settings by using the eclipse-code-formatter.xml file from the Spring Cloud Build project. If using IntelliJ, you can use the Eclipse Code Formatter Plugin to import the same file.

  • Make sure all new .java files have a simple Javadoc class comment with at least an @author tag identifying you, and preferably at least a paragraph describing the class’s purpose.

  • Add the ASF license header comment to all new .java files (to do so, copy from existing files in the project).

  • Add yourself as an @author to the .java files that you modify substantially (more than cosmetic changes).

  • Add some Javadocs and, if you change the namespace, some XSD doc elements.

  • A few unit tests would help a lot as well. Someone has to do it, and your fellow developers appreciate the effort.

  • If no one else uses your branch, rebase it against the current master (or other target branch in the main project).

  • When writing a commit message, follow these conventions. If you fix an existing issue, add Fixes gh-XXXX (where XXXX is the issue number) at the end of the commit message.


1. HATEOAS stands for Hypermedia as the Engine of Application State