This section covers auto-configuration of Spring Session using either Apache Geode or Pivotal GemFire to manage (HTTP) Session state in a reliable (consistent), highly-available (replicated) and clustered manner.
Spring Session provides an API and several implementations for managing a user’s session
information. It has the ability to replace the javax.servlet.http.HttpSession
in an application container neutral
way along with proving Session IDs in HTTP headers to work with RESTful APIs.
Furthermore, Spring Session provides the ability to keep the HttpSession alive even when working with WebSockets and reactive Spring WebFlux WebSessions.
A full discussion of Spring Session is beyond the scope of this document, and the reader is encouraged to learn more by reading the docs and reviewing the samples.
Of course, Spring Boot for Apache Geode & Pivotal GemFire provides auto-configuration support to configure either Apache Geode or Pivotal GemFire as the user’s session information management provider and store when Spring Session for Apache Geode or Pivotal GemFire is on your Spring Boot application’s classpath.
![]() | Tip |
---|---|
You can learn more about Spring Session for Apache Geode & Pivotal GemFire in the docs. |
There is nothing special that you need to do in order to use either Apache Geode or Pivotal GemFire as a Spring Session provider, managing the (HTTP) Session state of your Spring Boot application.
Simply include the appropriate Spring Session dependency on your Spring Boot application’s classpath, for example:
Maven dependency declaration.
<dependency> <groupId>org.springframework.session</groupId> <artifactId>spring-session-data-geode</artifactId> <version>2.3.0.M2</version> </dependency>
Alternatively, you may declare the provided spring-geode-starter-session
dependency in your Spring Boot application
Maven POM or Gradle build file:
Maven dependency declaration.
<dependency> <groupId>org.springframework.geode</groupId> <artifactId>spring-geode-starter-session</artifactId> <version>1.3.0.M2</version> </dependency>
![]() | Tip |
---|---|
You may replace Apache Geode with Pivotal Cloud Cache or Pivotal GemFire by changing the artifact ID
from |
After declaring the required Spring Session dependency, then begin your Spring Boot application as you normally would:
Spring Boot Application.
@SpringBootApplication public class MySpringBootApplication { public static void main(String[] args) { SpringApplication.run(MySpringBootApplication.class, args); } ... }
That is it!
Of course, you are free to create application-specific, Spring Web MVC Controllers
to interact with the HttpSession
as needed by your application:
Application Controller using HttpSession.
@Controller class MyApplicationController { @GetRequest(...) public String processGet(HttpSession session) { // interact with HttpSession } }
The HttpSession
is replaced by a Spring managed Session
that will be stored in either Apache Geode
or Pivotal GemFire, or even Pivotal Cloud Cache.
By default, Spring Boot for Apache Geode & Pivotal GemFire (SBDG) applies reasonable and sensible defaults when configuring Apache Geode or Pivotal GemFire as the provider in Spring Session.
So, for instance, by default, SBDG set the session expiration timeout to 30 minutes. It also uses a
ClientRegionShortcut.PROXY
as the client Region data management policy for the Apache Geode/Pivotal GemFire
Region managing the (HTTP) Session state when the Spring Boot application is using a ClientCache
, which it does
by default.
However, what if the defaults are not sufficient for your application requirements?
Spring Session for Apache Geode/Pivotal GemFire publishes well-known configuration properties for each of the various Spring Session configuration options when using Apache Geode or Pivotal GemFire as the (HTTP) Session state management provider.
You may specify any of these properties in a Spring Boot application.properties
file to adjust Spring Session’s
configuration when using Apache Geode or Pivotal GemFire.
In addition to the properties provided in and by Spring Session for Apache Geode/Pivotal GemFire,
Spring Boot for Apache Geode/Pivotal GemFire also recognizes and respects the spring.session.timeout
property
as well as the server.servlet.session.timeout
property as discussed here.
![]() | Tip |
---|---|
|
Spring Session for Apache Geode/Pivotal GemFire also provides the
SpringSessionGemFireConfigurer
callback interface, which can be declared in your Spring ApplicationContext
to programmatically control
the configuration of Spring Session when using Apache Geode or Pivotal GemFire.
The SpringSessionGemFireConfigurer
, when declared in the Spring ApplicationContext
, takes precedence over any of the
Spring Session (for Apache Geode/Pivotal GemFire) configuration properties, and will effectively override them when both
are present.
More information on using the SpringSessionGemFireConfigurer
can be found in the
docs.
There may be cases where you do not want your Spring Boot application to manage (HTTP) Session state using either
Apache Geode or Pivotal GemFire. In certain cases, you may be using another Spring Session provider,
such as Redis, to cache and manage your Spring Boot application’s (HTTP) Session state, while, even in other cases,
you do not want to use Spring Session to manage your (HTTP) Session state at all. Rather, you prefer to use your
Web Server’s (e.g. Tomcat) HttpSession
state management.
Either way, you can specifically call out your Spring Session provider using the spring.session.store-type
property
in application.properties
, as follows:
Use Redis as the Spring Session Provider.
#application.properties spring.session.store-type=redis ...
If you prefer not to use Spring Session to manage your Spring Boot application’s (HTTP) Session state at all, then do the following:
Use Web Server Session State Management.
#application.properties spring.session.store-type=none ...
Again, see Spring Boot docs for more details.
![]() | Tip |
---|---|
It is possible to include multiple providers on the classpath of your Spring Boot application. For instance, you might be using Redis to cache your application’s (HTTP) Session state while using either Apache Geode or Pivotal GemFire as your application’s persistent store (System of Record). |
![]() | Note |
---|---|
Spring Boot does not properly recognize |
Whether you are using Spring Session in a Spring Boot ClientCache
application connecting to an externally managed
cluster of Apache Geode or Pivotal GemFire servers, or connecting to a cluster of servers in a Pivotal Cloud Cache
instance managed by a Pivotal Platform environment, the setup is the same.
Spring Session for Apache Geode, Pivotal GemFire, and Pivotal Cloud Cache (PCC) expects there to exist a cache Region
in the cluster that will store and manage the (HTTP) Session state when your Spring Boot application is a ClientCache
application in a client/server topology.
By default, the cache Region used to store and manage (HTTP) Session state is called "ClusteredSpringSessions".
You can set the name of the cache Region used to store and manage (HTTP) Session state either by explicitly declaring
the @EnableGemFireHttpSession
annotation on your main @SpringBootApplication
class, like so:
Using `@EnableGemfireHttpSession.
@SpringBootApplication @EnableGemFireHttpSession(regionName = "MySessions") class MySpringBootSpringSessionApplication { ... }
Or alternatively, we recommend users to configure the cache Region name using the well-known and documented property
in Spring Boot application.properties
:
Using properties.
spring.session.data.gemfire.session.region.name=MySessions
Once you decide on the cache Region name used to store and manage (HTTP) Sessions, you must create the Region in the cluster somehow.
On the client, this is simple since SBDG’s auto-configuration will automatically create the client PROXY
Region
used to send/receive (HTTP) Session state between the client and server for you, when either Spring Session is on
the application classpath (e.g. spring-geode-starter-session
), or you explicitly declare
the @EnableGemFireHttpSession
annotation on your main @SpringBootApplication
class.
However, on the server-side, you currently have a couple of options.
First, you can create the cache Region manually using Gfsh, like so:
Create the Sessions Region using Gfsh.
gfsh> create region --name=MySessions --type=PARTITION --entry-idle-time-expiration=1800 --entry-idle-time-expiration-action=INVALIDATE
You must create the cache Region with the appropriate name and an expiration policy.
In this case, we created an Idle Expiration Policy with a timeout of 1800 seconds
(30 minutes
), after which,
the entry (i.e. Session object) will be "invalidated".
![]() | Note |
---|---|
Session expiration is managed by the Expiration Policy set on the cache Region used to store Session state. The Servlet Container’s (HTTP) Session expiration configuration is not used since Spring Session is replacing the Servlet Container’s Session management capabilities with its own and Spring Session delegates this behavior to the individual providers, like GemFire and Geode. |
Alternatively, you could send the definition for the cache Region from your Spring Boot ClientCache
application
to the cluster using the SBDG @EnableClusterAware
annotation,
which is meta-annotated with SDG’s @EnableClusterConfiguration
annotation.
![]() | Tip |
---|---|
See the Javadoc
on the |
Using @EnableClusterAware
.
@SpringBootApplication @EnableClusterAware class MySpringBootSpringSessionApplication { ... }
However, it is not currently possible to send Expiration Policy configuration metadata to the cluster yet. Therefore, you must manually alter the cache Region to set the Expiration Policy, like so:
Using Gfsh to Alter Region.
gfsh> alter region --name=MySessions --entry-idle-time-expiration=1800 --entry-idle-time-expiration-action=INVALIDATE
That is it!
Now your Spring Boot ClientCache
application using Spring Session in a client/server topology is configured to store
and manage user (HTTP) Session state in the cluster. This works for either standalone, externally managed Apache Geode
or Pivotal GemFire clusters, or when using PCC running in a Pivotal Platform environment.