HTTP Session Management

Before diving into the specifics of how to plug-in and configure various alternative HTTP session management modules, let’s review some useful terminology:

More details on these concepts can be found in the Programming Guide.

There are a number of modules that offer pluggable alternatives for http session management. You can design how you want to cache and store http sessions by selecting alternative combinations of session modules.

For example, Jetty ships with two alternative implementations of the SessionCache:

There are at least 6 alternative implementations of the SessionDataStore that you can use to persist/distribute your http sessions:

The $JETTY_BASE/start.d/sessions.ini file contains these configuration properties:

So, if you’re not running on Google AppEngine, and you haven’t configured one, the workerName will always be: node0.

The HouseKeeper is responsible for the periodic initiation of session scavenge cycles. The jetty.sessionScavengeInterval.seconds property in $JETTY_BASE/start.d/sessions.ini controls the periodicity of the cycle.

A session whose expiry time has been exceeded is considered eligible for scavenging. The session might be present in a SessionCache and/or present in the session persistence/clustering mechanism.

Scavenging occurs for all contexts on a server at every cycle. The HouseKeeper sequentially asks the SessionHandler in each context to find and remove expired sessions. The SessionHandler works with the SessionDataStore to evaluate candidates for expiry held in the SessionCache, and also to sweep the persistence mechanism to find expired sessions.

The sweep takes two forms: once per cycle the SessionDataStore searches for sessions for its own context that have expired; infrequently, the SessionDataStore will widen the search to expired sessions in all contexts. The former finds sessions that are no longer in this context’s SessionCache, and using some heuristics, are unlikely to be in the SessionCache of the same context on another node either. These sessions will be loaded and fully expired, meaning that HttpSessionListener.destroy() will be called for them. The latter finds sessions that have not been disposed of by scavenge cycles on any other context/node. As these will be sessions that expired a long time ago, and may not be appropriate to load by the context doing the scavenging, these are summarily deleted without HttpSessionListener.destroy() being called.

A combination of these sweeps should ensure that the persistence mechanism does not fill over time with expired sessions.

As aforementioned, the sweep period needs to be short enough to find expired sessions in a timely fashion, but not so often that it overloads the persistence mechanism.

If you wish to change any of the default configuration values you should enable the session-cache-hash module. The name "hash" harks back to historical Jetty session implementations, whereby sessions were kept in memory using a HashMap.

You may need to use the session-cache-null module if your clustering setup does not have a sticky load balancer, or if you want absolutely minimal support for sessions. If you enable this module, but you don’t enable a module that provides session persistence (ie one of the session-store-xxx modules), then sessions will neither be retained in memory nor persisted.

The $JETTY_BASE/start.d/sessions.ini file contains the following properties which may be modified to customise filesystem session storage:

After enabling the module, the $JETTY_BASE/start.d/session-store-jdbc.ini file contains the following customizable properties:

The $JETTY_BASE/start.d/session-store-mongo.ini file contains these configurable properties:

The session-store-infinispan-remote module configures Jetty to talk to an external Infinispan instance to store session data.

Because Infinispan is not a technology provided by the Eclipse Foundation, you will be prompted to assent to the licenses of the external vendor (Apache in this case).

Infinispan-specific jar files are download to the directory named $JETTY_BASE/lib/infinispan/.

In addition to adding these modules to the classpath of the server it also added several ini configuration files to the $JETTY_BASE/start.d directory.

The infinispan-remote-query module allows Jetty to scavenge expired sessions. Note that this is an additional module, to be used in conjunction with the session-store-infinispan-remote module.

There are no configuration properties associated with this module.

Enabling the session-store-infinispan-embedded module runs an in-process instance of Infinispan.

Because Infinispan is not a technology provided by the Eclipse Foundation, you will be prompted to assent to the licenses of the external vendor (Apache in this case). Infinispan-specific jar files will be downloaded and saved to a directory named $JETTY_BASE/lib/infinispan/.

The infinispan-embedded-query module allows Jetty to scavenge expired sessions.

There are no configuration properties associated with this module.

From Jetty-9.4.13 onwards, we have changed the format of the serialized session when using a remote cache (ie using hotrod). Prior to release 9.4.13 we used the default Infinispan serialization, however this was not able to store sufficient information to allow Jetty to properly deserialize session attributes in all circumstances. See issue https://github.com/jetty/jetty.project/issues/2919 for more background.

We have provided a conversion program which will convert any sessions stored in Infinispan to the new format.

How to use the converter:

To perform the conversion, run the InfinispanSessionLegacyConverter with just the cache-name, and optionally the host system property. The following command will attempt to convert all sessions in the cached named my-remote-cache on the machine myhost, ensuring that application classes in the /my/custom/classes directory are on the classpath:

If the converter fails to convert a session, an error message and stacktrace will be printed and the conversion will abort. The failed session should be untouched, however it is prudent to take a backup of your cache before attempting the conversion.

Enabling the session-store-hazelcast-remote module allows jetty to communicate with a remote Hazelcast instance to cluster session data.

Because Hazelcast is not a technology provided by the Eclipse Foundation, you will be prompted to assent to the licenses of the external vendor (Apache in this case).

Hazelcast-specific jar files will be downloaded and saved to a directory named $JETTY_BASE/lib/hazelcast/.

This will run an in-process instance of Hazelcast. This can be useful for example during testing. To enable this you enable the session-store-hazelcast-embedded module.

Because Hazelcast is not a technology provided by the Eclipse Foundation, you will be prompted to assent to the licenses of the external vendor (Apache in this case).

Hazelcast-specific jar files will be downloaded to a directory named $JETTY_BASE/lib/hazelcast/.

You will first need to create a project and enable the Google Cloud API: https://cloud.google.com/docs/authentication#preparation. Take note of the project id that you create in this step as you need to supply it in later steps.

Using some special, composite indexes can speed up session search operations, although it may make write operations slower. By default, indexes will not be used. In order to use them, you will need to manually upload a file that defines the indexes. This file is named index.yaml and you can find it in your distribution in $JETTY_BASE/etc/sessions/gcloud/index.yaml.

Follow the instructions here to upload the pre-generated index.yaml file.

To enable communication using the GCloud Emulator:

The session-store-gcloud module provides GCloud support for storing session data.

Because the Google Cloud DataStore is not a technology provided by the Eclipse Foundation, when enabling the module you will be prompted to assent to the licenses of the external vendor.

As GCloud requires certain Java Commons Logging features to work correctly, Jetty routes these through SLF4J. By default, Jetty implements the SLF4J api, but you can choose a different logging implementation by following the instructions here

The MemcachedSessionDataMap uses memcached to perform caching of SessionData.

To enable it with the Jetty distribution, enable the session-store-cache module, along with your chosen session-store-xxxx module.

The standard support for webapps in Jetty will use sessions cached in memory, but not persisted/clustered, with a scavenge for expired sessions that occurs every 10 minutes. If you wish to pare back support for sessions because you know your app doesn’t use them (or use JSPs that use them), then you can do the following:

If you wish to do any further minimization, you should consult the Programming Guide.

Preferably, your cluster will utilize a sticky load balancer. This will route requests for the same session to the same Jetty instance. In this case, the DefaultSessionCache can be used to keep in-use session objects in memory. You can fine-tune the cache by controlling how long session objects remain in memory with the eviction policy settings.

If you have a large number of sessions or very large session objects, then you may want to manage your memory allocation by controlling the amount of time session objects spend in the cache. The EVICT_ON_SESSION_EXIT eviction policy will remove a session object from the cache as soon as the last simultaneous request referencing it exits. Alternatively, the EVICT_ON_INACTIVITY policy will remove a session object from the cache after a configurable amount of time has passed without a request referencing it.

If your sessions are very long lived and infrequently referenced, you might use the EVICT_ON_INACTIVITY_POLICY to control the size of the cache.

If your sessions are small, or relatively few or stable in number or they are read-mostly, then you might select the NEVER_EVICT policy. With this policy, session objects will remain in the cache until they either expire or are explicitly invalidated.

If you have a high likelihood of simultaneous requests for the same session object, then the EVICT_ON_SESSION_EXIT policy will ensure the session object stays in the cache as long as it is needed.

Without a sticky load balancer requests for the same session may arrive on any node in the cluster. This means it is likely that the copy of the session object in any SessionCache is likely to be out-of-date, as the session was probably last accessed on a different node. In this case, your choices are to use either the NullSessionCache or to de-tune the DefaultSessionCache. If you use the NullSessionCache all session object caching is avoided. This means that every time a request references a session it must be read in from persistent storage. It also means that there can be no sharing of session objects for multiple requests for the same session: each will have their own independent session object. Furthermore, the outcome of session writes are indeterminate because the Servlet Specification does not mandate ACID transactions for sessions.

If you use the DefaultSessionCache, there is a risk that the caches on some nodes will contain out-of-date session information as simultaneous requests for the same session are scattered over the cluster. To mitigate this somewhat you can use the EVICT_ON_SESSION_EXIT eviction policy: this will ensure that the session is removed from the cache as soon as the last simultaneous request for it exits. Again, due to the lack of session transactionality, the ordering outcome of write operations cannot be guaranteed. As the session is cached while at least one request is accessing it, it is possible for multiple simultaneous requests to share the same session object.

For various reasons it might not be possible for the SessionDataStore to re-read a stored session. One scenario is that the session stores a serialized object in its attributes, and after a re-deployment there in an incompatible class change. Setting the $JETTY_BASE/start.d/session-cache-hash.ini or $JETTY_BASE/start.d/session-cache-null.ini property jetty.session.removeUnloadableSessions to true will allow the unreadable session to be removed from persistent storage. This can be useful for preventing the scavenger from continually generating errors on the same expired, but un-readable session.