5.2. Resolved Issues

After a connection failure to a JDG server, normal operations like get and put recover and work as expected. But a ClientListener stops working. After registering the listener using cache.addClientListener(listener) again, it recovers and works as expected.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

When inserting 20 key/values in a 10 sized cache (keys from key-1 to key-20) and then reinserting the first 10 keys (key-1 to key-10) while also using LIRS eviction strategy and listener, a NullPointerException is thrown in CacheNotifierImpl.notifyCacheEntriesEvicted. This issue existed in both Remote Client-Server mode and Library mode.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2

Previously, when using AtomicMaps and writeskew, lifespan as set in the configuration wasn't applied. This has been addressed in JBoss Data Grid 6.6.2 and now lifespan in the configuration is honored.

The expiration metadata is not applied to a newly created entry that was read in the same transaction as a null.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

@CacheEntryExpired listener method is not invoked for non auto-commit cache. When auto commit is true the listener method is invoked. @CacheEntryCreated is invoked in both configs.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

Methods PreferConsistencyStrategy and StateConsumerImpl contained non-typesafe printf errors.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

If the proto file has syntax errors it isn't placed in the ___protobuf_metadata cache as it should be. Additionally, a myFileWithSyntaxErrors.proto.errors key fails to be created and an exception is thrown.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

org.infinispan.rest.Server has a static field of DatePatternRfc1123LocaleUS, wihich is an instance of SimpleDateFormat. This causes a java.lang.ArrayIndexOutOfBoundsException under load.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

If infinispan.external_addr is defined in a server configuration file like clustered.xml, the JBoss Data Grid client will be unable to resolve the address.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

Clearing the cache map doesn't work under the following configuration: JBoss Data Grid is running in Library mode inside JBoss EAP 6.4.6 with passivation enabled. In this case the size will never be reduced to 0.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

When using LIRSEvictionPolicy in some situations a NullPointerException can occur in BoundedEquivalentConcurrentHashMapV8. The error is harmless but the issue has been fixed.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

The Hot Rod client uses more memory than it should. The buffering implementation of TcpTransport.socketOutputStream does not use a fixed sized input buffer. Instead it grows as BufferedInputStream does. A growing buffer can lead to excessive memory consumption in the heap, as well as in the native memory of the operating system.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

Previously, performing a rolling upgrade could fail with a java.lang.ClassCastException of either org.infinispan.container.entries.RepeatableReadEntry cannot be cast to org.infinispan.container.entries.InternalCacheEntry or SimpleClusteredVersion cannot be cast to NumericVersion.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

When doing a rolling upgrade that takes more than a few minutes to complete a TimeOutException could be thrown.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.

Previously, during Hot Rod rolling upgrades, write operations executed by the client on the target cluster were ignored. This could cause unexpected results in the application, such as entries being deleted that weren't intentionally deleted.

This issue is resolved as of Red Hat JBoss Data Grid 6.6.2.