Chapter 3. Resolved Issues

Previously in Red Hat JBoss Data Grid, TransactionTable.start() initialized the TxCleanupService thread pool when dealing with a non-transactional cache.

This is fixed in JBoss Data Grid 6.3.2 so that the TxCleanupService is disabled for non-transactional caches as expected.

Previously in Red Hat JBoss Data Grid, the OOB and/or thread pool became deadlocked during large state transfer operations (for example, 300 nodes with 3000 caches).

This is now fixed in JBoss Data Grid 6.3.2 and the state transfer thread no longer becomes deadlocked during large state transfer operations.

Previously in Red Hat JBoss Data Grid, the rebalancing "Finished cluster wide rebalance" message was not logged in INFO as expected. Users could access the log (started/enabled/suspended) messages, but could not view the expected "finished cluster wide rebalance" message in INFO.

This is now fixed in JBoss Data Grid 6.3.2 and the message is logged in INFO as expected.

Previously in Red Hat JBoss Data Grid, the Hot Rod Bulk operations used MapReduce which implied the EXEC permission even though conceptually they should be BULK_READ operations.

This is now fixed in JBoss Data Grid 6.3.2.

Previously in Red Hat JBoss Data Grid, Hot Rod authentication succeeded with a clustered cache, but executing the RemoteCache.size() command threw a an unexpected SecurityException.

This is now fixed in JBoss Data Grid 6.3.2 and Hot Rod authentication works as expected.

Previously in Red Hat JBoss Data Grid, when the server cache was clustered, the Hot Rod client was unable to authenticate itself despite using the correct credentials and displayed the following error:

09:44:00,343 INFO  [org.infinispan.AUDIT] (HotRodServerWorker-1) [DENY] null ADMIN cache[securedcache]

09:44:00,343 INFO  [org.infinispan.AUDIT] (HotRodServerWorker-1) [DENY] null ADMIN cache[securedcache]

The transport layer in JBoss Data Grid was breaking the authentication process. However, authentication worked as expected when a local cache was used instead of a clustered cache.

This is now fixed in JBoss Data Grid 6.3.2 and Hot Rod authentication for a clustered server works as expected.

Previously in Red Hat JBoss Data Grid, when using the FD_HOST protocol in JGroups for node failure detection (whether the node was alive was checked using ICMP pings), a node was suspected to be dead even if it was responsive. This issue was more likely to occur in larger clusters.

This issue is now fixed in JBoss Data Grid 6.3.2.