I have the following scenario: six embedded instances of Infinispan in two different computers (3 in one computer and 3 in the other) on a LAN, running on replicated mode. The following is the configuration is each instances:
<clustering mode="replication"> <sync /> <stateRetrieval fetchInMemoryState="false" /> </clustering> <locking useLockStriping="false" /> <deadlockDetection enabled="true" spinDuration="500" /> <transaction transactionManagerLookupClass="..." syncCommitPhase="true" syncRollbackPhase="true" useEagerLocking="false" useSynchronization="false" eagerLockSingleNode="false" > <recovery enabled="true" /> </transaction>
After running a number of put and get operations and stopping them, leaving the grid in a "repose" state, the value of the indicator "NumberOfLocksHeld" is above 0 (zero), when it should be 0 (zero), if i understood what the indicator means correctly.
Could you suggest me which could the causes for this happening?
What Infinispan version?
Can you attach the test you use?
When running the test, are any exceptions thrown?