Trying my luck with JBossCache 1.4.1.SP1 on a Websphere 126.96.36.199 cluster with 2 nodes each node with 1 server. The OS on both machines is
Windows XP Professional SP2.
I'm using the replSync-service.xml configuration file supplied in download ( etc\META-INF folder ). I've changed only the
cache mode to INVALIDATION_SYNC because I need the sync happen sync and REPL_SYNC was too slow.
I'm running some tests which do both cache read / write I'm getting lots of org.jboss.cache.lock.TimeoutException
when acquiring locks.
The scenario in which the TimeOut occurs is the following :
All the operations occur on a TreeCache with one node with one entry only in the node:
in Node1 modify a node
- the invalidation msg is snd
in Node2 the WL is acquired by a user thread
- the invalidation msg is rcv
- the JGroups thread tries to put a WL and fails with a time out which is send to Node1
..... and this keeps going on.
What are the recomandations in this situations ( indeed this is an extrem situations with some many updates on the same entry ):
1. increase the timeout period ( the current one is zero )
2. use optimistic locking ( how can I configure it ?)
3. other ?