[ https://issues.jboss.org/browse/ISPN-2751?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-2751: ---------------------------------------- In the past we had issues where the RPC dispatcher was stopping *after* the cache manager had stopped. This lead to NPEs in ExternalizerTable (ISPN-2025) as a result of trying to deserialize stuff after the cache managed had stopped. At the time we added a workaround to handle null commands, which effectively are the result of externalizer table not being able to deserialize objects (normally due to the sequence of events mentioned) and we reverted it (ISPN-2063) once AS7-3180 had been implemented. The problem now does not appear to be that we cannot deserialize something, but rather than while sending a response to *something*, the cache manager was stopped leading to a null externalizer table. The circumstances look very similar to ISPN-2752 actually and the result seems to be circumstantial: * In ISPN-2752 we see that the cache manager has started stopping but the externalizer table is still there, because the response is being marshalled. * In this JIRA, we see that the cache manager has fully stopped, the externalizer table is not there any more, but the JGroups transport (or RPC dispatcher) is still actively trying to send a reply to a command. Some TRACE logs, including not only org.jgroups and org.infinispan, but also org.jboss.as.clustering would be very handy to try to better understand how to fix this. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira