[jboss-jira] [JBoss JIRA] Updated: (JBMESSAGING-1861) Enhacement - Producer should be reconnected to another node in cluster when node to which he's sending lost connection to database

[ https://issues.jboss.org/browse/JBMESSAGING-1861?page=com.atlassian.jira.... ] Miroslav Novak updated JBMESSAGING-1861: ---------------------------------------- Description: There should be made an enhancement in JBM clustering. Producer should be reconnected to another node in cluster when node to which he's sending messages lost connection to database. Producer is using clustered connection factory and sends messages to clustered queue. From client point of view this is the same situation as when node crashed. Fail-over of all messages to another node should also occur and node should be stopped. Thanks, Mirek was: There should be made an enhancement in JBM clustering. For now the behaviour is following: Clustered queue is used and Session_Transacted mode is used. 1. cluster of 3 nodes - A - B - C 2. send 500 to each (db status - A:500, B:500, C:500) 3. disconnect B from DB (db unchanged) 4. read C - 1000 messages read (db status A:0, B:500, C:0) - hangs 5. read A - 0 messages read (db status A:0, B:500, C:0) 6. connect B to DB (db status A:0, B:417, C:83) 7. read B - 500 messages read (db status A:0, B:0, C:0) In step 4. consumer received 1000 from 1500 messages from nodes A and C and there are still 500 messages left in the db. As an enhancement there should be that client on node C in step 4. will read all the messages from clustered queue. Thanks, Mirek -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira