[ http://jira.jboss.com/jira/browse/JBMESSAGING-1173?page=comments#action_1... ] Tim Fox commented on JBMESSAGING-1173: -------------------------------------- Just to like to make a note that this won't prevent "split-brain" occurring in the first place. "Split -brain" is a condition that can happen to pretty much stateful clustered service with HA, it's not specific to JBM. Once in a split brain mode, both nodes will think they are in control of the queues, and therefore it's possible the same messages will be delivered to different consumers from the two servers. I believe this is what the customer is seeing in practice. There's not real a lot can be done once split brain is occurred, since, by definition the "islands" can't see each other. Standard approaches here are to add redundancy on the IP layer - e.g. IP bonding where you can a single virtual interface over two real network interfaces - if one is pulled the other takes over. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBMESSAGING-1173?page=all ] Ovidiu Feodorov updated JBMESSAGING-1173: ----------------------------------------- Description: JBM does not currently support the merging of cluster segments that have gotten into the "split-brain" situation. This occurs when network connectivity is lost between subsets of a cluster. Each subset assumes the other is gone and acts appropriately. When the network connectivity is restored, they do not merge back into a single cluster but remain as independent clusters (the spit-brain). This feature request is to add cluster merging to prevent the split-brain situation. was: JBM does not currently support the merging of cluster segments that have gotten into the "split-brain" situation. This occurs when network connectivity is lost between subsets of a cluster. Each subset assumes the other is gone and acts appropriately. When the network connectivity is restore, they do not merge back into a single cluster but remain as independent clusters (the spit-brain). This feature request is to add cluster merging to prevent the split-brain situation. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBMESSAGING-1173?page=com.atlassian.ji... ] Tim Fox closed JBMESSAGING-1173. -------------------------------- Resolution: Out of Date 1194 is handling split brain. With JBM 2.0 we shouldn't need to merge since we should prevent split brain from happening in the first place (using a quorum) -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira