Hi, > > I run this combination. > > > > # rpm -qa | grep heartbeat > > heartbeat-2.1.3-13.1 > > heartbeat-common-2.1.3-13.1 > > heartbeat-common-devel-2.1.3-13.1 > > heartbeat-resources-2.1.3-13.1 > > pacemaker-heartbeat-0.6.0-20.2 > > > > sequence of operation is here; > > (1) start up three nodes > > (node-a, node-b, node-c. DC = node-c) > > > > (2) down the interconnect LAN of node-c > > [node-c]# iptables -A INPUT -i eth2 -j DROP > > [node-c]# iptables -A OUTPUT -o eth2 -j DROP > > > > (3) split brain comes up > > <node-a + node-b> vs. <node-c> > > > > (4) up the interconnect LAN of node-c > > [node-c]# iptables -F > > > > It seems that Heartbeat can handle the recovering from a split brain, > > but I found the strange message in ha-log. > > During the recovering, node-b losts node-a for some reason. > > > > crmd[7463]: 2008/02/15_17:40:54 info: ccm_event_detail: NEW MEMBERSHIP: > > trans=6, nodes=2, new=1, lost=1 n_idx=0, new_idx=2, old_idx=5 > > crmd[7463]: 2008/02/15_17:40:54 info: ccm_event_detail: CURRENT: > > node-c [nodeid=2, born=1] > > crmd[7463]: 2008/02/15_17:40:54 info: ccm_event_detail: CURRENT: > > node-b [nodeid=1, born=6] > > crmd[7463]: 2008/02/15_17:40:54 info: ccm_event_detail: NEW: > > node-c [nodeid=2, born=1] > > crmd[7463]: 2008/02/15_17:40:54 info: ccm_event_detail: LOST: > > node-a [nodeid=0, born=2] > > crmd[7463]: 2008/02/15_17:40:54 info: crm_update_peer: Node node-a is now: > > lost > > crmd[7463]: 2008/02/15_17:40:54 info: crm_update_peer: Node node-c is now: > > member > > > > node-a and node-b were peer before and there is no need to fall away the > > membership. > > Instance ID is normal. > > Is it expected behavior or the CCM bug again ? > > Definitely not expected. Since there was no message about link > going up/down for node-b, it looks like a bug to me. Is this > reproducable? If so, could you set debug to 1 and rerun. And file > a bugzilla.
This symptom is reproducible. I posted this as bugzilla #1834. http://developerbugs.linux-foundation.org/show_bug.cgi?id=1834 Thanks! Junko > > Thanks, > > Dejan > > > I attached the hb_report. > > > > Best Regards, > > Junko Ikeda > > > > NTT DATA INTELLILINK CORPORATION > > > > _______________________________________________ > > Linux-HA mailing list > > [email protected] > > http://lists.linux-ha.org/mailman/listinfo/linux-ha > > See also: http://linux-ha.org/ReportingProblems > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
