At a guess, I think this might be related to the auto-nodeid code. If you set a fixed value in corosync.conf, this possibly wouldn't happen.
On Fri, Apr 16, 2010 at 10:21 AM, Alain.Moulle <alain.mou...@bull.net> wrote: > Hi, > > Sorry but this was not due to new releases, but only to the fact that > one subnet for > the redundant heartbeat was no more available on the node, whereas I > always had it in > corosync.conf, I was not aware of this "de-configuration" of IF, and it > takes me a > while to identify the problem ... So , to be clear for all people who > will eventually > encounter the same problem : > if you have two interfaces in corosync.conf: > ringnumber 0 (bindnetaddr:11.1.0.0) > and > ringnumber 1 (bindnetaddr:11.2.0.0) > > but if only one is listed in the IF list (with /ip addr/ command for > example), > > that's create strange problem meaning that you can start successfully > corosync/Pacemaker on one node, and when you start on the second > one, it stops corosync/Pacemaker on the first one, and nothing can > be starded correctly. > > But messages in syslog or messages files do not enable to quickly identify > that it is a problem of missing network interface with regard to the > interfaces > configured in corosync.conf (at least for me who does not know the > internals > of corosync etc. !). > > Hope this description will help some of you if one day you face the same > problem. > > Regards > Alain Moullé > > _______________________________________________ > Linux-HA mailing list > Linux-HA@lists.linux-ha.org > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems > _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems