Hi, On Wed, Jun 17, 2009 at 05:39:18PM -0700, c smith wrote: > Hi Dejan- > > I apologize for creating the hb_report with experimental timeouts and fail > counts not reset. I found the issue was with the clustered file system. > When node2 disappeared, OCFS2 I/O would hang while the file system recovered > from the lost node. When the start timeouts were set higher, resources > would start as soon as I/O resumed which explains the delay in failover
Ah, I was wondering how were you doing live migration without shared storage. BTW, you should include ocfs2 mounts in the cluster configuration. > You don't have stonith configured, which makes a two-node > > configuration impossible. > > > I'm interested to know what you mean by this. I've configured several 2 > node heartbeat clusters without stonith since data divergance wasn't a huge > worry. This is my first time working with pacemaker/openais. What > difference does stonith make if the second node is not available to be shot? > ie, power failure. How do you know that it's a power failure and not a split brain? Thanks, Dejan > > Thanks again > _______________________________________________ > 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
