On Tue, Oct 16, 2012 at 8:00 PM, M K <m.kowalski1104@xxxxxxxxx> wrote: > You mean that the fact that I can't start corosync again with message > "corosync dead but subsys locked" and need to remove this file manually is > the intended behavior? You only just mentioned that, in a comment directed specifically to Dan so I didn't read it. What prints that message and when? We're not mind readers. > > Expected behavior for me would be that I don't have to remove anything > manually and it is handled by start/stop scripts. > > > On Tue, Oct 16, 2012 at 10:40 AM, Andrew Beekhof <andrew@xxxxxxxxxxx> wrote: >> >> On Tue, Oct 16, 2012 at 6:02 PM, M K <m.kowalski1104@xxxxxxxxx> wrote: >> > @Dan: I am setting up database failover cluster and checking how >> > software >> > reacts to stopping, killing, rebooting, etc. Trying to stop corosync >> > first >> > does not stop serving resources, however if I stop pacemaker after >> > trying to >> > stop corosync I cannot start corosync anymore without manual >> > intervention - >> > "corosync dead but subsys locked" >> > >> > @Andrew: Can you please advise which version handles this situation >> > correctly? >> >> What I'm saying is that they already are handling the situations >> correctly. >> >> > Do I need to use 2.x.x branch? >> > >> > Best Regards, >> > Marcin >> > >> > >> > On Thu, Oct 11, 2012 at 9:53 AM, Andrew Beekhof <andrew@xxxxxxxxxxx> >> > wrote: >> >> >> >> On Thu, Oct 11, 2012 at 6:12 PM, M K <m.kowalski1104@xxxxxxxxx> wrote: >> >> > Hello List, >> >> > >> >> > I have corosync/pacemaker on RHEL (6.3) machines with 3 resources >> >> > configured: >> >> > >> >> > - VirtualIP >> >> > - FS (SAN) >> >> > - MySQL >> >> > >> >> > Scenario 1: >> >> > - make sure that corosync/pacemaker is up and running on two nodes >> >> > and >> >> > serving these three resources >> >> > - run /etc/init.d/corosync stop - never returns, but cluster >> >> > continues >> >> > serving resources on the same node >> >> >> >> You can't stop corosync without stopping pacemaker first. >> >> Corosync will refuse. >> >> >> >> > - run /etc/init.d/pacemaker stop - pacemaker is stopped and corosync >> >> > is >> >> > dead >> >> > >> >> > >> >> > >> >> > Scenario 2: >> >> > - make sure that corosync/pacemaker is up and running on two nodes >> >> > and >> >> > serving these three resources >> >> > - run /etc/init.d/pacemaker stop >> >> > - run /etc/init.d/corosync stop >> >> > >> >> > >> >> > Should corosync/pacemaker handle both scenarios correctly? >> >> >> >> It already is. >> >> >> >> > >> >> > Best Regards, >> >> > Marcin >> >> > >> >> > _______________________________________________ >> >> > discuss mailing list >> >> > discuss@xxxxxxxxxxxx >> >> > http://lists.corosync.org/mailman/listinfo/discuss >> >> > >> > >> > > > _______________________________________________ discuss mailing list discuss@xxxxxxxxxxxx http://lists.corosync.org/mailman/listinfo/discuss
