[Pacemaker] Issues with fence and corosync crash

Dejan Muhamedagic dejanmm at fastmail.fm
Tue Jan 4 17:21:23 UTC 2011


Hi again,

On Fri, Dec 24, 2010 at 12:05:27PM +0100, Simone Felici wrote:
>
> Hi to all!
>
> I've an issue with my cluster env. First of all my config:
>
> Two Cluster CentOS5.5 Active+Standby with one DRBD partition managing a Nagios service, ip, and storage.
> The config files at the bottom.
>
> I'm trying to test fence option to prevent split brain and problems on double access on drbd partition.
> Starting on a sane situation, manual switching of the resources or 
> simulating kernel-panic, crash of process or whatever, all works well. If 
> I try to shutdown the eth1 (192.168.100.0 as well as cross cable to drbd 
> mirroring) the active stay as it is, it calls the fence option adding the 
> entry to crm config:

Oh, forgot to say that doing 'ifconfig down' is an invalid
condition in corosync/pacemaker clusters.

Thanks,

Dejan




More information about the Pacemaker mailing list