[ClusterLabs] Antw: Re: Antw: Trying to Understanding crm-fence-peer.sh

Bryan K. Walton bwalton+1546953805 at leepfrog.com
Thu Jan 17 16:27:46 UTC 2019


On Thu, Jan 17, 2019 at 07:54:28AM +0100, Ulrich Windl wrote:
> 
> Jan 11 08:49:54 storage2 crmd[13173]: warning: No reason to expect node 1 to be down
> Jan 11 08:49:54 storage2 crmd[13173]:  notice: Stonith/shutdown of storage1 not matched
> 
> So it seems you reboot did not shut down pacemaker properly, so the other node will assume some failure and fence (to be sure the other side is dead).

This makes sense, and would explain why, when I issued a reboot test
yesterday, I didn't see crm-fence-peer.sh getting called, at all. I will
attribute blame to an improper shutdown that one time.

Thanks!
Bryan


More information about the Users mailing list