On 10/05/2013, at 11:37 PM, Ferenc Wagner <[email protected]> wrote:

> Hi,
> 
> An hour ago one node (n02) of our 4-node cluster started to shutdown.

Someone, probably the init script, sent SIGTERM to pacemakerd.

> No idea why.  But during shutdown, it asked another node (n01) to shut
> down as well:

No it didn't.  It asked n01 to perform an orderly shutdown (resources first, 
then the pacemaker daemons) of n02.

> 
> May 10 13:59:42 n02 pacemakerd: [10851]: info: crm_signal_dispatch: Invoking 
> handler for signal 15: Terminated
> May 10 13:59:42 n02 pacemakerd: [10851]: notice: pcmk_shutdown_worker: 
> Shuting down Pacemaker
> May 10 13:59:42 n02 pacemakerd: [10851]: notice: stop_child: Stopping crmd: 
> Sent -15 to process 10860
> May 10 13:59:42 n02 crmd: [10860]: info: crm_signal_dispatch: Invoking 
> handler for signal 15: Terminated
> May 10 13:59:42 n02 crmd: [10860]: notice: crm_shutdown: Requesting shutdown, 
> upper limit is 1200000ms
> May 10 13:59:42 n02 crmd: [10860]: debug: crm_timer_start: Started Shutdown 
> Escalation (I_STOP:1200000ms), src=50
> May 10 13:59:42 n02 crmd: [10860]: debug: s_crmd_fsa: Processing I_SHUTDOWN: 
> [ state=S_NOT_DC cause=C_SHUTDOWN origin=crm_shutdown ]
> May 10 13:59:42 n02 crmd: [10860]: debug: do_fsa_action: actions:trace: 
> #011// A_SHUTDOWN_REQ
> May 10 13:59:42 n02 crmd: [10860]: info: do_shutdown_req: Sending shutdown 
> request to n01
> 
> Then hell broke loose, and I'm still pondering over the logs,

Look for resource stop actions that failed.

> but
> meanwhile, could somebody please provide an explanation for crmd
> "Sending shutdown request to n01"?  It switched off another node
> of the cluster, but why?
> 
> As a note, STONITH was disabled at the moment, because I forgot to
> reenable it after yesterday's maintenance session...
> 
> Corosync 1.4.2, Pacemaker 1.1.7, plugin version 1.
> -- 
> Thanks,
> Feri.
> _______________________________________________
> 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

Reply via email to