[ClusterLabs] crm_mon watch fencing option - no notification

[Ken Gaillot]

On 03/03/2015 10:14 AM, Gao,Yan wrote:
> Hi Diane,
> 
> On 03/03/2015 01:44 PM, Schaefer, Diane E wrote:
>> Hello,
>>
>>    We are running on SLES 11 SP3 pacemaker 1.1.11-0.7.53 and are
>> attempting to get notifications of fencing events for an audit trail we
>> are constructing.  We are also hoping to get events for resource
>> start/stops and failure. We have constructed a script that writes to a
>> special log when triggered and started crm_mon with the following
>>
>>             crm_mon -d -W --external-agent=/root/bin/stamp.sh
>>
>>  
>>
>> We clearly see the fencing message get written to the syslog, but we
>> were hoping for a notification.
>>
>>  
>>
>> Mar  2 21:24:13 usrv-fsm21 crmd[15624]:   notice: te_fence_node:
>> Executing reboot fencing operation (88) on usrv-fsm11 (timeout=150000)
> 
> When a fencing action returns, the external agent will be invoked with
> some environment variables set like:
> 
> 
> CRM_notify_desc='Operation st_notify_fence requested by ... for peer
> ...: OK ...'
> CRM_notify_node=...
> CRM_notify_rc=0
> CRM_notify_status=0
> CRM_notify_target_rc=0
> CRM_notify_task=st_notify_fence
> 
> Regards,
>   Yan

And, there is an ocf:pacemaker:ClusterMon resource agent that will let
pacemaker manage your external agent as a cluster resource.

Documentation for that and the environment variables is at
http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html/Pacemaker_Explained/s-notification-external.html

>>
>> How is the –W supposed to function?  Any advice on capturing fencing events?