[Pacemaker] restart of crmd?

[Andrew Beekhof]

On Wed, Mar 17, 2010 at 10:36 PM, Alan Jones <falancluster at gmail.com> wrote:
> On Wed, Mar 17, 2010 at 1:39 PM, Andrew Beekhof <andrew at beekhof.net> wrote:
>>
>> On Wed, Mar 17, 2010 at 7:23 PM, Alan Jones <falancluster at gmail.com>
>> wrote:
>> > Is there any interest among people working with Pacemaker to provide for
>> > restarting crmd locally without failover and rediscovering resouce agent
>> > states through their monitor scripts?
>>
>> Out of interest, what would that achieve?
>
>
> ... greater resilience to bugs or problems in crmd.
> Failover is disruptive to the application.  If we can recover from a
> "transient" crmd process failure by locally restarting, there is less
> disruption.

But we do this already... try: killall -9 crmd

> Of course, like the processes it monitors, a repeated crmd failure should be
> converted to a "hard" failure that results in failover.