[Pacemaker] stonith suice / chaining stonith agents

[Dejan Muhamedagic]

Hi,

On Thu, Jul 23, 2009 at 05:19:56PM +0200, Bernd Schubert wrote:
> Hello,
> 
> for suicides I would prefer to have a stonith agent that resets the system 
> with sysrq-s, sysrq-u and sysrq-b, since that would at least allow to sync 
> filesystems. 

A suicide stonith plugin is by definition not reliable. There is
no way to figure out if it succeeded or not. Should be used only
in 1-node clusters.

> An option would be to check in all agents if the local node node is to be 
> reset, if try to use sysrq, if not or if sysrq fails use stonith. 
> But it is annyoing to modify each agent for that purpose. So it would be 
> helpful to have a chain of stonith agents. E.g. first try sysrq, if that fails 
> try ipmi, if that fails try PDU.
> 
> I also somehow remember to have read/head that it is possible to do that, but 
> I can't find web reference to that (maybe I'm searching for the wrong 
> keywords?).

You can have more than one stonith resource and they will be
tried in a round robin fashion until one succeeds. You can also
set priorities for stonith resources to defined the order: just
set the "priority" instance attribute. Smaller numbers are higher
priority.

Thanks,

Dejan

> Any ideas?
> 
> 
> Thanks,
> Bernd
> 
> 
> -- 
> Bernd Schubert
> DataDirect Networks
> 
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker