[ClusterLabs] Antw: Re: Antw: Re: Pacemaker not always selecting the right stonith device

Ulrich Windl Ulrich.Windl at rz.uni-regensburg.de
Fri Jul 22 06:52:26 UTC 2016


>>> Andrei Borzenkov <arvidjaar at gmail.com> schrieb am 21.07.2016 um 18:39 in
Nachricht <a3f88e68-5ddb-f292-6d2a-96fe69e02416 at gmail.com>:
> 21.07.2016 09:49, Ulrich Windl пишет:
>>>>> Ken Gaillot <kgaillot at redhat.com> schrieb am 19.07.2016 um 16:17 in
Nachricht
>> <d82b1b0b-cfde-d478-4287-36b4a81631ac at redhat.com>:
>> 
>> [...]
>>> You're right -- if not told otherwise, Pacemaker will query the device
>>> for the target list. In this case, the output of "stonith_admin -l"
>> 
>> In sles11 SP4 I see the following (surprising) output:
>> "stonith_admin -l" shows the usage message
> 
> That's correct.
> 
>> "stonith_admin -l any" shows the configured devices, independently
>> whether the given name is part of the cluster or no. Even if that
>> host does not exist at all the same list is displayed:
>>  prm_stonith_sbd:0
>>  prm_stonith_sbd
>> 
>> Is that the way it's meant to be?
>> 
> 
> Well, SBD can in principle fence any node, so yes, I'd say it is. In my

I'd like to object: Don't you have to reserve a SBD slot for every host, and
despite of that if a host doesn't mount the shared storage SBD cannot fence it
;-) Saying "SBD can fence any node" is equivalent to say any fence agent can
fence any node. So why mess with nodes then?

> case (stonith:external/ipmi) it returns correct information. I am a bit
> surprised that it also does it for non-existing node, but as far as I
> understand if agent returns nothing host is not even checked and it is
> assumed agent can fence anything.

That could be. Should there be a node list to configure, or can't the agent
find out itself (for SBD)?

Regards,
Ulrich

> 
> 
> 
> _______________________________________________
> Users mailing list: Users at clusterlabs.org 
> http://clusterlabs.org/mailman/listinfo/users 
> 
> Project Home: http://www.clusterlabs.org 
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf 
> Bugs: http://bugs.clusterlabs.org 







More information about the Users mailing list