[Pacemaker] how to determin crm version / features from within a RA

Andrew Beekhof andrew at beekhof.net
Thu Jun 25 18:34:27 UTC 2009


On Thu, Jun 25, 2009 at 5:13 PM, Lars
Ellenberg<lars.ellenberg at linbit.com> wrote:
> On Thu, Jun 25, 2009 at 04:47:55PM +0200, Andrew Beekhof wrote:
>> Can we back up a moment though and restate the problem?
>
> It was also sort of a philosophical question. But anyways ;)

As a philosophical question, I'd have to admit that the current
situation sucks, promise to expose the true version in an accessible
way.
I think I've done that for 1.0.4 (now that the bugs you mentioned are
hopefully fixed).

>> Is there a specific version barrier you're looking for?
>
> DRBD is one of the few Multi State (or Master Slave or whatever) RA
> people actually try to use.
>
> It is my understanding that, because these Master RAs have been less
> used/tested, older CRM do not handle them as well as "ordinary" RAs.

well there were certain aspects of M/S handling that were sub-optimal.
which is the long way of saying "you're correct"

> I'd like to have a check in there that says: forget it,
> this CRM is known to not work as intended with Master RAs.
>
> I'm open for suggestions which version to trigger on,
> my intention was to try and say:
> heartbeat built-in crm 2.1.4, or pacemaker.
>
> If you say trigger on pacemaker 1.0 is fine, I'd just check on that.

IIRC, Pacemaker 0.6.4 or 5 was the first version (thanks to kliend's
testing) that handled m/s "well".
But all things considered, I'd trigger on feature set >= 3.0.0 (aka.
pacemaker 1.0.0) _or_ the crmadmin --version output.

>
> If you say: BS, Master RAs are handled just as well or bad in 2.0.8-SLES10,
> ok, I'll just skip this check.
>
> --
> : Lars Ellenberg
> : LINBIT | Your Way to High Availability
> : DRBD/HA support and consulting http://www.linbit.com
>
> DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>




More information about the Pacemaker mailing list