On Tue, Dec 2, 2008 at 10:59 PM, Adrian Overbury <[EMAIL PROTECTED]> wrote:
> I have a pgsql resource set up on my 2-node heartbeat 2.1.3 cluster in CRM.
>  I'd like to be able to restart Postgres when I made configuration changes,
> but I've found that if the database is busy (thus making the Postges restart
> take some time) the nodes can fail over because Heartbeat thinks that
> Postgres has died.

Do you need to make changes in Postgres configuration and restart it
without Heartbeat noticing it? In this case you need to put your
resource into "unmanaged" state while you are doing this and then put
back to "managed" state when you are done.

Also I'd suggest to upgrade youre Heartbeat. 2.1.3 has some well known
bugs. Alsi pgsql RA in that version has some issues too.

>
> I could lengthen the monitor and deadtime intervals, but I like how
> responsive it is to a real problem right now, so is there any easy, or at
> least possible, way that I can tell heartbeat to take a resource down and
> start it up again on the same node?  Nothing depends upon Postgres in the
> ordering constraints, so I won't have to worry about it taking other
> resources with it, but I would very much like to be able to restart without
> failing over.
>
> --
>
> Adrian Overbury
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>



-- 
Serge Dubrouski.
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to