[ClusterLabs] Antw: [EXT] Moving multi-state resources

[Andrei Borzenkov]

On Wed, May 12, 2021 at 8:15 PM Alastair Basden <a.g.basden at durham.ac.uk> wrote:
>
>
> > On 12.05.2021 20:02, Alastair Basden wrote:
> >>>> Oddly enough, if I:
> >>>> pcs resource move resourcedrbdClone node2
> >>>> it moves it to node 2 okay.
> >>>>
> >>>> But then if I
> >>>> pcs resource clear resourcedrbdClone
> >>>> it moves it back to node1.
> >>>>
> >>>> Which is odd, given its score for role Master is higher on node 2.
> >>>>
> >>>
> >>> How do you know it?
> >> Because I have set score=100 for node2, and score=50 for node1.
> >>
> >
> > I told you several time that resource agent sets promotion scores. If
> > you want to see actual scores, use crm_simulate -Ls.
>
> Thanks, yes, but I was talking about MY score.

Pacemaker combines scores from multiple sources and it is the final
total that is used for deciding resource placement.

>  I have set a higher score
> for the node where I want it, and therefore, I would expect it to move
> back there (or at least, stay there), not move away from it!
>
> The fact that it isn't presumably means that something is wrong in the
> configuration that I posted.  The question is, what.  i.e. how does one
> specify a preferred node for the master of a master/slave clone.
>
> Once I can work this out, it will all work correctly!
>
> >>> Every transaction is saved in a file and you can replay it with
> >>> crm_simulate. Check logs for these file names.
> >>
> >> Which logs should be checked?
> >
> > pacemaker logs.
> Thanks.  So:
> /var/log/pacemaker.log, which then suggests /var/log/cluster/corosync.log.
>
> Thanks,
> Alastair.
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users
>
> ClusterLabs home: https://www.clusterlabs.org/