Hi
On Wed, 23 Sep 2009 13:29:52 +0200
Dejan Muhamedagic <[email protected]> wrote:
> > So my cluster is working good, resource A is running on n1,
> > resource B on n2. For some reason i decide to make
> > some maintenance on node n1 and tell the crm to set that node to
> > 'standby' state. During the next several seconds i see the following
> > actions:
> >
> > The B resource is stopped on n2 node
>
> I guess you mean here 'n1 node'.
Nope. B were running on n2 node all the time. A was on n1 and n1 is now
set to 'standby'
> > The A resource is migrated to n2 node
> > The B resource is started on n2 node
>
> Since these are virtual machines, perhaps you could use the real
> migration (though I think you need shared storage for that).
I have shaed storage and resource A is really migrated (vm didn't
rebooted, ssh sessions aren't broken, uptime is big :)
> > I was expecting that B resource wouldn't be touched at all. The A
> > resource was unavailable for less then a second and it didn't
> > rebooted and it's execution state doesn't changed during migration.
>
> Yes, that makes sense. Would setting symmetrical (that's an order
> collocation attribute) to false make a difference? If not, I
> don't really have an idea how to configure what you need. And in
> that case perhaps you could open an enhancement bugzilla and
> paste your configuration and this description.
No, setting 'symmetrical' to false doesn't helps.
Have to write a feature request :-) and my be even implement it
sometime :)
--
Denis Chapligin
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
