On 11/22/05, Joachim Banzhaf <[EMAIL PROTECTED]> wrote:
> Am Dienstag, 22. November 2005 03:29 schrieb Andrew Beekhof:
> > On 11/11/05, [EMAIL PROTECTED] <[EMAIL PROTECTED]>
> wrote:
> > > Hi,
> > >
> > > I have a problem with migrating resources. The config is the same as
> > > with my earlier postings:
> > >
> > > I started my 3 nodes with 3 groups, group 1 defaults to node 1 and so on.
> > > Due to a previous (known) problem during shutdown, the db2 resources did
> > > not start on the first attempt. So I ended up with resources of group 3
> > > running on node 2 - fine. At that point I started crm_resource to
> > > migrate group 3 to node 3. This failed. The resources where stopped on
> > > node 2 but not started on group 3.
> > >
> > > Can it be, that the cluster remembers earlier failures of group 3 on
> > > node 3 and thus refuses to start them there again?
> >
> > yes - this is exactly what it does
>
> OK, fine. Can I somehow avoid the stop on node 2 then, without manual
> simulation of a pe run?
> I think of something like a crm_resource parameter
> --migrate-but-only-if-start-seems-possible.
>

there should be some way to be able to provide this feature.  could
you add it to bugzilla as an enhancement so its not lost?
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/

Reply via email to