[Linux-HA] a fail over with some group resources

Andrew Beekhof beekhof at gmail.com
Tue Oct 16 05:17:44 MDT 2007


On 10/16/07, Junko IKEDA <ikedaj at intellilink.co.jp> wrote:
> > > node prec370e was DC first.
> > > after checking the above status,
> > > I stopped the heartbeat on prec370d.
> > > so prec370d took over the all services successfully.
> > >
> > > but... I found the following;
> > > Dummy01 had started on prec370d before prec370d noticed that it LOST its
> > > buddy.
> > > when the DC was elected again, Dummy01 has started already.
> > > is it expected?
> > > I thought that the resource would start after DC election in this case.
> >
> > It is expected.
> >
> > When you stop heartbeat cleanly, it migrates the resources cleanly to
> > the remaining nodes first, so it'll complete before the DC election.
> >
> > Anything else would be a bug.
>
> do you have any plans to change this behavior in the next release? (or
> sometime in the future?)

no. why?

the old DC puts it there before it leaves the cluster.
not doing this would unnecessarily increase resource unavailability.



More information about the Linux-HA mailing list