On 8/15/07, Klemens Kittan <[EMAIL PROTECTED]> wrote:
> Am Tuesday, 14. August 2007 16:54 schrieb James Oakley:
> > On Tuesday 14 August 2007 8:29 am, Andrew Beekhof wrote:
> > > I'm guessing that the resource failed to stop
> > > And since stonith is not enabled there is nothing we can do (since
> > > it's not safe to start it somewhere else while its still, potentially,
> > > active on the first node).
> >
> > Ah. I see that the stop failure is a special case. I was trying to induce a
> > start failure. I just tried again, renaming the cups config file instead of
> > the daemon binary so that the stop action would succeed.
> >
> > It still doesn't failover, though.
> >
> > > But thats just a guess since you didn't include logs or the complete CIB
> >
> > I have attached both this time.
> >
> > Thanks,
>
> Hi,
>
> we have no solution, but a problem that looks very similar:
> We have a drbd master/slave resource and a resource group
> including an IPAddr.
>
> We want the resource group to run on the node that is drbd-master
> and need these failover behaviours:
> 1. Master role moves (e.g. after stopping drbd on the Master node),
>     the group should follow.
> 2. If the group fails (e.g. after ifconfig bond0:0 down),
>     drbd Master and after that the resource group itself
>     should move to the other node.

if the group cant run anywhere, should drbd run (or at least me in master mode)?

>
> With these colocation and order constraints:
>
> <rsc_colocation id="group_on_drbd0" from="group" to_role="Master" to="ms_drbd"
> score="INFINITY"/>
> <rsc_colocation id="group_on_drbd1" from="ms_drbd" from_role="Stopped"
> to="group" score="-INFINITY"/>
> <rsc_colocation id="group_on_drbd2" from="ms_drbd" from_role="Slave"
> to="group" score="-INFINITY"/>
> <rsc_order id="drbd_promote" from="group" action="start" to="ms_drbd"
> to_action="promote"/>
> <rsc_order id="drbd_demote" from="ms_drbd" action="demote" to="group"
> to_action="stop"/>
>
> only behaviour 1. is achieved (in fact, the first colocation and the first
> order rule are enough for that).
> How do we achieve behaviour 2.? Do we need any rule with from="ms_drbd"
> at all? You have none such rule in your cib.xml but if I understand right you
> also expect the drbd Master to move in case of a failure of any other
> resource, right?
>
> Thanks for your help!
> Klemens
>
> --
> Klemens Kittan
> Systemadministrator
>
> Uni-Potsdam, Inst. f. Informatik
> August-Bebel-Str. 89
> 14482 Potsdam
>
> Tel.    :   +49-331-977/3125
> Fax.    :   +49-331-977/3122
> eMail   : [EMAIL PROTECTED]
>
> gpg --recv-keys --keyserver wwwkeys.de.pgp.net 6EA09333
>
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>
>
_______________________________________________
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