[Pacemaker] Problems with Pacemaker 1.1.8 on F17

[Andrew Beekhof]

On Fri, Feb 22, 2013 at 2:21 PM, Lars Kellogg-Stedman <lars at oddbit.com> wrote:
> The "switch" is a standard Linux bridge device, and the two systems are KVM
> virtual machines attached to the same bridge.

You'd think that would help, but
https://bugzilla.redhat.com/show_bug.cgi?id=880035 suggests otherwise.
I have one remaining fedora machine where KVM clusters still work, I
don't think I'll ever update it now.

> Are there known problems with
> this configuration?  I' can't get corosync to stay up for more than a few
> minutes.  I would simply use a vrrp or carp solution, but I need a service
> to follow the vip and there aren't many alternatives out there.
>
> I'll see if using udpu stabilizes things.
>
>
> On Thu, Feb 21, 2013 at 10:16 PM, Andrew Beekhof <andrew at beekhof.net> wrote:
>>
>> On Fri, Feb 22, 2013 at 1:00 PM, Lars Kellogg-Stedman <lars at oddbit.com>
>> wrote:
>> >
>> >> Did you define a recurring monitor action?
>> >>
>> >
>> > You know, upon reflection, probably not.  I started with "op monitor
>> > interval=10s", but because of all the problems I ran into with pcs
>> > complaining about systemd: services I think I dropped everything except
>> > the
>> > service name and agent in order to simplify the problem. I'll try it
>> > again
>> > with an explicit monitor action.
>> >
>> >>
>> >> What does corosync.conf look like?
>> >
>> >
>> > Like this: https://gist.github.com/larsks/5010169
>> >
>> > This was generated using the "pcs corosync configure puppet puppet0
>> > puppet1".
>> >
>>
>> Perhaps enable udpu (unicast) there.  Corosync doesn't seem to like your
>> switch.
>
>