Hi,

On Thu, Jun 19, 2008 at 09:50:51AM +0200, Navatala wrote:
> Hi Dejan
> 
> thanks for the pointer, I went back to look at the evms side of
> things and as a simple troubleshooting test I tried mounting
> the file system from the second node (using evmsgui) it mounted
> successfully. The I restarted the cluster and guess what voila
> the resource now happily migrates to either node and will load
> correctly on either node.

Strange. However, the dangerous thing is that this evms_failover
resource agent reported the resource (volume) as started whereas
it was not. The RA is not allowed to lie. That needs to be
investigated.

> I don't know if this is normal, ie you need to manually load
> the volume on both nodes before HA can do it automatically
> during a failover or not.

I guess not. That would sort of defeat the purpose of the
solution.

> I will go over my notes again and perform a new install from
> scratch to see if I've missed something.

I doubt that a new installation would help.

> By the way does anyone know of a good active EVMS list I can
> subscribe to?
> 
> Finally thanks to everyone who helped out on this list, it was
> my first contact with the list and I have to say it was very
> positive, helpful and instructive.

Thanks,

Dejan

> TFTH

> 
> 
> 
> >>> 
> From: Dejan Muhamedagic<[EMAIL PROTECTED]>
> To:[email protected]
> Date: 18/06/08 12:15
> Subject: Re: [Linux-HA] resource only runs on one node
> Hi,
> 
> On Tue, Jun 17, 2008 at 07:56:04PM +0200, Navatala wrote:
> [snip]
> > remember I new to all of this, I assume it is coming from the
> > cib.xml.......however I may have misunderstood your question :)
> 
> Yes, sorry, I was not precise enough, I meant the evms_failover
> resource agent. It's not part of the heartbeat package.
> 
> > in the group I have three resources the first is
> > resource_evms_failover the second is resource_evms_fs and the
> > third is resource_ip. All three work fine on node1. 
> 
> The evms_failover resource starts on node2, but the Filesystem
> resource fails which implies that the former one probably didn't
> really start the resource, i.e. make the volume available. 
> 
> Thanks,
> 
> Dejan
> 
> > 
> > tia
> > 
> > 
> > >>> 
> > From: Dejan Muhamedagic<[EMAIL PROTECTED]>
> > To:[email protected]
> > Date: 17/06/08 18:53
> > Subject: Re: [Linux-HA] resource only runs on one node
> > Hi,
> > 
> > On Tue, Jun 17, 2008 at 04:43:57PM +0200, Navatala wrote:
> > > I knew I should have added something but was not sure which
> > > ones, here goes the cib.xml, pe-warn and messages log.
> > 
> > You can try hb_report next time to create a complete report.
> > 
> > > the problem node as you can see from messages log is sles10-node2
> > > I'm wondering if it might be a sticky resource issue
> > 
> > Jun 17 16:24:30 sles10-node2 ccm: [3101]: info: Hostname: sles10-node2
> > Jun 17 16:24:30 sles10-node2 heartbeat: [3016]: WARN: duplicate client add 
> > request [ccm] [3108]
> > Jun 17 16:24:30 sles10-node2 ccm: [3108]: info: Hostname: sles10-node2
> > 
> > Looks like heartbeat tries to start the ccm twice. What does your
> > ha.cf look like?
> > 
> > Jun 17 16:24:51 sles10-node2 crmd: [3106]: info: process_lrm_event: LRM 
> > operation resource_evms_failover_start_0 (call=5, rc=0) complete 
> > 
> > Looks like your evms resource starts fine. This is probably a
> > resource agent problem:
> > 
> > <primitive class="heartbeat" type="evms_failover" provider="heartbeat" 
> > id="resource_evms_failover">
> > 
> > Where is that one coming from?
> > 
> > Thanks,
> > 
> > Dejan
> > 
> > > let me know if  I need to add anything else
> > > 
> > > tia
> > > 
> > > 
> > > >>> 
> > > From: Andrew Beekhof<[EMAIL PROTECTED]>
> > > To:[email protected]
> > > Date: 17/06/08 15:57
> > > Subject: Re: [Linux-HA] resource only runs on one node
> > > On Tue, Jun 17, 2008 at 11:28, Navatala <[EMAIL PROTECTED]> wrote:
> > > > Hi
> > > >
> > > > as a HA newbie I've been trying to set up a simple cluster enabled file 
> > > > system.
> > > 
> > > is there such a thing? :)
> > > 
> > > >
> > > > I'm using evms and so far I have set everything up and it seems to be 
> > > > working with the caveat that it only works on one node. It wont run on 
> > > > the other node regardless of what I try to do, ie bring up the other 
> > > > node only, place the firs node in standby node, create a placement 
> > > > rule.... nothing seems to work.
> > > >
> > > > Both machines can see the shared disk.
> > > 
> > > logs? config?  give us something to go on :)
> > > _______________________________________________
> > > 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
> > _______________________________________________
> > 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
> _______________________________________________
> 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
_______________________________________________
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