[Pacemaker] How to tell pacemaker to start exportfs after filesystem resource

[Aleksander Malaev]

I fixed it.
Reason was in exportfs resource monitor. On startup it check was it already
started:
        # "grep -z" matches across newlines, which is necessary as
        # exportfs output wraps lines for long export directory names
        exportfs | grep -zqs
"${OCF_RESKEY_directory}[[:space:]]*${OCF_RESKEY_clientspec}"
But after node reboot nfs-state files are not in clean state for a reason
and it caused the monitor action to return 0 state that was a reason of
failed cluster.

I fixed it by moving /var/lib/nfs to share, and i placed pipefs in /var/lib
- so now exportfs, if /var/lib/nfs/etab isn't accessible, returns no exports
on startup.

Filesystem resource is working as designed.

Thanks all for help!

2011/6/22 Aleksander Malaev <amalaev at alt-lan.ru>

> Also I've seen that resources started in parralel:
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> ping:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> ping:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-fs:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-fs:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-dlm:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-dlm:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-o2cb:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-o2cb:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> st-null:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> st-null:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> portmap:0#011(msk-nfs-gw01)
>  Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> nfs-kernel-server:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> portmap:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> nfs-kernel-server:1#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-nfs2-ip#011(msk-nfs-gw02)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-nfs1-ip#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-share:0#011(msk-nfs-gw01)
> Jun 22 16:18:49 msk-nfs-gw01 pengine: [1907]: notice: LogActions: Start
> res-share:1#011(msk-nfs-gw02)
>
> How can it be?
>
> 2011/6/22 Aleksander Malaev <amalaev at alt-lan.ru>
>
>> I created bug in bugzilla:
>> http://developerbugs.linux-foundation.org/show_bug.cgi?id=2609
>>
>> It seems all is like I thought, Filesystem resource is returned success
>> and pacemaker continue to start resources after it, but it have not mounted
>> fs yet and exportfs failed on export operation.
>> In logs i can see:
>> Jun 22 16:18:51 msk-nfs-gw02 exportfs[1955]: ERROR:
>> /media/media0/nfsroot/export1 does not exist or is not a directory
>>
>> Jun 22 16:18:55 msk-nfs-gw02 exportfs[2042]: ERROR:
>> /media/media0/nfsroot/export1 does not exist or is not a directory
>>
>> 2011/6/21 Dejan Muhamedagic <dejanmm at fastmail.fm>
>>
>>> On Tue, Jun 21, 2011 at 08:30:39PM +0200, Pawel Warowny wrote:
>>> > Dnia Tue, 21 Jun 2011 16:23:07 +0200
>>> > Dejan Muhamedagic <dejanmm at fastmail.fm> napisał(a):
>>> >
>>> > Hi
>>> >
>>> > Sorry to bother in this thread, but:
>>> >
>>> > > If you need to do so (there's actually start-delay, but it
>>> > > should be deprecated),
>>> >
>>> > I use start-delay for starting kvm virtualized guest one after another.
>>> > If they all start at once and in the background drbd is syncing, the
>>> > load is immense.
>>> >
>>> > How can I achieve this without start-delay?
>>>
>>> I think you cannot, not yet. There should be some support for
>>> this coming, but I'm not sure about the current plans.
>>>
>>> In this case the use of start-delay is fine. What I referred to
>>> was using it to make sure that the previous resource started.
>>>
>>> Thanks,
>>>
>>> Dejan
>>>
>>> > Best regards
>>> > --
>>> > Pawel Warowny
>>>
>>>
>>>
>>> > _______________________________________________
>>> > Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> > http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>> >
>>> > Project Home: http://www.clusterlabs.org
>>> > Getting started:
>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> > Bugs:
>>> http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>>>
>>>
>>> _______________________________________________
>>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>>
>>> Project Home: http://www.clusterlabs.org
>>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> Bugs:
>>> http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>>>
>>
>>
>>
>> --
>> С уважением,
>> Александр Малаев
>> +7-962-938-9323
>>
>
>
>
> --
> С уважением,
> Александр Малаев
> +7-962-938-9323
>



-- 
С уважением,
Александр Малаев
+7-962-938-9323
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20110624/b7bd02df/attachment-0003.html>