I've found in the mailing-list messages the syntax I could have written
with crm configure edit , something like :
order order-g-FS inf: ( fs-A fs-B fs-C fs-D fs-E ) ( exportfs-fs-A
exportfs-fs-B exportfs-fs-C exportfs-fs-D exportfs-fs-E )
right ?
But with my pacemaker release , crm configure edit returns a syntax error
around the first "("
so I think it is not supported with my release 1.1.5-5 , right ?
Thanks for confirmation.
Regards
Alain
De : [email protected]
A : General Linux-HA mailing list <[email protected]>
Date : 27/07/2012 12:47
Objet : Re: [Linux-HA] Antw: How to configure ordered sets of unordered
resources as described in Pacemaker doc ?
Envoyé par : [email protected]
Hi,
OK for mount of FS , that's not the real thing which matters for me,
but I'm quite sure that the parallelisation of exportfs stop, when the
OCF_RESKEY_wait_for_leasetime is set, is
valuable and even quite mandatory, as I do not want to add the sleep of
the 5 exportfs even with a reduced value
for the wait_for_leastime which is 90s by default. I would like to set it
to around 10s, but in parallel for the 5 exportfs.
Without paralleization, NFS clients will for sure get timeouts before the
end of migration of the FS and exportfs resource group.
Anyway, my question was more about the configuration of 6.6 example than
the behavior of server nfs in HA ... ;-)
Regards
Alain
De : "Ulrich Windl" <[email protected]>
A : <[email protected]>
Date : 27/07/2012 11:47
Objet : [Linux-HA] Antw: How to configure ordered sets of unordered
resources as described in Pacemaker doc ?
Envoyé par : [email protected]
Hi!
While your idea sounds good, I doubt whether parallel mounts being tried
are actually being performed in parallel, just as the exportfs operations.
They all access some common data structures in the kernel, I guess. In
that case, the timeout values may need adjustments.
Despite of that some RAs may show amazing behavior if executed in parallel
(I guess) ;-)
Regards,
Ulrich
>>> <[email protected]> schrieb am 27.07.2012 um 09:15 in Nachricht
<of7cf1dd89.6edcc5c6-onc1257a48.0025bf70-c1257a48.0027c...@bull.net>:
> Hi
>
> For now I had a group with several Filesystem resources followed by the
> exportfs like this :
> group g-FS-EXPORTED fs-A fs-B fs-C fs-D fs-E exportfs-fs-A
> exportfs-fs-B exportfs-fs-C exportfs-fs-D exportfs-fs-E \
>
> Now, I would like to have all the FS mounted before all the exportfs BUT
> with sequential=false for all Filesystem primitives and sequential=false
> also for all exportfs primitives.
>
> I saw in the Pacemaker Configuration Explained documentation the
> Example 6.11. Ordered sets of unordered resources
> with two ressources A & B starting in parallel and before two ressources
C
> & D starting also starting in parallel. I think this
> is exactly what I need.
>
> But :
>
> 1/ I have to remove the group configuration g-FS-EXPORTED , right ?
> or could I have such constraints "inside" the group itself ?
> (based on documentation, I don't think so)
>
> 2/ How can I enter the ordered set of unordered resources in the
> configuration ?
> (in documentation, the examples are given in xml, whereas we can't
edit
> the xml cib file,
> and in crm configure order, I can't see the way to do it :
> usage: order <id> score-type: <first-rsc>[:<action>]
> <then-rsc>[:<action>] [symmetrical=<bool>]
>
> 3/ After this configuration, that means that I can't manage the start or
> stop of all these resources with only one command
> as it was the case with the group ? meaning that I have to launch a
> start command on the 10 primitives ? instead of
> the start command on the group ?
>
> Thanks for your help on this.
> Alain
>
> _______________________________________________
> 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
