Andrew Beekhof a écrit :
> On Mon, Sep 21, 2009 at 5:40 PM, Jean Baptiste Favre
> <[email protected]> wrote:
>> Hello,
>> Finally succeed to update pacemaker. Got a problem reported 2 days ago
>> to the mailinglist:
>> Failed to fetch
>> http://people.debian.org/~madkiss/ha-corosync/dists/lenny/main/binary-amd64/cluster-agents_1.0+hg20090915-1~bpo50+1_amd64.deb
>> Size mismatch
>>
>> Workaround was to completely uninstall previous heartbeat version, then
>> download and manually install cluster-agents package. It complains
>> because heartbeat is not installed. Then install heartbeat and
>> pacemaker-heartbeat. cluster-agents install will be completed at this step.
>>
>> Then I figure I have to keep old repository (deb
>> http://people.debian.org/~madkiss/ha lenny main) in
>> /etc/apt/sources.list.d/pacemaker.list because of
>> pacemaker-mgmt-heartbeat and pacemaker-mgmt-heartbeat-client I currently
>> use. They are not in the new repository.
>>
>> Well, after all I began to read you howto.
>> As it's writen for openais, will I have to install it instead of openais
>> to get ocfs better integration or can I do it as well with heartbeat ?
>
> Nope. The only distro that ever had heartbeat/ocfs2 integration was
> SLES10 but the patches were rejected by upstream and so dropped for
> SLES11.
>
>> I will try to adapt your howto for heartbeat this evening and see what
>> happen ;-)
>
> big explosions. openais is the required stack for the dlm and ocfs2 pieces.
Well, hopefully I read your mail before testing ;-)
For obvious reasons, I had to keep one more time the old repository:
openais get installed without init script :-)
BTW, got my test cluster back based on openais. First tests for OCFS2
don't seem to work: controld pacemaker plugin complains about missing
dlm_controld.pcmk which I can not find. dlm_controld is available in
Lenny as part of cman package, but conflict with openais.
Decided to get it a chance with following configuration. The idea is to
allow XEN VM start if, and only if, /cluster is mounted (OCFS2 filsystem
sharing XEN VM configurations files). A second mandatory prerequisite
will be that XEN VM DRBD's resource must be (of course) in primary state
in order the XEN VM can start.
Any comment welcome ;-)
JB
========================================================================
primitive Dom0ClusterOcfsDrbd ocf:heartbeat:drbd \
params drbd_resource="cluster-ocfs" \
op monitor interval="60s" \
meta target-role="started"
primitive Dom0ClusterOcfsMount ocf:heartbeat:Filesystem \
params device="/dev/drbd0" directory="/cluster" fstype="ocfs2" \
meta $id="ClusterOcfsMount-meta_attributes"
ms Dom0ClusterOcfsDrbdMaster Dom0ClusterOcfsDrbd \
meta master-max="2" master-node-max="1" clone-max="2"
clone-node-max="1" notify="true" target-role="Started" \
globally-unique="false"
clone Dom0ClusterOcfsMountClone Dom0ClusterOcfsMount \
meta master-max="2" master-node-max="1" clone-max="2"
clone-node-max="1" notify="true" target-role="Started"
colocation Dom0ClusterOcfsMountOnMaster inf: Dom0ClusterOcfsMountClone \
Dom0ClusterOcfsDrbd:Master
order Dom0ClusterOcfsMountAfterMaster inf: Dom0ClusterOcfsDrbd:promote \
Dom0ClusterOcfsMountClone:start
rsc_defaults $id="rsc-options" \
resource-stickiness="100"
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
