[Pacemaker] RHEL6 - clam + pacemaker + clvmd

Vladislav Bogdanov bubble at hoster-ok.com
Mon Jul 1 12:10:17 UTC 2013


01.07.2013 14:04, Nikola Ciprich wrote:
>> I actually did that myself, but I wouldn't recommend that way unless you
>> are familiar with all that. You may search through archives and look at
>> Andrew's blog (blog.clusterlabs.org, notably
>> http://blog.clusterlabs.org/blog/2012/pacemaker-and-cluster-filesystems/) for
>> additional details. You may also look at my slides at
>> http://lvee.org/uploads/image_upload/file/267/Linux_Clusters_at_LVEE_2013.pdf
>>
>> In short - that is really complicated to backport everything to RHEL6.
>>
>> Vladislav
> 
> Hello once more,
> 
> thanks a lot for the links, interesting reading.. So I'd bother You with one
> last question if I can..
> 
> I'd like to stick to RHEL6, don't have problem with compiling my own (newer)
> pacemaker/corosync/whatever packages. I'd like to keep using pacemaker though
> and I need clvmd as well. Don't really care whether I keep using corosync or
> switch to cman. Would You recommend one combination to me? According to Your

Sorry, no. That's up to you to decide. I go with corosync2 only because
I'm pretty sure I can put that all into "Works for me" state now and in
a future.

> slides, to me safe choice seems to be cman + pacemaker + (patched) clvmd,
> better for futer could be corosync + pacemaker + clvmd (+ future patch).

Not exactly.
now - cman + pacemaker + future patch for clvmd,
future - corosync2 + pacemaker + dlm4 + clvmd with existing patch.

If you go with corosync2 you should realize that there are big issues
running clustered filesystems with new corosync and old kernel.

Actually, in future RHEL (and in fedora 17+) gfs and dlm communicate in
kernel space rather than in userspace fro EL6. That's why gfs_controld
(which is not needed in future releases) and ocfs_controld need to be
ported to corosync2 for EL6. I did port former (thanks to David for not
removing fs_control functionality from dlm), but I won't support that
port except for myself. OCFS part is completely unclear for me.


> 
> While I'm not a cluster guru at all, I've got some experience with pacemaker
> and don't have problem building and maintaining my own packages as well as
> patching stuff...

You may try.
I would say that it is possible to make everything work.

> 
> thanks a lot in advance for this last advice :)
> 
> BR
> 
> nik
> 
> 
> 
> 
> 
>>
>>
>> _______________________________________________
>> 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://bugs.clusterlabs.org
>>
> 
> 
> 
> _______________________________________________
> 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://bugs.clusterlabs.org
> 





More information about the Pacemaker mailing list