[Pacemaker] Live migration with order constraints

[Andreas Kurz]

Replying to myself ....

On 11/11/2011 12:43 AM, Andreas Kurz wrote:
> On 11/10/2011 07:59 PM, Dmitry Golubev wrote:
>>>>> Actually, it's called ManageVE. Well, I wonder what else an RA
>>>>> could do other than manage :->
>>>>
>>>> ah, yes .. ManageVE ... thx 
>>>
>>> What I meant to say is that it's really hard to recall such a
>>> name.
>>
>> With all due respect, ManageVE does not have live migration. It just manages
>> start/stop for a single virtual machine. What I need is to live-migrate a group
>> of virtual machines which reside on one filesystem all together to a different
>> node. Given that there might be more than one such group. Yes, it can be done
>> having a resource for each virtual machine - i wouldn't object to that at all,
>> but they must have live migration together with their underlying filesystem. I
>> can't have cluster filesystem there, which could be mounted on both nodes and
>> deliver the possibility to live-migrate everything without a glich or remounting,
>> but unfortunately OpenVZ supports quotas only on ext2/3/4... so no OCFS or
>> similar things for me :(
> 
> ManageVE has migration support using chkpt/restore since resource-agents
> version 1.0.4. .... but if I understand the OpenVZ migration concept
> correct ... please someone correct me if I'm wrong! ... there is no need
> for a shared storage.

Focused too much on the migration feature ... of course the not-shared
storage should be highly available.

> 
> The vzmigrate script rsyncs complete data, config and state between
> nodes .... no shared storage needed.
> 
> Of course you would need twice the diskspace, but this is also true for
> DRBD replication. Extending ManageVE to use vzmigrate for live migration
> looks quite straight forward to me.

So this would mean four times the disk space ... suboptimal ... maybe
someone else has a better idea.

Regards,
Andreas

-- 
Need help with Pacemaker?
http://www.hastexo.com/now

> 
>>
>> My resource agent correctly does that thing with OpenVZ restart on the second
>> node, but no live migration for now. Live migration (no matter, whether it is
>> OpenVZ or KVM) depends on filesystem being either present on both nodes
>> simultaneously (impossible with ext2/3/4) or unmounted/remounted (which requires
>> order and location constraints). If there is no clean solution, is it really
>> impossible to have one resource agent moving another resource (by issuing a
>> commandline instruction to the cluster resource manager)?
> 
> So in fact you would only need to have two independent filesystems per
> VM group running, one on each node with local storage.
> 
> Regards,
> Andreas
> 
> 
> 
> 
> _______________________________________________
> 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