[Openstack-operators] Live Migration and instances stores

Jian Wen wenjianhn at gmail.com
Fri Nov 30 02:23:13 UTC 2012


2012/11/28 Dave Spano <dspano at optogenics.com>

> Not sure if this is sufficient for you or not, but I currently use the rbd
> portion of ceph for Openstack. I don't currently do live migrations because
> it's not feasible for me to setup the extra storage. Instead, I do block
> migrations, and these work rather well.
>
> The only bad part is that you have to release the floating ip, migrate the
> instance, and then reassign it to the vm. Otherwise, the vm's networking
> gets messed up. I use nova-network in multi-host mode, so I'm not sure if
> the same behavior would happen with Quantum.
>
> I fixed the bad part.
Fix will be released in grizzly.
https://bugs.launchpad.net/nova/+bug/966529

> Sébastian Han wrote a great article on block migration if you're
> interested.
>
> http://www.sebastien-han.fr/blog/2012/07/12/openstack-block-migration/
>
> I noted at the bottom of the article that the block migration
> functionality won't take affect with currently running instances. You have
> to terminate, then re-spawn them after you make your changes to nova.conf.
>
> Dave Spano
>
>
> ------------------------------
> *From: *"Jonathan Abdiel Gonzalez Valdebenito" <jonathan.abdiel at gmail.com>
> *To: *openstack-operators at lists.openstack.org
> *Sent: *Tuesday, November 27, 2012 11:38:24 AM
> *Subject: *[Openstack-operators] Live Migration and instances stores
>
>
> Hi all,
>
> First at all I'm glad to write to the list :D thanks for reading!
>
> I'm currently experimenting with lot of ideas to store the
> instances(/var/lib/nova/instances) and do a live migration with them, well
> here it's a tiny resume:
>
> DRBD + OCFS2: Works pretty well but as it depends on DRBD which it's a
> raid 1 system it can't grow any further than the initial size... bad idea
> if you want some scalability
>
> GlusterFS: A bit slow, it can take some minutes to replicate but works,
> the only problem that it has it's that you can't perform a live migration,
> the reason, cache != none or that's what the logs says
>
> Ceph: It runs pretty good tests, but it's slow with the machines in my
> lab, and throw some errors about being too slow. Anyway, I tried the live
> migration and it didn't work either.
>
> Now I'm currently starting my test with MooseFS but don't know yet if it
> will work, according to the openstack manual it's a good idea, but I will
> like to have the experience of the list on this matter, does anyone try any
> of this options or another successfully? or have other options that I
> haven't found yet?
>
>
> Thanks in advance for all you answer!
>
> --
> Greets,
>
>
>
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>
>
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>
>


-- 
Best,

Jian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20121130/f2cdd1f8/attachment.html>


More information about the OpenStack-operators mailing list