[Users] Lun snapshots by a san
Itamar Heim
iheim at redhat.com
Sat Apr 20 17:37:44 UTC 2013
On 04/18/2013 12:08 PM, pierre Labanowski wrote:
> Le 18/04/2013 04:50, Shu Ming a écrit :
>> pierre Labanowski:
>>> Hello everybody,
>>>
>>> I have a question, or more precisely information or best practice.
>>>
>>> I use oVirt 3.1 with an attachment on fiber channel on my SAN. Makes it
>>> even the San snapshots that are replicated on a remote site.
>>>
>>> it's a procedure for adding a new "Storage" Domain to another Data
>>> Centers (oVirt) in order to recover the state of a VM snapshoted by
>>> the SAN?
>>
>> If I understand correctly, the remote SAN of you can snapshot the
>> storage domain(physical lun in the SAN) lively. And you hope to
>> recover the storage domain with the previous snapshot stored in the SAN.
>> So you create a new "storage" domain for another data center and
>> restart the VM from the newstorage domain. I am not sure if the VM
>> disks in the storage domain snapshot is in consistent date if the SAN
>> snapshot the VM in a running state. I think the VM OS should be
>> notified and make sure it is in consistent state before the SAN can do
>> the snapshot, also the VM OS should be notified with the finish of the
>> snapshotting..
>
> Thank you for your answer,
is the engine still intact and the problem is only with the storage (or
you can restore the engine and its db)?
currently you'd need to edit the LUN in the db, but iirc, alissa has
been working on ability to edit it in a less hacky manner (for nfs and
posixfs its already merged)
>
> Yes of course, yet a little fsck is not a problem when you can restore
> an entire VM after a big crash. But, create a new "storage" domain for
> another data center is forbiden because this a new "storage" domain is
> not empty, i have a error in ovirt when create a new "storage" domain
> with my snapshot, it's on this point that I have a problem, adding of
> lun snapshoted on a new "storage" domain
>
> Pierre
>
>>>
>>> I've read the documentation on Disaster Recovery but it's for the
>>> site-to-site failover (i don't have dual architecture of ovirt) while my
>>> demand here is for the state recovered a current domain. And only to
>>> export the VM(snapshoted by a san) for restore in the current storage
>>> (production).
>>>
>>> thank you
>>> Pierre
>>> _______________________________________________
>>> Users mailing list
>>> Users at ovirt.org
>>> http://lists.ovirt.org/mailman/listinfo/users
>>>
>>
>>
>
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
More information about the Users
mailing list