[one-users] Datastore maintenance

[André Monteiro]

Hi Jaime,

For the datastore maintenance, I've stopped all VMs on ONE 3.6 and all
images were saved with checkpoint. I upgraded to ONE 3.8.1 and
re-established the disk paths and NFS, and all machines resumed well!

I just had one problem on 3.8.1 migration, registered images don't work
with current .../remotes/tm/shared/ln script.

Sat Oct 27 12:41:39 2012 [TM][E]: clone: Command "cd
/opt/opennebula/var/datastores/0/143; cp -r
/opt/opennebula/var/datastores/ubuntu/snapshot.qcow2
/opt/opennebula/var/datastores/0/143/disk.0" failed: cp: cannot stat
`/opt/opennebula/var/datastores/ubuntu/snapshot.qcow2': No such file or
directory

My DATASTORE_LOCATION is correct - "/opt/opennebula/var/datastores" -  but
my registered image source was
"/opt/opennebula_shared/golden-images/ubuntu/snapshow.qcow2" and the
.../remotes/tm/shared/ln script converted into
"/opt/opennebula/var/datastores/ubuntu/snapshot.qcow2". I've looked into
the script and I don't see how it can use images on some user folder, I had
to change it:

#modification by Andre Monteiro on 3.8.1 on /var/remotes/tm/shared/ln and
clone scripts on line 53
if [[ "$SRC" == *datastore* ]]; then
   SRC_PATH="${DST_DS_PATH}${SRC_ARG_PATH##$SRC_DS_PATH}"
else
   SRC_PATH=${SRC#*:}
Regards,

--
André Monteiro




On Fri, Oct 26, 2012 at 4:33 PM, Jaime Melis <jmelis at opennebula.org> wrote:

> Hi,
>
> In ONE 3.6 if you resume a VM and it doesn't work, it will go into FAIL,
> and once it's in fail you cannot bring it back to RUNNING again.
>
> I suggest you upgrade to 3.8.1 (not 3.8.0) so you don't risk losing VMs.
>
> We're about to release 3.8.1 (couple of hours luckily) which fixes a bug
> that affects VMs with CONTEXT
>
> cheers,
> Jaime
>
>
> On Fri, Oct 26, 2012 at 4:36 PM, André Monteiro <andre.monteir at gmail.com>wrote:
>
>> Hi Jaime,
>>
>> Thank you for the prompt response. I have a optic fiber multipath disk
>> shared with NFS to datastore and my images are very heterogeneous. I've
>> started to stop the 69 VMs page by page (10 at a time), and it is working
>> fine. I'll shutdown ONE 3.6 and Sunstone also.
>>
>> --
>> André Monteiro
>>
>>
>>
>>
>> On Fri, Oct 26, 2012 at 3:30 PM, Jaime Melis <jmelis at opennebula.org>wrote:
>>
>>> Hi,
>>>
>>> It really depends on your storage configuration.
>>>
>>> If you are using TM SSH (or using SHARED and all your VMS are CLONE =
>>> YES => images: persistent => no) AND you are *not* moving your system
>>> datastore, just the regular datastores, then you can just stop opennebula
>>> during the operations so no deploy or shutdown/save is performed in the
>>> meantime.
>>>
>>> Otherwise,  to be on the safe side I recommed you stop the vms and
>>> shutdown OpenNebula.
>>>
>>> cheers,
>>> Jaime
>>>
>>> On Fri, Oct 26, 2012 at 11:58 AM, André Monteiro <
>>> andre.monteir at gmail.com> wrote:
>>>
>>>> Hello again,
>>>>
>>>> My LUN disks, where datastore is located, are being moved to other
>>>> physical location. What is the best way to prepare this migration and all
>>>> VMs startup smoothly after remounting the disks?
>>>>
>>>>
>>>>    - Shutdown all VMS and restart them after?
>>>>    - Shutdown ONE main host/frontend blade?
>>>>    - Other?
>>>>
>>>>
>>>> --
>>>> André Monteiro
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Users mailing list
>>>> Users at lists.opennebula.org
>>>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>>>>
>>>>
>>>
>>>
>>> --
>>> Jaime Melis
>>> Project Engineer
>>> OpenNebula - The Open Source Toolkit for Cloud Computing
>>> www.OpenNebula.org | jmelis at opennebula.org
>>>
>>
>>
>
>
> --
> Jaime Melis
> Project Engineer
> OpenNebula - The Open Source Toolkit for Cloud Computing
> www.OpenNebula.org | jmelis at opennebula.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20121029/cc820314/attachment-0002.htm>