[Openstack] nova backup - instances unreachable
John Petrini
jpetrini at coredial.com
Fri Jan 20 18:18:12 UTC 2017
Hi All,
Following up after making this change. Adding write permissions to the
images pool in Ceph did the trick and RBD snapshots now work. However the
instance is still paused for the duration of the snapshot. Is it possible
to do a live snapshot without pausing the instance?
Thanks,
John
On Fri, Jan 13, 2017 at 5:49 AM, Eugen Block <eblock at nde.ag> wrote:
> Thanks,
>
> for anyone interested in this issue, I filed a bug report:
> https://bugs.launchpad.net/nova/+bug/1656242
>
>
> Regards,
> Eugen
>
>
> Zitat von Mohammed Naser <mnaser at vexxhost.com>:
>
> It is likely because this has been tested with QEMU only. I think you
>> might want to bring this up with the Nova team.
>>
>> Sent from my iPhone
>>
>> On Jan 12, 2017, at 11:28 AM, Eugen Block <eblock at nde.ag> wrote:
>>>
>>> I'm not sure if this is the right spot, but I added some log statements
>>> into driver.py.
>>> First, there's this if-block:
>>>
>>> if (self._host.has_min_version(MIN_LIBVIRT_LIVESNAPSHOT_VERSION,
>>> MIN_QEMU_LIVESNAPSHOT_VERSION,
>>> host.HV_DRIVER_QEMU)
>>> and source_type not in ('lvm')
>>> and not CONF.ephemeral_storage_encryption.enabled
>>> and not CONF.workarounds.disable_libvirt_livesnapshot):
>>> live_snapshot = True
>>> [...]
>>> else:
>>> live_snapshot = False
>>>
>>> And I know that it lands in the else-statement. Turns out that
>>> _host.has_min_version is "false", because of host.HV_DRIVER_QEMU. We are
>>> running on Xen hypervisors. So I tried it with host.HV_DRIVER_XEN and now
>>> nova-compute says:
>>>
>>> [instance: 14b75237-7619-481f-9636-792b64d1be17] instance snapshotting
>>> [instance: 14b75237-7619-481f-9636-792b64d1be17] Beginning live
>>> snapshot process
>>>
>>> Now I'm waiting for the result, but at least the VM is still running, so
>>> it looks quite promising...
>>>
>>> And there it is:
>>>
>>> [instance: 14b75237-7619-481f-9636-792b64d1be17] Snapshot image upload
>>> complete
>>>
>>> I'm testing the image now, and it works!
>>>
>>> Now the question is, why is it defaulting to HV_DRIVER_QEMU and is it
>>> really necessary to change this directly in the code? Is there any other
>>> way?
>>>
>>> Regards,
>>> Eugen
>>>
>>> Zitat von Eugen Block <eblock at nde.ag>:
>>>
>>> Yes, I truncated the file and uploaded it:
>>>>
>>>> http://dropcanvas.com/ta7nu
>>>> (First time I used this service, please give me feedback if this
>>>> doesn't work for you)
>>>>
>>>> I see the "Beginning cold snapshot process" message, but I don't know
>>>> why. Any help is appreciated!
>>>>
>>>> Regards,
>>>> Eugen
>>>>
>>>>
>>>> Zitat von Mohammed Naser <mnaser at vexxhost.com>:
>>>>
>>>> Would you be able to share the logs of a full snapshot run with the
>>>>> compute node in debug?
>>>>>
>>>>> Sent from my iPhone
>>>>>
>>>>> On Jan 12, 2017, at 7:47 AM, Eugen Block <eblock at nde.ag> wrote:
>>>>>>
>>>>>> That's strange, I also searched for this message, but nothing there.
>>>>>> I have debug logs enabled on compute node but I don't see anything
>>>>>> regarding ceph. No matter, what I do, my instance is always shutdown before
>>>>>> a snapshot is taken. What else can I try?
>>>>>>
>>>>>>
>>>>>> Zitat von John Petrini <jpetrini at coredial.com>:
>>>>>>
>>>>>> Mohammed,
>>>>>>>
>>>>>>> It looks like you may be right. Just found the permissions issue in
>>>>>>> the
>>>>>>> nova log on the compute node.
>>>>>>>
>>>>>>> 4-e8f52e4fbcfb 691caf1c10354efab3e3c8ed61b7d89a
>>>>>>> 49bc5e5bf2684bd0948d9f94c7875027 - - -] Performing standard snapshot
>>>>>>> because direct snapshot failed: no write permission on storage pool
>>>>>>> images
>>>>>>>
>>>>>>> I'm going to test the change and will send an update you all with the
>>>>>>> results.
>>>>>>>
>>>>>>> Thank You,
>>>>>>>
>>>>>>> ___
>>>>>>>
>>>>>>> John Petrini
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>> Yes, we are also running Mitaka and I also read Sebastien Han's
>>>>>>>> blogs ;-)
>>>>>>>>
>>>>>>>> our snapshots are not happening at the RBD level,
>>>>>>>>
>>>>>>>>> they are being copied and uploaded to glance which takes up a lot
>>>>>>>>> of space
>>>>>>>>> and is very slow.
>>>>>>>>>
>>>>>>>>>
>>>>>>>> Unfortunately, that's what we are experiencing, too. I don't know if
>>>>>>>> there's something I missed in the nova configs or somewhere else,
>>>>>>>> but I'm
>>>>>>>> relieved that I'm not the only one :-)
>>>>>>>>
>>>>>>>> While writing this email I searched again and found something:
>>>>>>>>
>>>>>>>> https://specs.openstack.org/openstack/nova-specs/specs/mitak
>>>>>>>> a/implemented/rbd-instance-snapshots.html
>>>>>>>>
>>>>>>>> https://review.openstack.org/#/c/205282/
>>>>>>>>
>>>>>>>> It seems to be implemented already, I'm looking for the config
>>>>>>>> options to
>>>>>>>> set. If you manage to get nova to make rbd snapshots, please let me
>>>>>>>> know ;-)
>>>>>>>>
>>>>>>>> Regards,
>>>>>>>> Eugen
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Zitat von John Petrini <jpetrini at coredial.com>:
>>>>>>>>
>>>>>>>> Hi Eugen,
>>>>>>>>
>>>>>>>>>
>>>>>>>>> Thanks for the response! That makes a lost of sense and is what I
>>>>>>>>> figured
>>>>>>>>> was going on but I missed it in the documentation. We use Ceph as
>>>>>>>>> well and
>>>>>>>>> I had considered doing the snapshots at the RBD level but I was
>>>>>>>>> hoping
>>>>>>>>> there was someway to accomplish this via nova. I came across this
>>>>>>>>> Sebastien
>>>>>>>>> Han write-up that claims this functionality was added to Mitaka:
>>>>>>>>> http://www.sebastien-han.fr/blog/2015/10/05/openstack-nova-
>>>>>>>>> snapshots-on-ceph-rbd/
>>>>>>>>>
>>>>>>>>> We are running Mitaka but our snapshots are not happening at the
>>>>>>>>> RBD
>>>>>>>>> level,
>>>>>>>>> they are being copied and uploaded to glance which takes up a lot
>>>>>>>>> of space
>>>>>>>>> and is very slow.
>>>>>>>>>
>>>>>>>>> Have you or anyone else implemented this in Mitaka? Other than
>>>>>>>>> Sebastian's
>>>>>>>>> blog I haven't found any documentation on this.
>>>>>>>>>
>>>>>>>>> Thank You,
>>>>>>>>>
>>>>>>>>> ___
>>>>>>>>>
>>>>>>>>> John Petrini
>>>>>>>>>
>>>>>>>>> On Wed, Jan 11, 2017 at 3:32 AM, Eugen Block <eblock at nde.ag>
>>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> this seems to be exptected, the docs say:
>>>>>>>>>>
>>>>>>>>>> "Shut down the source VM before you take the snapshot to ensure
>>>>>>>>>> that all
>>>>>>>>>> data is flushed to disk."
>>>>>>>>>>
>>>>>>>>>> So if the VM is not shut down, it's freezed to prevent data loss
>>>>>>>>>> (I
>>>>>>>>>> guess). Depending on your storage backend, there are other ways to
>>>>>>>>>> perform
>>>>>>>>>> backups of your VMs.
>>>>>>>>>> We use Ceph as backend for nova, glance and cinder. Ceph stores
>>>>>>>>>> the
>>>>>>>>>> disks,
>>>>>>>>>> images and volumes as Rados block device objects. We have a
>>>>>>>>>> backup script
>>>>>>>>>> that creates snapshots of these RBDs, which are exported to our
>>>>>>>>>> backup
>>>>>>>>>> drive. This way the running VM is not stopped or freezed, the user
>>>>>>>>>> doesn't
>>>>>>>>>> notice any issues. Unlike a nova snapshot, the rbd snapshot is
>>>>>>>>>> created
>>>>>>>>>> immediately within a few seconds. After a successful backup the
>>>>>>>>>> snapshots
>>>>>>>>>> are removed.
>>>>>>>>>>
>>>>>>>>>> Hope this helps! If you are interested in Ceph, visit [1].
>>>>>>>>>>
>>>>>>>>>> Regards,
>>>>>>>>>> Eugen
>>>>>>>>>>
>>>>>>>>>> [1] http://docs.ceph.com/docs/giant/start/intro/
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Zitat von John Petrini <jpetrini at coredial.com>:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Hello,
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> I've just started experimenting with nova backup and discovered
>>>>>>>>>>> that
>>>>>>>>>>> there
>>>>>>>>>>> is a period of time during the snapshot where the instance
>>>>>>>>>>> becomes
>>>>>>>>>>> unreachable. Is this behavior expected during a live snapshot?
>>>>>>>>>>> Is there
>>>>>>>>>>> any
>>>>>>>>>>> way to prevent this?
>>>>>>>>>>>
>>>>>>>>>>> ___
>>>>>>>>>>>
>>>>>>>>>>> John Petrini
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>> Eugen Block voice : +49-40-559 51
>>>>>>>>>> 75
>>>>>>>>>> NDE Netzdesign und -entwicklung AG fax : +49-40-559 51
>>>>>>>>>> 77
>>>>>>>>>> Postfach 61 03 15
>>>>>>>>>> D-22423 Hamburg e-mail : eblock at nde.ag
>>>>>>>>>>
>>>>>>>>>> Vorsitzende des Aufsichtsrates: Angelika Mozdzen
>>>>>>>>>> Sitz und Registergericht: Hamburg, HRB 90934
>>>>>>>>>> Vorstand: Jens-U. Mozdzen
>>>>>>>>>> USt-IdNr. DE 814 013 983
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> _______________________________________________
>>>>>>>>>> Mailing list: http://lists.openstack.org/cgi
>>>>>>>>>> -bin/mailman/listinfo/openstac
>>>>>>>>>> k
>>>>>>>>>> Post to : openstack at lists.openstack.org
>>>>>>>>>> Unsubscribe : http://lists.openstack.org/cgi
>>>>>>>>>> -bin/mailman/listinfo/openstac
>>>>>>>>>> k
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> Eugen Block voice : +49-40-559 51 75
>>>>>>>> NDE Netzdesign und -entwicklung AG fax : +49-40-559 51 77
>>>>>>>> Postfach 61 03 15
>>>>>>>> D-22423 Hamburg e-mail : eblock at nde.ag
>>>>>>>>
>>>>>>>> Vorsitzende des Aufsichtsrates: Angelika Mozdzen
>>>>>>>> Sitz und Registergericht: Hamburg, HRB 90934
>>>>>>>> Vorstand: Jens-U. Mozdzen
>>>>>>>> USt-IdNr. DE 814 013 983
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Eugen Block voice : +49-40-559 51 75
>>>>>> NDE Netzdesign und -entwicklung AG fax : +49-40-559 51 77
>>>>>> Postfach 61 03 15
>>>>>> D-22423 Hamburg e-mail : eblock at nde.ag
>>>>>>
>>>>>> Vorsitzende des Aufsichtsrates: Angelika Mozdzen
>>>>>> Sitz und Registergericht: Hamburg, HRB 90934
>>>>>> Vorstand: Jens-U. Mozdzen
>>>>>> USt-IdNr. DE 814 013 983
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Mailing list: http://lists.openstack.org/cgi
>>>>>> -bin/mailman/listinfo/openstack
>>>>>> Post to : openstack at lists.openstack.org
>>>>>> Unsubscribe : http://lists.openstack.org/cgi
>>>>>> -bin/mailman/listinfo/openstack
>>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Eugen Block voice : +49-40-559 51 75
>>>> NDE Netzdesign und -entwicklung AG fax : +49-40-559 51 77
>>>> Postfach 61 03 15
>>>> D-22423 Hamburg e-mail : eblock at nde.ag
>>>>
>>>> Vorsitzende des Aufsichtsrates: Angelika Mozdzen
>>>> Sitz und Registergericht: Hamburg, HRB 90934
>>>> Vorstand: Jens-U. Mozdzen
>>>> USt-IdNr. DE 814 013 983
>>>>
>>>
>>>
>>>
>>> --
>>> Eugen Block voice : +49-40-559 51 75
>>> NDE Netzdesign und -entwicklung AG fax : +49-40-559 51 77
>>> Postfach 61 03 15
>>> D-22423 Hamburg e-mail : eblock at nde.ag
>>>
>>> Vorsitzende des Aufsichtsrates: Angelika Mozdzen
>>> Sitz und Registergericht: Hamburg, HRB 90934
>>> Vorstand: Jens-U. Mozdzen
>>> USt-IdNr. DE 814 013 983
>>>
>>>
>
>
> --
> Eugen Block voice : +49-40-559 51 75
> NDE Netzdesign und -entwicklung AG fax : +49-40-559 51 77
> Postfach 61 03 15
> D-22423 Hamburg e-mail : eblock at nde.ag
>
> Vorsitzende des Aufsichtsrates: Angelika Mozdzen
> Sitz und Registergericht: Hamburg, HRB 90934
> Vorstand: Jens-U. Mozdzen
> USt-IdNr. DE 814 013 983
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20170120/e465d07d/attachment.html>
More information about the Openstack
mailing list