[Openstack] [OpenStack][Nova] Live Migration Error

Sébastien Han han.sebastien at gmail.com
Tue Jul 10 10:30:07 UTC 2012 

It's production ready, RedHat offers a commercial support on it.
Just keep in mind that it's owned by Redhat ;)


On Tue, Jul 10, 2012 at 12:24 PM, Leander Bessa Beernaert <
leanderbb at gmail.com> wrote:

> Is GlusterFS be more viable for a production environment?
>
>
> On Tue, Jul 10, 2012 at 11:17 AM, Sébastien Han <han.sebastien at gmail.com>wrote:
>
>> Great!
>>
>> The last time I ran the live-migration, it was with GlusterFS and CephFS
>> and I didn't changed any permissions in libvirt. I did the live-migration
>> with NFS once but it was in Diablo (horrible), I don't really remember my
>> setup. Maybe you should consider to try GlusterFS.
>>
>>
>> On Tue, Jul 10, 2012 at 12:07 PM, Leander Bessa Beernaert <
>> leanderbb at gmail.com> wrote:
>>
>>> That did! Thanks :)
>>>
>>> Do you by change have any pointer on getting the live-migration to work
>>> without running libvirt under root?
>>>
>>>
>>> On Tue, Jul 10, 2012 at 10:55 AM, Sébastien Han <han.sebastien at gmail.com
>>> > wrote:
>>>
>>>> Change the vncserver_listen to 0.0.0.0 and re-try the live-migration,
>>>> you should get better results :)
>>>>
>>>>
>>>>
>>>> On Tue, Jul 10, 2012 at 11:52 AM, Leander Bessa Beernaert <
>>>> leanderbb at gmail.com> wrote:
>>>>
>>>>> If i don't run libvirt with root, it can't write to the shared folder.
>>>>> It's the only way i've been able to get this to work. :S
>>>>>
>>>>> Below follows the configuration of one of the compute nodes. 10.0.1.1
>>>>> is the controller and 10.0.1.2 is the compute node.
>>>>>
>>>>> novncproxy_base_url=http://10.0.1.1:6080/vnc_auto.html
>>>>>> xvpvncproxy_base_url=http://10.0.1.1:6081/console
>>>>>> vncserver_proxyclient_address=10.0.1.2
>>>>>> vncserver_listen=10.0.1.2
>>>>>
>>>>>
>>>>> On Tue, Jul 10, 2012 at 10:49 AM, Sébastien Han <
>>>>> han.sebastien at gmail.com> wrote:
>>>>>
>>>>>> I forgot to ask, did you enable the vnc console?
>>>>>>
>>>>>> If so, with which parameters?
>>>>>>
>>>>>>
>>>>>> On Tue, Jul 10, 2012 at 11:48 AM, Sébastien Han <
>>>>>> han.sebastien at gmail.com> wrote:
>>>>>>
>>>>>>> Hi!
>>>>>>>
>>>>>>> Usually you get:
>>>>>>>
>>>>>>> 2012-07-09 13:58:27.179+0000: 10227: warning :
>>>>>>>> qemuDomainObjTaint:1134 : Domain id=2 name='instance-00000002'
>>>>>>>> uuid=57aca8a6-d062-4a08-8d87-e4d11d259ac7 is tainted: high-privileges
>>>>>>>
>>>>>>>
>>>>>>> when you change permission in libvirt (root I presumed) which is not
>>>>>>> necessary.
>>>>>>>
>>>>>>> 2012-07-10 09:26:02.244+0000: 10229: error : virExecWithHook:328 :
>>>>>>>> Cannot find 'pm-is-supported' in path: No such file or directory
>>>>>>>
>>>>>>>
>>>>>>> This error is harmless and can be easily solved by installing the
>>>>>>> following package:
>>>>>>>
>>>>>>> sudo apt-get install pm-utils -y
>>>>>>>
>>>>>>>
>>>>>>> Do you have something in the nova-scheduler logs?
>>>>>>>
>>>>>>> Cheers!
>>>>>>>
>>>>>>> On Tue, Jul 10, 2012 at 11:29 AM, Leander Bessa Beernaert <
>>>>>>> leanderbb at gmail.com> wrote:
>>>>>>>
>>>>>>>> Well i've checked the libvirt logs on both nodes and i found these
>>>>>>>> two lines:
>>>>>>>>
>>>>>>>> 2012-07-09 13:58:27.179+0000: 10227: warning :
>>>>>>>>> qemuDomainObjTaint:1134 : Domain id=2 name='instance-00000002'
>>>>>>>>> uuid=57aca8a6-d062-4a08-8d87-e4d11d259ac7 is tainted: high-privileges
>>>>>>>>> 2012-07-09 13:58:27.736+0000: 10226: error : qemuMonitorIORead:513
>>>>>>>>> : Unable to read from monitor: Connection reset by peer
>>>>>>>>
>>>>>>>>
>>>>>>>> The log is alos filled with the message below, it gets repeated
>>>>>>>> over and over and over.
>>>>>>>>
>>>>>>>> 2012-07-10 09:26:02.244+0000: 10229: error : virExecWithHook:328 :
>>>>>>>>> Cannot find 'pm-is-supported' in path: No such file or directory
>>>>>>>>> 2012-07-10 09:26:02.244+0000: 10229: warning : qemuCapsInit:856 :
>>>>>>>>> Failed to get host power management capabilities
>>>>>>>>
>>>>>>>>
>>>>>>>> On Tue, Jul 10, 2012 at 8:16 AM, Razique Mahroua <
>>>>>>>> razique.mahroua at gmail.com> wrote:
>>>>>>>>
>>>>>>>>> Hi Leander,
>>>>>>>>> try to check libvirtd.log files,
>>>>>>>>> is the instance still running on the first node while you are
>>>>>>>>> lauching the migration process ?
>>>>>>>>>
>>>>>>>>> Razique
>>>>>>>>>  *Nuage & Co - Razique Mahroua** *
>>>>>>>>> razique.mahroua at gmail.com
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Le 9 juil. 2012 à 16:09, Leander Bessa Beernaert a écrit :
>>>>>>>>>
>>>>>>>>> Ok, so i've updated to the test packages from
>>>>>>>>>
>>>>>>>>> The migration still fails, but i see no errors in the logs. I'm
>>>>>>>>> trying to migrate a VM with the m1.tiny flavor from one machine to another.
>>>>>>>>> Their hardware are identical and they have more than enough resources to
>>>>>>>>> support the m1.tiny flavor:
>>>>>>>>>
>>>>>>>>> cloud35         (total)                         4    3867     186
>>>>>>>>>> cloud35         (used_now)                      0     312       5
>>>>>>>>>> cloud35         (used_max)                      0       0       0
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> These are the logs from the origin compute node:
>>>>>>>>> http://paste.openstack.org/show/19319/  and  the destination
>>>>>>>>> compute node: http://paste.openstack.org/show/19318/ . The
>>>>>>>>> scheduler's log has no visible errors or stack traces.
>>>>>>>>>
>>>>>>>>> I'm still using nfsv4.
>>>>>>>>>
>>>>>>>>> Any ideas?
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On Fri, Jul 6, 2012 at 7:57 PM, Leander Bessa Beernaert <
>>>>>>>>> leanderbb at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> Thanks for the tip, it's a better than nothing :)
>>>>>>>>>>
>>>>>>>>>> Regards,
>>>>>>>>>> Leander
>>>>>>>>>>
>>>>>>>>>> On Fri, Jul 6, 2012 at 6:32 PM, Mandar Vaze / मंदार वझे <
>>>>>>>>>> mandarvaze at gmail.com> wrote:
>>>>>>>>>>
>>>>>>>>>>> Not sure if you are able to debug this, but a while ago there
>>>>>>>>>>> was a bug where instance.id was passed where instance.uuid was
>>>>>>>>>>> expected. This used to cause some problem.
>>>>>>>>>>> It looks like you are using distribution package rather than
>>>>>>>>>>> devstack installation, so it is likely that the issue is now fixed. Can you
>>>>>>>>>>> try latest packages (and/or try devstack if you can)
>>>>>>>>>>>
>>>>>>>>>>> I wish I could help more.
>>>>>>>>>>>
>>>>>>>>>>> -Mandar
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On Fri, Jul 6, 2012 at 3:26 PM, Leander Bessa Beernaert <
>>>>>>>>>>> leanderbb at gmail.com> wrote:
>>>>>>>>>>>
>>>>>>>>>>>> Hello,
>>>>>>>>>>>>
>>>>>>>>>>>> I've recently setup a system to test out the live migration
>>>>>>>>>>>> feature. So far i've been able to launch the instances with the shared nfs
>>>>>>>>>>>> folder. However, when i run the live-migration command i encounter this
>>>>>>>>>>>> error in the destination compute node:
>>>>>>>>>>>>
>>>>>>>>>>>> 2012-07-05 09:33:48 ERROR nova.manager [-] Error during
>>>>>>>>>>>>> ComputeManager.update_available_resource: Domain not found: no domain with
>>>>>>>>>>>>> matching id 2
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager Traceback (most recent
>>>>>>>>>>>>> call last):
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager   File
>>>>>>>>>>>>> "/usr/lib/python2.7/dist-packages/nova/manager.py", line 155, in
>>>>>>>>>>>>> periodic_tasks
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager     task(self, context)
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager   File
>>>>>>>>>>>>> "/usr/lib/python2.7/dist-packages/nova/compute/manager.py", line 2409, in
>>>>>>>>>>>>> update_available_resource
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager
>>>>>>>>>>>>> self.driver.update_available_resource(context, self.host)
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager   File
>>>>>>>>>>>>> "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/connection.py", line
>>>>>>>>>>>>> 1936, in update_available_resource
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager     'vcpus_used':
>>>>>>>>>>>>> self.get_vcpu_used(),
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager   File
>>>>>>>>>>>>> "/usr/lib/python2.7/dist-packages/nova/virt/libvirt/connection.py", line
>>>>>>>>>>>>> 1743, in get_vcpu_used
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager     dom =
>>>>>>>>>>>>> self._conn.lookupByID(dom_id)
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager   File
>>>>>>>>>>>>> "/usr/lib/python2.7/dist-packages/libvirt.py", line 2363, in lookupByID
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager     if ret is
>>>>>>>>>>>>> None:raise libvirtError('virDomainLookupByID() failed', conn=self)
>>>>>>>>>>>>> 2012-07-05 09:33:48 TRACE nova.manager libvirtError: Domain
>>>>>>>>>>>>> not found: no domain with matching id 2
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Any ideas on how to solve this?
>>>>>>>>>>>>
>>>>>>>>>>>> Regards,
>>>>>>>>>>>> Leander
>>>>>>>>>>>>
>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>> Mailing list: https://launchpad.net/~openstack
>>>>>>>>>>>> Post to     : openstack at lists.launchpad.net
>>>>>>>>>>>> Unsubscribe : https://launchpad.net/~openstack
>>>>>>>>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Mailing list: https://launchpad.net/~openstack
>>>>>>>>> Post to     : openstack at lists.launchpad.net
>>>>>>>>> Unsubscribe : https://launchpad.net/~openstack
>>>>>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Mailing list: https://launchpad.net/~openstack
>>>>>>>> Post to     : openstack at lists.launchpad.net
>>>>>>>> Unsubscribe : https://launchpad.net/~openstack
>>>>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20120710/b3fe14c3/attachment.html>

More information about the Openstack mailing list