[Rdo-list] Icehouse multi-node - Centos7 - live migration failed because of "network qbr no such device"

[Miguel Angel]

Hi Benoit,

  A manual virsh migration should fail, because the
network ports are not migrated to the destination host.

  You must investigate on the authentication problem itself,
and let nova handle all the underlying API calls which should happen...

  May be it's worth setting nova.conf to debug=True



---
irc: ajo / mangelajo
Miguel Angel Ajo Pelayo
+34 636 52 25 69
skype: ajoajoajo


2014-07-11 14:41 GMT+02:00 Benoit ML <ben42ml at gmail.com>:

> Hello,
>
> cat /etc/redhat-release
> CentOS Linux release 7 (Rebuilt from: RHEL 7.0)
>
>
> Regards,
>
>
> 2014-07-11 13:40 GMT+02:00 Boris Derzhavets <bderzhavets at hotmail.com>:
>
>  Could you please post /etc/redhat-release.
>>
>> Boris.
>>
>> ------------------------------
>> Date: Fri, 11 Jul 2014 11:57:12 +0200
>> From: ben42ml at gmail.com
>> To: rdo-list at redhat.com
>> Subject: [Rdo-list] Icehouse multi-node - Centos7 - live migration failed
>> because of "network qbr no such device"
>>
>>
>> Hello,
>>
>> I'm working on a multi-node setup of openstack Icehouse using centos7.
>> Well i have :
>>   - one controllor node with all server services thing stuff
>>   - one network node with openvswitch agent, l3-agent, dhcp-agent
>>   - two compute node with nova-compute and neutron-openvswitch
>>   - one storage nfs node
>>
>> NetworkManager is deleted on compute nodes and network node.
>>
>> My network use is configured to use vxlan.  I can create VM,
>> tenant-network, external-network, routeur, assign floating-ip to VM, push
>> ssh-key into VM, create volume from glance image, etc... Evrything is
>> conected and reacheable. Pretty cool :)
>>
>> But when i try to migrate VM things go wrong ...  I have configured
>>  nova, libvirtd and qemu to use migration through libvirt-tcp.
>> I have create and exchanged ssh-key for nova user on all  node. I have
>> verified userid and groupid of nova.
>>
>> Well nova-compute log, on the target compute node, :
>> 2014-07-11 11:45:02.749 6984 TRACE nova.compute.manager [instance:
>> a5326fe1-4faa-4347-ba29-159fce26a85c] RemoteError: Remote error:
>> Unauthorized {"error": {"m
>> essage": "The request you have made requires authentication.", "code":
>> 401, "title": "Unauthorized"}}
>>
>>
>> So well after searching a lots in all logs, i have fount that i cant
>> simply migration VM between compute node with a simple virsh :
>> virsh migrate instance-00000084 qemu+tcp://<dest>/system
>>
>> The error is :
>> erreur :Cannot get interface MTU on 'qbr3ca65809-05': No such device
>>
>> Well when i look on the source hyperviseur the bridge "qbr3ca65809"
>> exists  and have a network tap device.  And moreover i manually create
>> qbr3ca65809 on the target hypervisor, virsh migrate succed !
>>
>> Can you help me plz ?
>> What can i do wrong ? Perhpas neutron must create the bridge before
>> migration but didnt for a mis configuration ?
>>
>> Plz ask anything you need !
>>
>> Thank you in advance.
>>
>>
>> The full nova-compute log attached.
>>
>>
>>
>>
>>
>>
>> Regards,
>>
>> --
>> --
>> Benoit
>>
>> _______________________________________________ Rdo-list mailing list
>> Rdo-list at redhat.com https://www.redhat.com/mailman/listinfo/rdo-list
>>
>
>
>
> --
> --
> Benoit
>
> _______________________________________________
> Rdo-list mailing list
> Rdo-list at redhat.com
> https://www.redhat.com/mailman/listinfo/rdo-list
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rdoproject.org/pipermail/dev/attachments/20140711/c7223d5c/attachment.html>