Ask Your Question
1

live migration fails when a cinder volume is attached to a running instance

asked 2013-08-02 12:47:45 -0500

Sifty gravatar image

updated 2013-08-03 15:57:50 -0500

I have glusterfs setup for nova volumes and cinder volumes and all appears to work fine, /var/lib/nova/instances on /export/gluster however when live migrating with a cinder volume attached - i get the following errors and the gluster nova volume for the instance disappears.

fe4eb488fb053ea4b2884d] [instance: f1c30a49-01c5-4fa1-9026-26b97098a04e] Unexpected error while running command.
Command: scp /var/lib/nova/instances/f1c30a49-01c5-4fa1-9026-26b97098a04e_resize/volume-c578a3dc-d24a-40d4-bef6-62d63357542b 172.18.6.14:/var/lib/nova/mnt/c18bbc08d124e44b956e836600894ec9/volume-c578a3dc-d24a-40d4-bef6-62d63357542b

ProcessExecutionError: Unexpected error while running command.
2013-08-02 18:37:15.728 8732 TRACE nova.openstack.common.rpc.amqp Command: scp /var/lib/nova/instances/f1c30a49-01c5-4fa1-9026-26b97098a04e_resize/volume-c578a3dc-d24a-40d4-bef6-62d63357542b 172.18.6.14:/var/lib/nova/mnt/c18bbc08d124e44b956e836600894ec9/volume-c578a3dc-d24a-40d4-bef6-62d63357542b

2013-08-02 18:37:15.728 8732 TRACE nova.openstack.common.rpc.amqp Stderr: '/var/lib/nova/instances/f1c30a49-01c5-4fa1-9026-26b97098a04e_resize/volume-c578a3dc-d24a-40d4-bef6-62d63357542b: No such file or directory\n'
2013-08-02 18:37:15.728 8732 TRACE nova.openstack.common.rpc.amqp
2013-08-02 18:37:36.265 8732 AUDIT nova.compute.resource_tracker [-] Auditing locally available compute resources
2013-08-02 18:37:36.299 8732 ERROR nova.virt.libvirt.driver [-] Getting disk size of instance-00000008: [Errno 2] No such file or directory: '/var/lib/nova/instances/f1c30a49-01c5-4fa1-9026-26b97098a04e/disk'

why is it using scp when the gluster filesystem is mounted on both systems so it can live migrate. shouldn't it just check the images are in the /var/lib/nova/instances folder.

Is there something I am missing ?

Thanks all

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2013-08-05 08:01:58 -0500

updated 2013-08-05 08:02:23 -0500

#1 Live Migration from gluster volume to another gluster volume would involve copy (scp) operation unless there is "sync" operation comes in action. <may be="" container="" to="" container="" sync="" type="" of="" feature?="">. #2 Mouting a same gluster volume hosting same cinder volumes on "different" host and resuming VM activities from the new server could be another flavor of live migration. And that would not require any copy operation but some consistency marker for cinder volume. So can you please shed some light on what type of setup you had, then it would be easier to debug the scp issues if any. Thanks, Chetan Risbud.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2013-08-02 12:47:45 -0500

Seen: 722 times

Last updated: Aug 03 '13