Ask Your Question
0

Live VM migration (block-migration) fails

asked 2015-09-15 18:43:14 -0500

hardik italia gravatar image

updated 2015-09-16 03:15:43 -0500

Hi,

I need some help about VM live migration (block migration). I am running 1 controller and 2 compute nodes in a devstack environment with stable/kilo branch. I have configured nova.conf with live_migration_flag by following openstack documentation (http://docs.openstack.org/admin-guide... configure libvirtd to listen on tcp but I do not see vm migrate to new compute node when I use command

nova live-migrate --block-migrate UUID NEW_NODE

nova.conf (libvirt section)

[libvirt]
vif_driver = nova.virt.libvirt.vif.LibvirtGenericVIFDriver
inject_partition = -2
live_migration_uri = qemu+ssh://stack@%s/system
live_migration_flag=VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,VIR_MIGRATE_LIVE
use_usb_tablet = False
cpu_mode = none
virt_type = qemu

libvirtd.conf

listen_tls=0
listen_tcp=1
tcp_auth="none"

libvirtd is listening on 16509

$ sudo netstat -anp | grep 165
tcp        0      0 0.0.0.0:16509           0.0.0.0:*               LISTEN      18946/libvirtd
tcp6       0      0 :::16509                :::*                    LISTEN      18946/libvirtd

Errors under nova log.

    2015-09-14 18:46:18.378 ERROR nova.compute.manager [req-d35efe82-3fd6-4c84-953c-256248a34f5b admin admin] [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc713
    9e5] Pre live migration failed at nova1
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5] Traceback (most recent call last):
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]   File "/opt/stack/nova/nova/compute/manager.py",
    line 5316, in live_migration
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]     block_migration, disk, dest, migrate_data)
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]   File "/opt/stack/nova/nova/compute/rpcapi.py", l
    ine 623, in pre_live_migration
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]     disk=disk, migrate_data=migrate_data)
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]   File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/rpc/client.py", line 156, in call
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]     retry=self.retry)
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]   File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/transport.py", line 90, in _send
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]     timeout=timeout, retry=retry)
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]   File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/_drivers/amqpdriver.py", line 350, in send
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]     retry=retry)
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]   File "/usr/local/lib/python2.7/dist-packages/oslo_messaging/_drivers/amqpdriver.py", line 341, in _send
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5]     raise result
    2015-09-14 18:46:18.378 TRACE nova.compute.manager [instance: 2ae2fc70-eb23-45f0-8fcb-d65bdc7139e5] NoLiveMigrationForConfigDriveInLibVirt_Remote: Live migration of instances with config drives is not supported in libvirt unless libvirt instance path and drive data is shared across compute nodes.

Any help?

Thanks, Hardik

edit retag flag offensive close merge delete

2 answers

Sort by ยป oldest newest most voted
1

answered 2015-09-16 03:19:21 -0500

updated 2015-09-16 03:21:05 -0500

Hi,

the error message is clear "Live migration of instances with config drives is not supported in libvirt unless libvirt instance path and drive data is shared across compute nodes."

What kind of shared storage do you use ? NFS or Ceph ? If you don't use a shared storage you cannot use live migration.

Otherwise, if you have a shared storage you can check if you have setting force_config_drive different from None in nova.conf.

What version of openstack do you use ?

Salvo.

edit flag offensive delete link more

Comments

Hi Salvo,

Thanks for your reply. I do not have any shared storage and that is the reason I am trying block migration. My understating is that you do not need shared storage if you do block migration. is that correct?

I am using stable/kilo openstack version.

hardik italia gravatar imagehardik italia ( 2015-09-16 13:46:39 -0500 )edit

Have you enable libvirtd_opts="-d -l" in /etc/default/libvirt-bin ?

salvorapi gravatar imagesalvorapi ( 2015-09-17 05:45:51 -0500 )edit

yes it is already enabled. I tired with following "live migration url" in the nova.conf but no success in any case.

live_migration_uri = qemu+ssh://stack@%s/system 
live_migration_uri = qemu+tcp://%s/system
hardik italia gravatar imagehardik italia ( 2015-09-17 14:46:49 -0500 )edit
0

answered 2015-12-04 15:45:00 -0500

hardik italia gravatar image

updated 2015-12-04 15:46:06 -0500

Resolved issued by setting up following in the nova.conf with the latest release of openstack.

force_config_drive = False
config_drive_format = vfat

Reference:
http://docs.openstack.org/user-guide/cli_config_drive.html
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: 2015-09-15 18:43:14 -0500

Seen: 2,662 times

Last updated: Dec 04 '15