[Openstack-operators] Live-migrations stuck in MIGRATING state

William Josefsson william.josefson at gmail.com
Wed Oct 26 12:16:54 UTC 2016


Hi,

Some of my VMs fail when I issue, nova live-migration while others
work. The pattern seem to be common to instances that are running more
intensive workloads such as database MQ etc.  if I try with an idle
small instance I don't see these issues.

The VM get's stuck in MIGRATING state, and I cannot do anything,

When I issue virsh list on the source and designation host, I can in
fact see that the VM is already running on the destination, but the DB
in nova show <ID> still shows the old host, and the state is stuck in
MIGRATING.

I also notice this in my logs as if nova database and hypervisors got
out of sync about what is running where,

2016-10-26 19:34:42.947 1438627 WARNING nova.compute.manager
[req-c321c991-7c58-4b78-924e-773678d27b1a - - - - -] While
synchronizing instance power states, found 2 instances in the database
and 3 instances on the hypervisor.


pls advice on how to get this resolved. is there a safer way of
migrating the instances, should I try to pause the cpu/memory
intensive instances, live-migrate and than un-pause them when
completed on the destination? thx will



More information about the OpenStack-operators mailing list