co-ordinated periodic live migration
Hi
I am trying to implement periodic live migration b/e multiple compute nodes.
I was thinking I can code the migration in one of the compute node that is the "controller" node as a periodic task.
For instance, I hard-code under compute/manager.py:
instance_uuid = "blah"
host_to_move = "server1"
instance = self.conductor_
self.
The output of nova list from server0 shows that the VM is in server0 and the state is "SHUTOFF". However virsh shows that the instance does not exist in server0 and is "running" in server 1. It looks like the DB is not being updated?
Also, I see that trying to run the live migration in the "controller" compute node when the instance resides in other nodes give me "instance not found" error. Does this mean that this sort of task that requires the "global" view needs to be done outside the compute? I see that there are many live-migration methods also in conductor nodes that I am starting to get confused by what is the exact flow of "live migration". Essentially, I want to live-migrate instance from any compute node to any other compute node, any idea how this can be done - is this via nova-conductor?
Question information
- Language:
- English Edit question
- Status:
- Expired
- Assignee:
- No assignee Edit question
- Last query:
- Last reply: