Proper handling of suspended VMs in migration
Bug #1791682 reported by
Tobias Rydberg
This bug affects 3 people
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OpenStack Public Cloud WG |
Confirmed
|
Undecided
|
Tobias Rydberg | ||
Bug Description
It is currently impossible to dead-migrate a suspended guest. If a compute node needs to be evacuated while there are suspended guests on it, the only way to do that is to resume the guest, live-migrate, and then suspend the guest again. That might be entirely contrary to what the user wants — they might want that guest to be down for a reason.
It would be great if Nova live-migration could detect that a guest is currently suspended, and rather than resume the VM, just copy over the domain configuration XML and the suspend file.
| Changed in openstack-publiccloud-wg: | |
| status: | New → Confirmed |
Revision history for this message
| Matt Riedemann (mriedem) wrote | #1 |
To post a comment you must log in.
This would arguably require a microversion since it would be an API change to allow live migrating suspended VMs:
https:/
And we'd have an upgrade impact to that as well since the API would need to check with the compute to see if it can handle live migrating a suspended VM, because you could have old computes that don't have the code, or the instance is running on a virt driver that doesn't support this.