Shelve on stop
From the public cloud SIG:
https:/
Operators want stop (and suspend?) operations to actually offload the guest and then unshelve on start (resume). The problem with shelve is users don't know what it is, nor care to use it, but operators definitely want users to use it.
Suggest adding a compute API microversion to the stop/suspend APIs for an offload behavior which has values:
* auto
* shelve
* retain (?)
The API could default to auto which would check a config option per-deployment and that would default to retain behavior (what we have today - stop/suspend on the same host), but could be configured to shelve. That way deployments can configure their desired default behavior, but users could still have an override option if they really care.
Blueprint information
- Status:
- Not started
- Approver:
- None
- Priority:
- Undefined
- Drafter:
- Matt Riedemann
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- Discussion
- Series goal:
- None
- Implementation:
-
Unknown
- Milestone target:
- None
- Started by
- Completed by
Related branches
Related bugs
| Bug #1791679: Proper handling of suspended/stopped VMs in scheduling | Won't Fix |
| Bug #1791681: Configurable rescheduling when waking from suspend, or starting from shutdown | Won't Fix |
