Heat stack stop or reboot
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OpenStack Public Cloud WG |
Confirmed
|
Undecided
|
Tobias Rydberg | ||
Bug Description
Currently, all resources in a Heat stack can be simultaneously suspended (openstack stack suspend), snapshotted (openstack stack snapshot), rolled back to a previous snapshot (openstack stack rollback), checked (openstack stack check), and deleted (openstack stack delete).
What we cannot do is actually shut down and start up an existing stack, something like openstack stack stop/start, in the same manner as openstack server stop/start. Stopping a stack (from my point of view) would mean stopping all of its VMs, setting all its Neutron resources' admin state to DOWN, and detaching all volumes. Having this would be exceptionally useful in all sorts of ways, not least doing a full reboot of a stack. Is this something that could be discussed, or is perhaps someone already working on this?
| Changed in openstack-publiccloud-wg: | |
| status: | New → Confirmed |
| Zane Bitter (zaneb) wrote | #1 |
This is something that could certainly be discussed. The actions API in Heat is set up in such a way that it should be comparatively easy to add new actions. However, there is a fair amount of implementation involved, since each resource type that needs to do something on stop/start would need to have the correct actions implemented.
However, there are some open issues with suspend/resume:
https:/
https:/
I'd be very reluctant to add any new actions before those are resolved.
The addition of extra states also contributes to disproportionately increased complexity (as illustrated by that second issue above), so it would require a compelling use case. A comprehensive spec where somebody has worked out in advance how each potential state transition should be treated (e.g. what should happen if you stop a suspended stack, &c.) would be the most helpful thing to make it more likely to be accepted.