[Optional] Update Instance/Task status if instance stucks at BUILD status to ERROR
Taskmanager should perform complete rollback if instance stucks at BUILD status. It means that each component of instance should be deleted and quota should be updated.
Workflow is next:
1. If instance stacks at BUILDING status after poll_untill fells to timeout set TaskStatus to Error, set InstanceStatus to Error.
2. Update rows in DB.
All clean-up should be delegated to instance forceDelete API call
Blueprint information
Related branches
Related bugs
Sprints
Whiteboard
[Illia Khudoshyn] I would rather have an ability to rollback stuff manually too.
[Haomai Wang ] Why not give the user force delete option?
I think the workflow you list is actually doing deleting instance.
[Denis M.] - My point is to perform complete rollback and report user that creation failed. I think allowing user deleting stucked instances is not so good idea. I'm suggesting to do rollback aumaticaly, without user. Or we could do it optionaly
Gerrit topic: https:/
Addressed by: https:/
Optional rollback
Gerrit topic: https:/
Addressed by: https:/
Instance rollback
Gerrit topic: https:/
Addressed by: https:/
Update statuses on GA timeout
Addressed by: https:/
Update statuses on GA timeout
Work Items
Dependency tree
* Blueprints in grey have been implemented.
