Playbook to ensure service bounced after config change
To ensure idempotent state of deployment, service need to be bounced after configuration change.
This part is still missing in current playbook.
Blueprint information
- Status:
- Complete
- Approver:
- None
- Priority:
- Medium
- Drafter:
- Qiu Yu
- Direction:
- Needs approval
- Assignee:
- Qiu Yu
- Definition:
- Superseded
- Series goal:
- Accepted for mitaka
- Implementation:
- Started
- Milestone target:
- None
- Started by
- Steven Dake
- Completed by
- Sam Yaple
Related branches
Related bugs
Sprints
Whiteboard
The direction of this needs to be discussed and approved. We have two separate cases to consider, COPY_ONCE and COPY_ALWAYS. Additionally, using the task to register if there is a change is _not_ acceptable. There are several instances where the config file won't show updated (so no restart of container) and will show updated even though the running config doesn't differ. Then we have the matter of nohup reloads versus full restart of service. All of these issues need to be discussed and agreed upon before we start merging code. --SamYaple
Sam - I agree re needing discussion. Our agenda for the midcycle is heavy atm but perhaps we can fit this in on the last day in the parking lot. Would you mind leading this midcycle session if we can move it from the parking lot to the agenda?
Gerrit topic: https:/
Addressed by: https:/
Add `restart_container` action in kolla_docker module
Gerrit topic: https:/
Addressed by: https:/
Bounce cinder service after config change
Addressed by: https:/
Add dryrun support for config action
New blueprint is kolla-reconfig