Config changes do not cause a container restart when using COPY_ALWAYS
Bug #1500858 reported by
Paul Bourke
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
kolla |
Invalid
|
Critical
|
Steven Dake |
Bug Description
When updating an augment file or template in Ansible, one would expect running containers to be restarted on deploy, when using the COPY_ALWAYS config strategy. Right now this doesn't seem to be the case - you have to find a way of manually restarting containers across hosts.
Is this intended behavior to lay the path for some form of upgrade strategy? If so it should probably be documented, if not, we need to update the tasks to restart containers on config change (when strategy == COPY_ALWAYS)
Changed in kolla: | |
status: | New → Triaged |
milestone: | none → liberty-rc2 |
importance: | Undecided → Wishlist |
Changed in kolla: | |
milestone: | liberty-rc2 → liberty-rc3 |
Changed in kolla: | |
milestone: | liberty-rc3 → mitaka-1 |
Changed in kolla: | |
milestone: | mitaka-1 → mitaka-2 |
Changed in kolla: | |
milestone: | mitaka-2 → mitaka-3 |
To post a comment you must log in.
The intended behavior for COPY_ALWAYS is to pull in a (possibly new) config everytime the container restarts. This does not have anything to do with a requirement to restart the container when configs have changed.
As initially discussed in the spec, COPY_ALWAYS was more intended for testing environments where you make a manual change on one node and restart the container for it to take effect.
As a whole, Kolla has not agreed on what the best way to pull in new configs should be. For example, we could trigger a handler based on a config change in the regular playbooks, or we could write a new playbook specific for just updating configs. Many options. Whatever is decided the same thing would happen for both COPY_ONCE and COPY_ALWAYS methods tasks and what not.