stopped proxy can be inadvertently started by config change

[Edward Hope-Morley]

During a ring sync across proxy units we stop proxy services until rings are synced to ensure consistency for api calls. We are not, however, disabling the service (we just stop it) which means that it could be restarted by an config changes made during a sync that would result in a restart_on_change(). We should permanently stop and disable the service until we are ready to start it.

If we don' do this we easily run into a situation where the charm does a start() following a sync and this has no effect because the service is already started but has failed due to rings not yet existing e.g. https://pastebin.canonical.com/200596/

I believe that this could actually be critical since it could result in the swift-proxy service never being restarted following sync.

[David O Neill (dmzoneill)]

Please consider:

https://review.openstack.org/#/c/636622/

[Edward Hope-Morley (hopem)]

@dmzoneill please ensure that the LP associated with a patch has its status correctly updated. Normally this is taken care of by gerrit but since in this case the format was nor correct on first submit the update didn't happen. I'll update this one but please remember for future patches.

[OpenStack Infra (hudson-openstack)]

Change abandoned by James Page (<email address hidden>) on branch: master
Review: https://review.opendev.org/636622
Reason: No response in 4 months to review feedback as requested - closing review.

Please re-open and address feedback if still required.

[Edward Hope-Morley (hopem)]

Assuming this is no longer being worked on since the patch was abandoned several months ago. I'll go ahead and remove the current assignee to allow someone else to take it on.