pre-backup-lock/post-backup-unlock scripts are not idempotent

[x6j8x]

Issue

pre-backup-lock/post-backup-unlock scripts are not idempotent.

Context

If for whatever reason you have to run the scripts again if the instance is already locked/unlocked they will fail.

Steps to Reproduce

Run bbr backup and then bbr backup-cleanup - the latter will fail due to the api instance being already unlocked.

Error attempting to run post-backup-unlock for job cloud_controller_ng on api/...

Expected result

Running pre-backup-lock or post-backup-unlock should always succeed even if the respective state (instance being in maintenance mode or not) is already active.

Current result

It fails if the instance is already in the desired state.

Possible Fix

Use the already existing /var/vcap/data/cloud_controller_ng/tmp/proxy_temp/bbr_maintenance "semaphore" file as indicator whether or not the instance is in the desired state is already (or employ a more sophisticated algorithm checking for the existence of the nginx maintenance process).

[cf-gitbot]

We have created an issue in Pivotal Tracker to manage this:

https://www.pivotaltracker.com/story/show/165556252

The labels on this github issue will be updated when the story is started.

[tcdowney]

Thanks for reporting this @x6j8x!

The following two commits should fix this issue and will be available in the next capi-release:

• 49d6977
• 123bb4c

Basically it just checks with bpm list to see if the nginx_maintenance job is already in the stopped state. Not super sophisticated, but gets the job done since bpm already knows how to track the lifecycle of the process.