Open chrisys opened 3 years ago
thank you very much @chrisys for providing all the context here and not making us dig it up in recordings/threads.
I'm cool with changing the config simply on the basis that the Supervisor is not the one controlling the persistent logging it's the host as the new variable allows you to correctly infer. I got lost on the why the reboot maters but believe it's just further reason to use host since it affects the host and not just the Supervisor.
What further discussion do you think there should be ? I think it's a clear cut case of just rename it tbh.
As per arch call (software 2) on 6th May, consider if the
BALENA_SUPERVISOR_PERSISTENT_LOGGING
config var should be renamedBALENA_HOST_PERSISTENT_LOGGING
. This likely warrants a discussion on it's own to make sure this rename is as correct as we can tell to make sure we avoid another similar situation in the future.The thinking behind this proposal is that currently persistent logging is the only config var within the supervisor namespace that causes a reboot, and the action it carries out is related to mount points on the host OS rather than supervisor functionality.