We are running older versions of BalenaOS (Resin OS 2.9.7+rev4) and supervisor (6.6.0), and we usually encounter an issue when the host ends up with a double container when moved across applications.
We can workaround this by removing the existing
/tmp/resin-supervisor/<containerid>/resin-supervisor.lock from the host. However, the change is not immediate, and we have to wait for several minutes for the change to kick in, and then the issue is fixed.
I believe this waiting period is controlled by the variable
RESIN_SUPERVISOR_POLL_INTERVAL, which is 15 minutes by default. My question is: without changing this variable, is there a way we can shorten the waiting length for this change to apply, or is there a way to “force” supervisor to update their containers when we remove the lockfiles? Unfortunately we cannot upgrade OS for our production fleet, so I’m looking for ways to just make the app migration more manageable by us
If you require more context I can gladly help you replicate the issue so you can observe and test
Thanks in advance