Comment 4 for bug 1811147

There are multiple issues at hand here:

1. Any Juju agent can cause the controller logs collections to grow unbounded and run the controller(s) out of disk space.

2. A Juju agent spews logs when it encounters an ENOSPC situation.

3. Juju worker failures result in a crash loop with no backoff - this exacerbates (2), which results in a out of disk space instance causing (1).

I've retitled this bug to specifically target the unbounded growth of the logs collections on the controllers (which is presumably due to the log pruner failing or failing to keep up with the rate of growth). This one is also the most critical from an operational perspective.

I've retitled https://bugs.launchpad.net/juju/+bug/1827664 to target the fact that Juju agents spew logs on ENOSPC. A separate bug probably should be raised for (3).