Description

Up for grabs for anyone who'd like to contribute, no rush :)

When reviewing a PR that introduced widespread mail delivery failures by changing UID/GID (as Dovecot no longer had permission to write the mail to /var/mail), I suggested a test-case that would better catch that common failure (technically two, as checking the mail logs for failures would be complimentary at catching another set of errors).

It would be nice to eventually do some cleanup across tests to have a common helper method for consistently checking successful mail delivery.

• At least when only one mail is delivered in a test this approach works well (using a timeout with tail to avoid blocking on mail queue emptying, fails faster in a more helpful way)
• An alternative technique is to filter and group by account (ideally waiting for mail queue to empty first)

Failures were all shared here, which should help track down which tests need the refactoring. I also identified some common keywords in the failure logs that could be checked. An earlier comment provided more verbosity in logs (for additional context)