I'm not sure this really deals with them, but they are generally indicative of a non-recoverable error, so probably better to exit than to keep trying.

Granted, both of these have Restart=on-failure in their systemd service config, so it ends up being a heavier-handed form of retry. In the case where a session has been expired, this may well resolve the issue. For other types of AuthError faults (disabled user, access issues, bad client code), a full restart will probably not help.

Longer term, we should probably try to have a cleaner reauth mechanism for expired sessions, but this simple change is fine for now.

Commit 3cf8cc50 fixes this pull-request