Whoops - an actual outage occurred! That's the first one this year.

Machine ran out of disk due to some large files in /root/.cache/* which were unanticipated.

Fixed by moving that directory to a ZFS volume with bounded quota.