I hadn't been closely monitoring the backups from SQL dumps, and the root filesystem ran out of disk space (see below). Once that happened, mailman, mysql and confluence had nowhere to write temp files or storage to, so they all crapped out. The nightly SQL dumps use a HUGE amount of disk space, so my plan to cut this crap down in the future is to write some scripts which instead of dumping SQL every night, do a 'diff' on the dump from $tonight against $last_night, save the diff, and then discard the full dump. Each month there will be a full dump saved, but no "full incrementals'. Other suggestions are welcome.... ============= Filesystem 1K-blocks Used Available Use% Mounted on /dev/sda1 239778532 228145200 0 100% / tmpfs 2027020 0 2027020 0% /lib/init/rw udev 10240 56 10184 1% /dev tmpfs 2027020 0 2027020 0% /dev/shm /dev/sdb1 307663800 168479668 123555700 58% /mnt