We have to remember that the BOINC project servers (and admins) are distributed all over the world. They have day/night cycles that do not coincide with ours, but they also have families, and holidays, and weekends, etc.

Recently the SETI@Home server was threatened by a fire. There are many things outside the control of the admin teams, and we can hardly expect them to keep an eye on the server 24/7. If you want to ensure steady work, an easy solution is to increase the backlog stored on your machines. I have at least 2 days worth of work on all my equipment.

As an interesting side note, at the conclusion of the Enigma@Home outage recently I had just over 18 thousand WUs waiting to be returned!