Sorry for the recent downtime.
My host says the explanation for “Saturday morning’s downtime was caused by the hardware failure of a not-as-redundant-as-claimed power supply. Monday morning’s downtime was caused by a software error triggered by the rebuild process that occurred after the system came back online. On Saturday morning, we fixed the hardware problem, and now we are addressing the software problem.”
Things are stable now, and they’ll be moving the disk cluster from the existing hardware onto new hardware in the near future, hopefully increasing reliability.