We're back up.
Hopefully there won't be any further problem.

Our attempt to upgrade the memory on Oscar (the BOINC database machine) and Carolyn (the replica database) took a bit of DIMM swapping to find pairs of DIMMs that worked. Carolyn is still short two DIMMs (32 GB), but we didn't want to extend the outage any further to try to get it up to the full 192 GB.

We might try to max out Carolyn during the outage next week.

[edit]And as is usual after a long outage, we've got a large number of hosts out there starving for work. It'll take a little while before "no work" messages start to get less frequent.[/edit]

More...