PLEASE NOTE

Anchor rebooted again last night at ~5:30pm, so preventative maintenance has been scheduled "on short notice" in an attempt to reduce the frequency of these crashes.

Anchor will be shut down for ~30 minutes today, from 12:30 to 1:00 (ie, during lunchtime, hopefully somewhat less disruptive a time). During this time, memory modules will be removed, which have been throwing errors, and which may be the ultimate source of the kernel panic / reboot events in the past few days.

Updates will be sent via POLAN mailing list once work is complete, as well as updates to this web page (of course!) :-)

UPDATE: POST MAINTENANCE:

Anchor is back online and operating "normally" following maintenance over lunch, which resulted in downtime from ~12:30 through to ~1:10.

I've removed a "suspected bad" memory module from the server, and this has resulted in reduced system RAM on anchor (a drop from 8gigs to 6 gigs). In theory operations will be normal in this configuration, although (clearly) applications which demand LARGE amounts of ram may not work properly (such as Matlab analysis of very large data sets?)

I intend to leave anchor "up and running normally" to see how things proceed from here.

If we remain crash-free for 24 hours (72 hours ideally) then I'm somewhat optimistic this will be an adequate workaround / temporary resolution.

IF crashes persist @ ~24hour regularity, then further measures will need consideration.

Updates will be sent as the situation evolves.

UPDATE: NEXT DAY - THURSDAY OCT.11th

Anchor has been up and running smoothly for ~24hours now. This is an improvement over the 48hours prior to the maintenance work. Hopefully this is a good sign. We'll see.