The drives are part of shared nothing filesystem (IBM GPFS FPO). We have 55 drives in all and only 2 of them have failed. Many more would need to fail before I became concerned about a loss of data.

As for why we cannot perform a hot swap, I have been inquiring about that. The servers support it, but when I opened the ticket to get the failed disks replaced I was told that it wasn't possible. So I am trying to get an answer as to what is going on, but in the meantime we are getting the disks replaced while we do the database change since we would be down during that time anyway.

will the forum stay up during the maintenance? I'm assuming not since it says the website will be down

This work has been completed and we are catching up on the backlog.

As for BobbyB's question. the peak load once we started up hit our configured limits for about 7-8 minutes (i.e. we hit the limit of 300 concurrent uploads) and during that time we were receiving about 1.1 Gbps of data during that time. The load is now rapidly dropping back to normal levels and users should see no issues uploading or downloading work*


* The one exception is that we are currently processing all of the results returned which means that we are going to generate 6 hours worth of resends in 20-40 minutes. As a result over the 30-40 mintues there will be times where users have trouble getting assigned work during a scheduler request because we might be clogged with work that needs a reliable host on a given platform. This will clear quickly and automatically and we should be fully back to normal soon.

It came back up about 16:45 UTC. So you uploaded about 4 minutes after it became available again.