Sorry for the outage, everyone. We had a catastrophic SCSI card failure in our core database server. We're back up for the moment, although we'll need to do some more maintenance to the server to be confident about having fixed the issue. Actually going on-site to the colo was required, and it took us a little while to do that.

Ironically, the database maintenance we were planning to do on Friday would have been really useful, as is often the case with murphy's law. As of the release tomorrow night, we should have a hot database mirror running again (we always had one before, but some recent changes made that more problematic). Thus, if Stuff Happens again, we should be able to bring the game back more quickly.

whoops. I was wondering what kind of SCSI card that is? I am wondering why there is still such a thing in use? :)
Hope you can work it out.

M. Duncan

It's an Adaptec. And after some further research, I think it may actually be a motherboard issue. We're trying to work around it in software.

As for why it's still in use.. what should we be using? Serial attached SCSI? Fiber channel? Our database server doesn't require a lot of space, just fast I/O, so we use SCSI with Cheetah drives. If I used SATA Raptors, the transfer times would probably be similar, but I think the I/O overhead would still be a bit higher (NCQ not being quite.. right, as of last I read).

does this mean a delay of guild banking features?

SCSI is the way to go. May want to consider 2x raid 5 array's in a mirror arrangement.

Ed

epadafunk: no, it should have no impact on that.

black: we would do 0+1 if we needed the space, but it's more of an I/O speed thing than space intensive. We're pretty write-heavy, so raid5 isn't necessarily the best, unless the we required the space.