On Jan 12, 2004, at 1:11 PM, Brion Vibber wrote:
We really, really need to move the database back to a machine with a decent hard drive. The wikis are very sluggish, and a fair chunk of it's from waiting on the database.
Ursula's sitting around with a 90% idle CPU, but everything's blocked on disk I/O to the point it's got a load average of about 16. At any given time from 8-20 processes are blocked and waiting. Operations that hit a lot of rows like history and watchlist are particularly badly hit since they don't play as well with caching.
If Geoffrin's not going to be up soon, and Pliny's still emitting spurious disk errors, what are our options out of the available machines?
-- brion vibber (brion @ pobox.com) _______________________________________________ Wikitech-l mailing list Wikitech-l@Wikipedia.org http://mail.wikipedia.org/mailman/listinfo/wikitech-l
It seems to me that one of the biggest problems we seem to have is reliable and speedy hard drives. Perhaps it might be wise to consider possible purchasing an external disk subsystem? Something like the Apple Xserve RAID systems are speedy (lots of internal hardware RAID), reliable (easy to swap out disks), and expandable (up to 3.5TB, 1TB in the cheapest configuration). If the database server dies, it should be fairly easy to plug the external disk system into another machine.
http://www.apple.com/xserve/raid/
Might be something worth considering...
-- Nick Reinking -- eschewing obfuscation since 1981 -- Minneapolis, MN