I truly had (i hope) a hell of a time with tremendous high iotimes on a Redhat enterprise 3 server.
Well, maybe some of you guys have experienced unusual high io times on a RedHat Server with Enterprise 3, well , I had (hopefully) my share of that during these past days.
The server is a Dual Pentium Xeon 2.4Ghz with 1Gb Ram and two SATA Hard Drives of 100GB of space each, no Raid or anything like that, i use it for a Plesk Reloaded 7.53 hosting service, currently hosting over 130 Domains which is not that much really. I started to have terrible problems with server slowness, mostly while performing intensive disk functions, like for example a Plesk upgrade or running the plesk backup utility which currently holds around 9GB of data using psadump. Initially i kind of thought there was a problem with the hard drive, but after reading a lot on google results, i found that literally thousands of people have had similar experiences.
Yesterday (May 20, 2005) , i upgraded the kernell to version kernel-smp-2.4.21-32.EL , downloaded it from redhat network and since then have been doing some intensive testing with big operations like the psadump, using a bash ftp script to transsfer over 10GB of data and the iotimes have normalized, consider that before my iowait times were around 320%, just crazy, and now its around 20% - 53% while doing such operations, server load average also normalized, before my load average was around 10 ! , or more !
Heres a copy of my top header:
13:56:34 up 38 min, 2 users, load average: 0.12, 0.10, 0.15
|