Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util sda 0.38 176.63 70.32 78.26 813.46 2044.82 406.73 1022.41 19.24 0.40 19.17 4.04 60.07 sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 5.28 0.00 23.61 19.33 0.00 sda2 0.38 176.63 70.32 78.26 813.45 2044.82 406.73 1022.41 19.24 0.40 19.17 4.04 60.07 dm-0 0.00 0.00 70.71 255.60 813.45 2044.82 406.73 1022.41 8.76 2.90 8.87 1.84 60.10 dm-1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 8.00 0.00 64.20 11.38 0.00
Try setting the scheduler to 'deadline' and see if the queue sizes shrink.
I have googled this and having a bit of trouble figuring how to change it under CentOS 4.
http://www.wlug.org.nz/LinuxIoScheduler
Does not seem to work on CentOS 4.
No raid1? Besides adding redundancy, it can help with read performance. I would probably put the mail on a raid 10 though if I had 4 disks to do so.
I plan on moving to faster disks and RAID 1 down the road. Just has not happened yet.
Matt