On Dec 8, 2008, at 4:55 PM, Matt lm7812@gmail.com wrote:
Cpu0 : 5.3% us, 4.0% sy, 0.0% ni, 81.5% id, 9.3% wa, 0.0% hi, 0.0% si Cpu1 : 11.3% us, 8.3% sy, 0.0% ni, 1.7% id, 78.1% wa, 0.7% hi, 0.0% si Mem: 8309188k total, 4761352k used, 3547836k free, 451464k buffers Swap: 2031608k total, 192k used, 2031416k free, 1564316k cached
If this is typical, you don't need more RAM... cpu1 is waiting a lot, I/O bottleneck?
A bit of bottle neck.
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.
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.
-Ross