[CentOS] Adding RAM

Ross Walker rswwalker at gmail.com
Mon Dec 8 23:04:53 UTC 2008


On Dec 8, 2008, at 4:55 PM, Matt <lm7812 at 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



More information about the CentOS mailing list