Disk top usage PIDs

[Eduardo Meyer]

On Tue, Nov 4, 2008 at 12:55 PM, Bill Moran <wmoran at potentialtech.com> wrote:
> In response to "Eduardo Meyer" <dudu.meyer at gmail.com>:
>
>> On Tue, Nov 4, 2008 at 12:27 PM, Eduardo Meyer <dudu.meyer at gmail.com> wrote:
>> > On Tue, Nov 4, 2008 at 12:18 PM, Bill Moran <wmoran at potentialtech.com> wrote:
>> >> In response to "Eduardo Meyer" <dudu.meyer at gmail.com>:
>> >>>
>> >>> I have some serious issue. Sometimes something happens and my disk
>> >>> usage performance find its limit quickly. I follow with gstat and
>> >>> iostat -xw1, and everything usually happens just fine, with %b around
>> >>> 20 and 0 to 1 pending i/o request. Suddely I get 30, 40 pending
>> >>> requests and %b is always on 100% (or more than this).
>> >>>
>> >>> fstat and lsof gives me no hint, because the type of programs as well
>> >>> as the amount of 'em is just the same.
>> >>>
>> >>> How can I find the PID which is hammering my disk? Is there an "iotop"
>> >>> or "disktop" tool or something alike?
>> >>
>> >> top -m io -o total
>> >
>> > Great, thats exactly what I was looking for, thank you a lot Mr Moran.
>>
>> I see syncer (40%) and bufaemon (10%) and after that, imapd. The first
>> ones are kernel PIDs (36 and 37).
>>
>>     PID USERNAME   VCSW  IVCSW   READ  WRITE  FAULT  TOTAL PERCENT COMMAND
>>      36 root          2      2      0     31      0     31  40.79% bufdaemon
>>      37 root          2      2      0     16      0     16  21.05% syncer
>> 71501 vmail         4      0      0      0      0      0   12.00% imapd
>>
>> I guess it a symptom of some hardware problems, kernel itself is not
>> supposed to do this many I/O, right?
>>
>> Sometimes PID 39, softdepflush, is always on top 3.
>
> Off the top of my head, it looks like you're exceeding what the hardware
> can do.  What kind of disks do you have in that system?  It may be time
> to get faster disks or expand to a high-performance RAID setup.

Its a:

Master:  ad4 <Hitachi HDP725050GLA360/GM4OA52A> Serial ATA II

I still think its a hw (disk or controller) problem.

I have raised the following:

kern.filedelay=90
kern.dirdelay=90
kern.metadelay=90

Now, my disk usage is back to 30%




>
> --
> Bill Moran
> http://www.potentialtech.com
>



-- 
===========
Eduardo Meyer
pessoal: dudu.meyer at gmail.com
profissional: ddm.farmaciap at saude.gov.br