On Thu, Jun 16, 2011 at 1:27 PM, Alexander Farber
<alexander.farber@gmail.com> wrote:
> For example pg_top shows (why is everything idle?):
>
> last pid: 5215; load avg: 0.65, 1.64, 2.13; up 0+00:46:48
>
> 20:16:37
> 22 processes: 22 sleeping
> CPU states: 12.4% user, 0.0% nice, 0.3% system, 87.4% idle, 0.1% iowait
> Memory: 1187M used, 2737M free, 34M buffers, 611M cached
> Swap: 7812M free
>
> PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND
> 3303 postgres 16 0 1170M 137M sleep 3:29 10.92% 20.16%
> postgres: pref pref [local] idle
> 5045 postgres 18 0 1169M 168M sleep 0:49 0.00% 0.00%
> postgres: pref pref [local] idle
> 5057 postgres 16 0 1169M 168M sleep 0:37 0.00% 0.00%
> postgres: pref pref [local] idle
> 5000 postgres 16 0 1169M 168M sleep 0:34 0.00% 0.00%
> postgres: pref pref [local] idle
> 5025 postgres 18 0 1169M 150M sleep 0:31 0.00% 0.00%
> postgres: pref pref [local] idle
> 5030 postgres 16 0 1171M 152M sleep 0:29 0.00% 0.00%
What do vmstat 1 and iostat -xd 1 (or equivalent for your OS) say? So
here's a healthy not working too hard machine:
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
4 0 0 552768 480820 121441792 0 0 1 24 0 0 8 1 91 0
5 0 0 556168 480820 121441664 0 0 0 468 4005 4763 11 1 88 0
2 0 0 561660 480820 121441680 0 0 0 164 4032 4555 10 1 88 0
1 0 0 555076 480820 121441680 0 0 12 300 4171 4710 5 0 94 0
Note that the in and cs values are pretty reasonable (interrupts and
context switches / sec) and idle CPU is ok. 0 wait.
If those cs and in start climbing and the wait starts climbing your IO
is getting saturated.
In iostat keep an eye on %util