From: Mihai Donțu on
Hi,

I started the postgresql daemon on my laptop, gave it a big query to chew and
then ran 'top' to see it struggle, when I noticed that several of my processes
had crazy utime-s. I selected the postgres one for exemplification:

$ cat /proc/5623/stat
5623 (postgres) S 1 5623 5623 0 -1 4202560 4340 1644345 5 92 39 1844674407336
92886 1844674402728 20 0 1 0 25983879 100798464 1302 18446744073709551615
4194304 8513148 0 0 0 0 0 19935232 84487 18446744073709551615 0 0 17 1 0 0 0 0

0

$ ps -p 5623 x
postgres 5623 39289650 0.1 98436 5208 ? Ss 02:31 21114581:29
/usr/lib64/postgresql-8.4/bin/postgres --silent-mode=true

$ ps aux | grep "21114581:" | wc -l
152

$ uptime
03:40:07 up 3 days, 1:18, 9 users, load average: 0.75, 0.74, 1.04

$ uname -a
Linux mdontu-dell 2.6.33.1 #4 SMP PREEMPT Mon Mar 22 04:06:35 EET 2010 x86_64
Intel(R) Core(TM)2 CPU T5500 @ 1.66GHz GenuineIntel GNU/Linux

(vanilla kernel + tuxonice patches)

The system time looks accurate though and, sadly, I have no idea how to
reproduce this. I attached a compressed dmesg dump because it was pretty big.

Thanks,

--
Mihai Donțu