[csw-buildfarm] High load with unknown source:
Török Edwin
edwin at clamav.net
Mon Feb 1 21:15:16 CET 2010
Hi,
One of my unit tests has timed out, it is taking 7-20s, while it
normally takes ~0.5s under Linux/x86-64:
b9s:
real 21.0
user 2.2
sys 2.4
Linux:
real 0m0.537s
user 0m0.228s
sys 0m0.320s
I noticed that there is a high load on the Solaris buildhosts
(b8s,b9s,b10s):
9:06pm up 13 day(s), 9:35, 5 users, load average: 14.52, 14.97, 15.93
Is this normal? Is some maintainance task running on the buildfarms
right now?
If so I can disable my automated builds during the maintainance, just
let me know what time intervals I should exclude.
I don't think this is my fault, all my processes are in sleep state:
PID USERNAME LWP PRI NICE SIZE RES STATE TIME CPU COMMAND
26726 edwin 1 59 0 4336K 2976K cpu/19 0:00 0.10% top
23909 edwin 1 59 0 7464K 4152K sleep 0:00 0.02% sshd
6818 edwin 1 59 0 6424K 936K sleep 2:40 0.00% ssh
12480 edwin 2 44 5 16M 2408K sleep 0:00 0.00% buildbot
23920 edwin 1 59 0 5000K 2832K sleep 0:00 0.00% bash
The system load is 30+ right now, appears to be some heavy disk activity:
load averages: 36.2, 22.7, 18.7; up 13+09:39:01
21:10:34
57 processes: 56 sleeping, 1 on cpu
CPU states: 0.0% idle, 74.1% user, 25.9% kernel, 0.0% iowait, 0.0% swap
Kernel: 6579 ctxsw, 8652 trap, 4935 intr, 20870 syscall, 12 fork, 1370 flt
Memory: 12G phys mem, 384M free mem, 8000M total swap, 8000M free swap
PID USERNAME LWP PRI NICE SIZE RES STATE TIME CPU COMMAND
24143 edwin 1 49 0 4272K 2256K cpu/11 0:00 0.03% top
23830 rocky 1 59 0 35M 16M sleep 22:03 0.01%
emacs-athena-23
23810 rocky 1 59 0 7640K 1544K sleep 8:50 0.01% sshd
10139 root 1 59 0 12M 4736K sleep 4:01 0.00% munin-node
23909 edwin 1 59 0 7464K 4616K sleep 0:00 0.00% sshd
10403 root 1 59 0 9184K 1128K sleep 0:56 0.00% httpd
Best regards,
--Edwin
More information about the buildfarm
mailing list