On 11/11/10 20:20, Коньков Евгений wrote:
Hi, all
"How to obtain what swi1:net is doing?"
The short answer is: depending on what your network card is, it could be
everything related to TCP/IP-level processing.
In your case, you are doing a lot of work in netgraph and dummynet,
probably shaping, but have high dummynet usage which probably means its
handling the lower level of network IO, probably with a high packet rate.
You might try including the following loader.conf tunables:
net.isr.direct_force=0
net.isr.maxthreads=2
... and report if it helps you. (but be careful: here you must measure
real-world performance not CPU usage!)
last pid: 65736; load averages: 3.54, 4.46, 3.92 up 4+07:51:26 21:19:08
215 processes: 8 running, 195 sleeping, 12 waiting
CPU 0: 2.9% user, 0.0% nice, 42.9% system, 11.4% interrupt, 42.9% idle
CPU 1: 0.0% user, 0.0% nice, 54.3% system, 17.1% interrupt, 28.6% idle
CPU 2: 2.9% user, 0.0% nice, 57.1% system, 5.7% interrupt, 34.3% idle
CPU 3: 0.0% user, 0.0% nice, 45.7% system, 17.1% interrupt, 37.1% idle
Mem: 502M Active, 87M Inact, 324M Wired, 24M Cache, 112M Buf, 1053M Free
Swap: 20G Total, 72K Used, 20G Free
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND
15 root -44 - 0K 8K CPU1 3 31.4H 71.39% swi1: net
35 root -68 - 0K 8K CPU0 0 21.7H 50.20% dummynet
14 root 171 ki31 0K 8K RUN 0 74.8H 42.87% idle: cpu0
11 root 171 ki31 0K 8K CPU3 3 78.4H 31.79% idle: cpu3
13 root 171 ki31 0K 8K RUN 1 80.3H 29.69% idle: cpu1
12 root 171 ki31 0K 8K RUN 2 76.9H 23.49% idle: cpu2
1698 root -68 - 0K 8K sleep 2 312:30 15.38% ng_queue0
1700 root -68 - 0K 8K sleep 3 313:02 15.09% ng_queue2
1699 root -68 - 0K 8K sleep 2 314:18 14.89% ng_queue1
1701 root -68 - 0K 8K sleep 1 312:54 14.06% ng_queue3
63829 www 50 0 185M 123M select 0 0:47 4.05% httpd
59213 root 96 0 400M 61940K CPU3 2 32:24 2.98% rtorrent
16 root -32 - 0K 8K WAIT 2 129:41 0.39% swi4: clock sio
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[email protected]"
