[RESOLVED]: upgrading to r230059 cause slow network throughput
Коньков Евгений
kes-kes at yandex.ru
Sat Jan 14 21:11:03 UTC 2012
Здравствуйте, Коньков.
Вы писали 13 января 2012 г., 21:32:14:
КЕ> I have tryed as ULE as SCHED_4BSD, no changes
КЕ> in both cases low throughput
КЕ> CPUload
КЕ> http://piccy.info/view3/2478224/e5d7f208538d05d813411c34eb493a8f/orig/
КЕ> if load
КЕ> http://piccy.info/view3/2478228/bf8dca5fad12c1436092f4d6aaf2356f/orig/
КЕ> looking on graphs CPU load it seems strange distribution...
КЕ> it seems almost not scheduling ng_queue and netisr
КЕ> last pid: 54347; load averages: 0.30, 0.28, 0.22 up 0+06:24:06 21:25:11
КЕ> 273 processes: 5 running, 237 sleeping, 31 waiting
КЕ> CPU 0: 0.0% user, 0.0% nice, 3.1% system, 0.4% interrupt, 96.5% idle
КЕ> CPU 1: 3.5% user, 0.0% nice, 1.2% system, 0.4% interrupt, 94.9% idle
КЕ> CPU 2: 2.0% user, 0.0% nice, 1.6% system, 0.4% interrupt, 96.1% idle
КЕ> CPU 3: 0.0% user, 0.0% nice, 0.8% system, 3.1% interrupt, 96.1% idle
КЕ> Mem: 310M Active, 1080M Inact, 179M Wired, 112M Buf, 354M Free
КЕ> Swap: 3926M Total, 3926M Free
КЕ> PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND
КЕ> 11 root 155 ki31 0K 32K CPU3 3 359:26 94.38% idle{idle: cpu3}
КЕ> 11 root 155 ki31 0K 32K CPU0 0 356:33 92.77% idle{idle: cpu0}
КЕ> 11 root 155 ki31 0K 32K RUN 1 345:23 89.60% idle{idle: cpu1}
КЕ> 11 root 155 ki31 0K 32K CPU2 2 340:20 85.35% idle{idle: cpu2}
КЕ> 3312 root 40 0 15468K 6488K select 2 15:00 4.25% snmpd
КЕ> 12 root -60 - 0K 248K WAIT 0 7:20 1.07% intr{swi4: clock}
КЕ> 12 root -92 - 0K 248K WAIT 3 12:06 0.29% intr{irq266: re0}
КЕ> 0 root -92 0 0K 152K - 3 6:45 0.05% kernel{dummynet}
КЕ> 13 root -92 - 0K 32K sleep 1 1:39 0.00% ng_queue{ng_queue1}
КЕ> 13 root -92 - 0K 32K sleep 1 1:39 0.00% ng_queue{ng_queue3}
КЕ> 13 root -92 - 0K 32K sleep 3 1:39 0.00% ng_queue{ng_queue0}
КЕ> 13 root -92 - 0K 32K sleep 0 1:39 0.00% ng_queue{ng_queue2}
КЕ> 6880 root 8 0 9592K 1300K nanslp 1 0:46 0.00% monitord
КЕ> 0 root -16 0 0K 152K sched 1 0:43 0.00% kernel{swapper}
КЕ> 95148 root 16 0 9756K 1452K pause 1 0:37 0.00% netstat
КЕ> 12 root -72 - 0K 248K WAIT 3 0:34 0.00% intr{swi1: netisr 3}
КЕ> 15 root -16 - 0K 8K - 3 0:28 0.00% yarrow
КЕ> 1070 root 40 0 10524K 4224K select 1 0:21 0.00% zebra
КЕ> 2020 root 30 -10 50664K 22824K select 3 0:20 0.00% mpd5{mpd5}
КЕ> 7611 firebird 30 -10 106M 65780K usem 2 0:14 0.00% fb_smp_server{fb_smp_server}
КЕ> 1766 root 40 0 9680K 1480K select 2 0:11 0.00% syslogd
КЕ> 1909 bind 40 0 69244K 55360K uwait 1 0:06 0.00% named{named}
КЕ> 1909 bind 40 0 69244K 55360K uwait 1 0:06 0.00% named{named}
КЕ> 1909 bind 40 0 69244K 55360K uwait 1 0:06 0.00% named{named}
КЕ> 1909 bind 40 0 69244K 55360K uwait 3 0:06 0.00% named{named}
КЕ> 8 root 16 - 0K 8K syncer 0 0:06 0.00% syncer
КЕ> 1909 bind 4 0 69244K 55360K kqread 2 0:06 0.00% named{named}
КЕ> in compare to FreeBSD-9,
КЕ> 10-CURRENT has only one {swi1: netisr 3}
КЕ> 9- has four process: {swi1: netisr 0} {swi1: netisr 1}
КЕ> {swi1: netisr 2} {swi1: netisr 3}
КЕ> last pid: 40679; load averages: 2.38, 2.39, 2.28 up 2+05:31:50 21:23:43
КЕ> 294 processes: 7 running, 269 sleeping, 18 waiting
КЕ> CPU 0: 1.2% user, 0.0% nice, 20.4% system, 23.9% interrupt, 54.5% idle
КЕ> CPU 1: 1.2% user, 0.0% nice, 10.6% system, 29.8% interrupt, 58.4% idle
КЕ> CPU 2: 0.4% user, 0.0% nice, 10.2% system, 26.7% interrupt, 62.7% idle
КЕ> CPU 3: 1.2% user, 0.0% nice, 16.1% system, 22.4% interrupt, 60.4% idle
КЕ> Mem: 750M Active, 2700M Inact, 307M Wired, 83M Cache, 112M Buf, 58M Free
КЕ> Swap: 4096M Total, 49M Used, 4047M Free, 1% Inuse
КЕ> PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND
КЕ> 11 root 155 ki31 0K 32K RUN 1 37.1H 59.23% {idle: cpu1}
КЕ> 11 root 155 ki31 0K 32K RUN 3 37.3H 58.79% {idle: cpu3}
КЕ> 11 root 155 ki31 0K 32K RUN 2 36.8H 57.62% {idle: cpu2}
КЕ> 11 root 155 ki31 0K 32K CPU0 0 34.9H 51.46% {idle: cpu0}
КЕ> 12 root -72 - 0K 160K CPU2 2 778:00 39.99% {swi1: netisr 3}
КЕ> 12 root -72 - 0K 160K CPU1 1 558:58 22.56% {swi1: netisr 1}
КЕ> 12 root -92 - 0K 160K WAIT 0 424:04 16.60% {irq256: re0}
КЕ> 12 root -72 - 0K 160K WAIT 3 204:04 14.36% {swi1: netisr 0}
КЕ> 12 root -72 - 0K 160K WAIT 1 224:14 7.62% {swi1: netisr 2}
КЕ> 13 root -16 - 0K 32K sleep 0 123:28 5.37% {ng_queue0}
КЕ> 6907 root 23 0 15392K 5348K select 2 123:59 5.22% snmpd
КЕ> 13 root -16 - 0K 32K sleep 0 123:32 5.18% {ng_queue3}
КЕ> 13 root -16 - 0K 32K sleep 0 123:20 5.08% {ng_queue1}
КЕ> 13 root -16 - 0K 32K sleep 0 123:20 5.03% {ng_queue2}
КЕ> 3605 root 25 0 10460K 3704K select 0 22:19 2.49% zebra
КЕ> 16519 firebird 20 -10 251M 158M usem 1 0:49 1.32% {fb_smp_server}
КЕ> 5553 root 20 0 205M 102M select 3 39:29 0.98% {mpd5}
КЕ> 61490 freeradius 20 -20 354M 317M usem 1 3:40 0.63% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 2 3:33 0.63% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 0 3:58 0.54% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 0 3:32 0.54% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 1 3:29 0.54% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 2 3:23 0.54% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 3 3:22 0.54% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 3 3:35 0.44% {radiusd}
КЕ> 61490 freeradius 20 -20 354M 317M usem 0 3:21 0.44% {radiusd}
I have made a patch. See here:
http://www.freebsd.org/cgi/query-pr.cgi?pr=164130
--
С уважением,
Коньков mailto:kes-kes at yandex.ru
More information about the freebsd-current
mailing list