PFsense SLow and High CPU
-
The right command is to use the fine console menu that's there (both via SSH and on physical console).
-
Hi,
I've restart the 2 still 100% cpu
last pid: 70334; load averages: 7.58, 6.90, 7.28 up 3+11:37:44 12:41:02 172 processes: 9 running, 135 sleeping, 28 waiting CPU: 38.2% user, 0.2% nice, 34.7% system, 26.9% interrupt, 0.0% idle Mem: 111M Active, 211M Inact, 119M Wired, 74M Buf, 1531M Free Swap: 4096M Total, 4096M Free PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND 12 root -92 - 0K 224K WAIT 0 235:41 24.27% [intr{irq259: 49598 root 45 0 86728K 35944K RUN 0 2:07 19.68% /usr/local/bin 93820 root 39 0 86728K 35680K accept 1 1:03 15.97% /usr/local/bin 12 root -92 - 0K 224K WAIT 0 239:00 11.67% [intr{irq262: 10691 root 29 0 86728K 35716K accept 1 1:03 11.38% /usr/local/bin 5449 root 46 0 82632K 28476K accept 1 0:05 9.57% /usr/local/bin 41372 root 42 0 86728K 35724K accept 1 2:20 7.76% /usr/local/bin 49355 root 74 0 86824K 33060K RUN 0 0:27 7.28% php-fpm: pool 14204 root 74 0 86824K 33940K RUN 1 0:30 6.88% php-fpm: pool 54828 root 52 0 86824K 33828K piperd 0 0:27 6.40% php-fpm: pool 96397 root 52 0 86728K 33052K sysctl 0 0:04 6.40% /usr/local/bin 48718 root 52 0 82632K 28496K accept 0 0:36 5.37% /usr/local/bin 2200 root 52 0 86824K 35476K piperd 1 0:35 5.08% php-fpm: pool 45926 root 25 0 17248K 6644K kqread 1 0:03 4.39% /usr/local/sbi 29995 nobody 22 0 11400K 4020K select 1 2:35 3.47% /usr/local/sbi 27091 root 26 0 82632K 28492K accept 0 0:09 2.49% /usr/local/bin 54198 admin-itp 20 0 11384K 2724K CPU1 1 0:00 0.59% top -aSH [/code] -
The output with truncated paths is completely useless.
-
Sorry
last pid: 78757; load averages: 7.10, 7.19, 7.14 up 3+11:53:35 12:56:53 167 processes: 9 running, 132 sleeping, 26 waiting CPU: 38.8% user, 0.0% nice, 44.4% system, 16.2% interrupt, 0.6% idle Mem: 143M Active, 202M Inact, 119M Wired, 85M Buf, 1509M Free Swap: 4096M Total, 4096M Free PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND 26911 root 49 0 82632K 28484K accept 1 2:01 25.59% /usr/local/bin/php 12 root -92 - 0K 224K CPU0 0 241:42 21.78% [intr{irq262: em2:rx 0}] 90617 root 33 0 82632K 28484K CPU0 0 1:48 17.77% /usr/local/bin/php 12 root -92 - 0K 224K CPU0 0 238:40 14.06% [intr{irq259: em1:rx 0}] 14204 root 75 0 99112K 50224K RUN 1 1:40 11.47% php-fpm: pool lighty (php-fpm) 2200 root 75 0 101M 50364K RUN 1 1:50 10.69% php-fpm: pool lighty (php-fpm) 93820 root 40 0 86728K 35728K accept 1 1:24 10.16% /usr/local/bin/php 54828 root 75 0 101M 51096K RUN 0 1:42 9.57% php-fpm: pool lighty (php-fpm) 49355 root 52 0 99112K 47252K nanslp 1 1:40 9.38% php-fpm: pool lighty (php-fpm) 5449 root 27 0 82632K 28484K accept 1 1:40 7.76% /usr/local/bin/php 10691 root 33 0 86728K 35732K accept 1 1:26 5.66% /usr/local/bin/php 48718 root 52 0 82632K 28496K accept 1 1:55 1.46% /usr/local/bin/php 29995 nobody 21 0 11400K 4020K select 1 2:54 1.27% /usr/local/sbin/dnsmasq --all-servers --rebind-localhost-ok --stop-dns-rebind --dns-forward-max=5000 --cache-size=100 54198 admin-itp 21 0 11384K 2908K CPU1 1 0:11 1.17% top -aSH 89404 root 26 0 82632K 28488K accept 0 0:42 0.49% /usr/local/bin/php 12 root -92 - 0K 224K WAIT 0 12:20 0.39% [intr{irq260: em1:tx 0}] 47366 root 20 0 13148K 5484K kqread 0 0:06 0.20% /usr/local/sbin/lighttpd -f /var/etc/lighty-test-CaptivePortal.confWe also have a lot of these logs, don't know if it's linked:
Sep 28 12:52:42 lighttpd[47366]: (connections.c.137) (warning) close: 21 Connection reset by peer Sep 28 12:52:51 lighttpd[47366]: (connections.c.137) (warning) close: 13 Connection reset by peer Sep 28 12:53:05 lighttpd[47366]: (connections.c.137) (warning) close: 13 Connection reset by peer Sep 28 12:53:15 lighttpd[47366]: (connections.c.137) (warning) close: 15 Connection reset by peer Sep 28 12:53:46 lighttpd[47366]: (connections.c.137) (warning) close: 15 Connection reset by peer Sep 28 12:53:49 lighttpd[47366]: (connections.c.137) (warning) close: 18 Connection reset by peer Sep 28 12:54:11 lighttpd[47366]: (connections.c.137) (warning) close: 18 Connection reset by peer Sep 28 12:54:35 lighttpd[47366]: (connections.c.137) (warning) close: 21 Connection reset by peer Sep 28 12:54:57 lighttpd[47366]: (connections.c.137) (warning) close: 14 Connection reset by peer Sep 28 12:55:02 lighttpd[47366]: (connections.c.137) (warning) close: 16 Connection reset by peer Sep 28 12:55:15 lighttpd[47366]: (connections.c.137) (warning) close: 14 Connection reset by peer Sep 28 12:55:33 lighttpd[47366]: (connections.c.137) (warning) close: 14 Connection reset by peer Sep 28 12:55:54 lighttpd[47366]: (connections.c.137) (warning) close: 13 Connection reset by peer Sep 28 12:55:57 lighttpd[47366]: (connections.c.137) (warning) close: 14 Connection reset by peer Sep 28 12:56:21 lighttpd[47366]: (connections.c.137) (warning) close: 22 Connection reset by peer Sep 28 12:56:23 lighttpd[47366]: (connections.c.137) (warning) close: 18 Connection reset by peer Sep 28 12:57:05 lighttpd[47366]: (connections.c.137) (warning) close: 16 Connection reset by peer Sep 28 12:57:18 lighttpd[47366]: (connections.c.137) (warning) close: 14 Connection reset by peer -
Disable the Captive portal thing.
-
Do you thin kthe Platform is not powerfull enough ? ( we have 100-150 users)
-
also might try
sysctl dev.em.0.debug=1 ```from console -
karlzre - Apparently not in your case when idiots/broken mobile shit keeps hammering the CP over and over again. (Other than that, you told us exactly NOTHING about your "platform"). And yeah, there's also something very broken about the NICs. Did you enable some nonsense like polling there?
-
also might try
Code: [Select]
sysctl dev.em.0.debug=1
from consoledev.em.0.debug: -1 -> -1Intel(R) Atom(TM) CPU N270 @ 1.60GHz Current: 800 MHz, Max: 1600 MHz 2 CPUs: 1 package(s) x 1 core(s) x 2 HTT threadsMemory usage
13% of 2006 MB
SWAP usage
0% of 4096 MB
Disk usage
/ (ufs): 2% of 23G
/var/run (ufs in RAM): 3% of 3.4M -
Did you enable some nonsense like polling there?
Enable device polling is disabled.
-
Intel(R) Atom(TM) CPU N270 @ 1.60GHz Current: 800 MHz, Max: 1600 MHz 2 CPUs: 1 package(s) x 1 core(s) x 2 HTT threadsThis CPU is just old slow POS, but that thing is running at half the freq it's spec'ed for with 100% CPU usage? :o Try fiddling with the powerd settings.
Also, post some information about the NICs.
-
This CPU is just old slow POS, but that thing is running at half the freq it's spec'ed for with 100% CPU usage? :o Try fiddling with the powerd settings.
Also, post some information about the NICs.
I have a bit more power behind my box and Ive run into the same thing.. Is powerd on minimum capable of causing this issue?
my machine gets loaded up with 60mbits of torrent/netflix traffic (that is my connection speed) from time to time and I see that [kernel{em0 taskq}] and [intr{swi5: fast taskq}] tasks taking up the majority of the available CPU.. Polling is turned off..
This machine as built from spare parts that I had laying around is it possible that im just running into a limitation with the hardware?
IC2D 2.13ghz, Asus OEM shitty mobo, 3gb ram, Intel Pro/1000 Legacy (external), Realtek 8169 (Internal)
-
I have a bit more power behind my box and Ive run into the same thing.. Is powerd on minimum capable of causing this issue?
You need to fiddle with whatever settings are available there. People here mentioned boxes where CPU is never run at full freq unless powerd is enabled somehow. Sure if set @minimum, that's a good candidate.
-
You may also want to look at https://forum.pfsense.org/index.php?topic=59028.0 to remove those messages from the log and help cut down the noise. Looks like they may not mean anything.
-
I have a bit more power behind my box and Ive run into the same thing.. Is powerd on minimum capable of causing this issue?
You need to fiddle with whatever settings are available there. People here mentioned boxes where CPU is never run at full freq unless powerd is enabled somehow. Sure if set @minimum, that's a good candidate.
Interestingly enough, I switched it to adaptive and the overall load on the machine has lowered a HUGE amount. Lets say from 6.0 to under 0.7-0.5.. Kind of ironic that the feature meant to save energy is actually causing an increase… (in this situation)
Seriously, Turn on the profile "minimum" and max out your connection, your load will raise slowly.. turn it to Adaptive and watch it lower almost immediately. "[kernel{em0 taskq}]" is the task using all the cpu..
With "Minimum" power setting on powerd last pid: 11607; load averages: 3.36, 1.79, 0.99 PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND 0 root -92 0 0K 192K CPU0 0 647:46 79.49% [kernel{em0 taskq}] 12 root -56 - 0K 272K WAIT 1 432:38 51.46% [intr{swi5: fast taskq}] 87987 root 100 20 25160K 6464K CPU0 0 0:04 26.27% /sbin/pfctl -ss 11 root 155 ki31 0K 32K RUN 1 390.8H 15.97% [idle{idle: cpu1}] 11 root 155 ki31 0K 32K RUN 0 385.3H 11.18% [idle{idle: cpu0}] With "HiAdaptive" power setting on powerd last pid: 61673; load averages: 0.53, 1.23, 0.89 PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND 11 root 155 ki31 0K 32K RUN 1 390.9H 97.56% [idle{idle: cpu1}] 11 root 155 ki31 0K 32K CPU0 0 385.3H 83.98% [idle{idle: cpu0}] 0 root -92 0 0K 192K - 0 648:19 10.16% [kernel{em0 taskq}] 12 root -56 - 0K 272K WAIT 1 433:00 5.66% [intr{swi5: fast taskq}] 12 root -92 - 0K 272K WAIT 1 58:56 0.98% [intr{irq19: em0 uhci1}] 12 root -92 - 0K 272K WAIT 0 40:54 0.78% [intr{irq16: re0 uhci3}]No other changes were made on the machine other then fiddling with the powerd settings.. I never would have thought to look there.. Kudos and thanks!
-
You're misunderstanding what're being reported. Load is relative. If your CPU is running at 100mhz, then it's not hard to reach 75% usage. The reason "load" is high when using minimum is because the CPU is being down-clocked, you are saving power.