OPNsense Forum
Archive => 21.1 Legacy Series => Topic started by: manuel on May 19, 2021, 11:59:59 am
-
Hello
I have collectd running on my opnsense box which is a apu2e4 with 4 cores (AMD Embedded G series GX-412TC) and 4GB RAM. As you can see in my Grafana Graphics there is a zigzag of the CPU Usage. Normally the cpu load raises over 2 or 3 days continuously and then drops to 1%.
Does anybody know the reason for this or has made same observation?
Thank you very much for a hint.
Greetings Manuel
-
It's normal...it's related to insight data (flowd_aggregate.py) and file processing.
https://github.com/opnsense/core/issues/3587
-
Hello gpb
Thank you for your answer and the link. This is it. :D
Regards Manuel
-
Hello
After updating to newest release 21.1.5 on 23.05. 12:00 the CPU Load looks now different. As mentioned by gpb python3.7 is using quite a lot of CPU time.
last pid: 16333; load averages: 0.79, 0.89, 0.89 up 3+20:47:14 08:57:55
49 processes: 1 running, 48 sleeping
CPU: 2.8% user, 0.0% nice, 7.0% system, 0.6% interrupt, 89.6% idle
Mem: 53M Active, 1518M Inact, 461M Wired, 217M Buf, 2095M Free
Swap: 2048M Total, 2048M Free
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
17984 root 1 20 0 29M 20M select 2 35.3H 0.03% python3.7
88970 root 1 20 0 1048M 8424K select 2 10:52 0.14% openvpn
68332 root 11 20 0 39M 7824K nanslp 3 8:45 0.00% collectd
8987 root 1 52 0 1037M 3332K wait 3 2:40 0.00% sh
90751 root 1 20 0 26M 11M select 2 1:53 0.04% python3.7
16747 root 1 20 0 21M 11M select 2 1:42 0.04% python3.7
90286 root 3 20 0 32M 11M kqread 2 1:33 0.08% syslog-ng
58694 root 1 52 0 54M 32M accept 0 1:29 0.97% python3.7
Greetings Manuel
-
The graphs look the similar to me. One thing you might check is do you have PowerD enabled...check under System | Settings | Miscellaneous. Adaptive works well for my setup...this was something I overlooked for quite a while after switching to OPNsense...so just in case you never checked it. Anyhow it might make a difference.
Also, I've run into a situation where the RRD netflow data gets corrupted or confused and resetting/repairing will resolve excess CPU in this case...so that's something else to check. That's under Reporting | Settings.