EVENTLIB-3-CPUHOG:Switch 1 R0/0: cli_agent

[senthilkumars]

Hi,
We observed below alarm on switch 3850 and couldn't find RCA for the same.Can anyone throw some light on this?

Alarm Log:

%EVENTLIB-3-CPUHOG:Switch 1 R0/0: cli_agent:  undefined: 1691ms, Traceback=1#988329bea1812b71673b732c2a899f01
  binos:FFE8409000+160F0 binos:FFE8409000+160F0 evutil:FFE79AE000+A082 uipeer:FFE7E08000+34AB4
uipeer:FFE7E08000+13178 evlib:FFE8476000+A5F8 evlib:FFE8476000+B3E0 :AAB69C9000+11A74 c:FFE3D44000+44298 :AAB69C9000+D6E0

Hardware&IOS detail:

Switch Model   WS-C3850-48T

SW Version   16.3.7 

SW Image    CAT3K_CAA-UNIVERSALK9

[Georg Pauwen]

Hello,

you are probably hitting the bug below:

https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvo47060/?rfs=iqvred

[senthilkumars]

Thanks for your reply.
The provided bug belongs to HPM main process but in my case cli_agent process cause cpu hogging.

[Leo Laohoo]

Post the complete output (page 1 only).

[senthilkumars]

During the normal operation of switch ,we observed only below provided alarm in "show log" and no other functionality affect during that period.As of now and post to the alarm ,cpu process is 2% only.

%EVENTLIB-3-CPUHOG:Switch 1 R0/0: cli_agent: undefined: 1691ms, Traceback=1#988329bea1812b71673b732c2a899f01 binos:FFE8409000+160F0 binos:FFE8409000+160F0 evutil:FFE79AE000+A082 uipeer:FFE7E08000+34AB4 uipeer:FFE7E08000+13178 evlib:FFE8476000+A5F8 evlib:FFE8476000+B3E0 :AAB69C9000+11A74 c:FFE3D44000+44298 :AAB69C9000+D6E0

[pkumari4]

Hello Senthil,

I can see that you are seeing CPU HOG message with a traceback. Can you tell me if this is just a one-time occurrence or you are seeing this again n again on the switch?

This error message looks to be purely cosmetic and is scheduled to be fixed in 16.3.8. Long story short it is a very deep CPU handoff false positive that generates that Syslog CPUHOG message. Upgrading to that should solve our issue.

Regards,

Priyanka

[senthilkumars]

Hi Priyanka,

Thanks for your response.
Totally four switches has same hardware and software ,on which 3 switches got this alarm only once.

[pkumari4]

Hello Senthil,

As discussed this error message looks to be purely cosmetic and is scheduled to be fixed in 16.3.8. Long story short it is a very deep CPU handoff false positive that generates that Syslog CPUHOG message. Upgrading to that should solve our issue.

I will suggest you perform the upgrade on the switch to 16.3.8.

Regards,

Priyanka

[Leo Laohoo]

The next time this happens, post the complete output to the command "show processes cpu platform sorted location switch 1 r0".
The output will tell us which process is hogging the CPU or which CPU is "hot spinning".

[senthilkumars]

Thank you for your reply.

The "cli_agent" process is hogging the CPU as per the alarm log.

%EVENTLIB-3-CPUHOG:Switch 1 R0/0: cli_agent:  undefined: 1691ms

[senthilkumars]

We observed same alarm in another switch (same H&W and software) and attached output of cmd show processes cpu platform sorted location switch 1 r0 for your perusal.

[Leo Laohoo]

Based on the attachment I see no evidence of high CPU.
The CPU has an average of 8%