Switching

Hello all.

We are seeing this ONLY on 2300s *we have 3400 and 4300 actually doing more and not showing any stress*

Ive disabled syslog and snmp on the switch .

Im looking for a good doc that explains what gets sent to eventd and why..  ALSO.. Im dabling on the shell side wanted to know if anyone

could give any guidace on ktrace / kdump ..   Im just trying to see what all the process is doing ..  

Any advice on how to dive in and see what is making the cpu so high , id be very greateful.

this one is running 18.1R1.9

run show system processes summary
last pid: 24212; load averages: 1.60, 1.67, 1.71 up 7+12:04:47 10:00:30
197 processes: 4 running, 174 sleeping, 1 zombie, 18 waiting

Mem: 87M Active, 986M Inact, 275M Wired, 21M Buf, 558M Free
Swap:

PID USERNAME PRI NICE SIZE RES STATE TIME WCPU COMMAND
4311 root 82 0 718M 276M RUN 35.9H 32.96% fxpc{fxpc}
4284 root 40 0 283M 11476K RUN 646:38 28.96% eventd

So i found another post about turning up task accounting to see if that will help me.. I get nothing from this output showing the issue

task has been running for about 3 hours.    Can anyone help point me in the right direction.

show task accounting
Task accounting is enabled.
Task Started User Time System Time Longest Run
Scheduler 12348 0.6 0.7 0.0
LMP Client 1995 0.1 0.2 0.0
Memory 221 0.0 0.0 0.0
hakr 35 0.0 0.0 0.0
MPLSOAMD I/O./var/run/mpl 998 0.1 0.2 0.0
MPLS_PM Proto I/O./var/ru 747 0.0 0.0 0.0
BFD I/O./var/run/bfdd_con 747 0.0 0.0 0.0
Mirror Task.128.0.0.5+572 2995 0.1 0.1 0.0
KRT 2996 0.0 0.0 0.0
Redirect 56 0.0 0.0 0.0
MGMT_Listen./var/run/rpd_ 40 0.0 0.0 0.0
SNMP Subagent 1398 0.1 0.1 0.0
Scheduler 35 4.4 3.6 7.9 {krtio-th}

 show task io | except "0 0 0 0 0"
Task Name Reads Writes Rcvd Sent Dropped
RPD Unix Domain Server./var/ru 3 0 3 4 0
RPD Unix Domain Server./var/ru 0 0 0 1 0
RPD Unix Domain Server./var/ru 2 0 0 0 0
MPLS_PM Proto I/O./var/run/ppm 32921 1 32921 16463 0
BFD I/O./var/run/bfdd_control 32921 1 32921 16462 0
Mirror Task.128.0.0.5+57266 65800 1 0 0 0
KRT_PURGE_TASK 0 0 105 107 2
KRT 7 0 7 0 0
krt unsolic client 1159 0 5920 0 0
krt solic client JUNOS 0 0 1 4 3
MGMT.local 0 0 1 0 0
MGMT_Listen./var/run/rpd_mgmt 43 0 43 0 0
SNMP Subagent 44068 0 0 0 0
krt ioth solic client 0 0 8 0 0 {krtio-th}

So i found a freebsd post that helps settle my nerousness about this. As LOAD is not a good eough metric to montior this type stuff.

http://www.undeadly.org/cgi?action=article&sid=20090715034920

Im still trying to find out what sends to eventd.  If anyone has that info .. please post.. found this on another post.. But that

may not be all the things eventd handles.

https://www.juniper.net/documentation/software/junos/junos94/syslog-messages/eventd-system-log-messages.html

Are you still having this issue? We've been fighting it for some time without success. At one point the engineer found evidence that it was the Junos Space VM causing the CPU spikes, but I've turned the VM off to confirm and the problem still exists.