high CPU usage

[jasonlsch]

Hi -

I recently discovered that everyday around the same time, 4:30 p.m., my Stack starts to have issues.

ran: top

Mem: 250468K used, 4168K free, 0K shrd, 20724K buff, 44432K cached
CPU: 6.8% usr 21.3% sys 0.0% nic 0.6% idle 0.0% io 17.0% irq 54.0% sirq
Load average: 11.67 11.66 10.96 5/173 1995
PID PPID USER STAT RSS %MEM CPU %CPU COMMAND
1275 2 root RW< 0 0.0 0 68.9 [tbcm_msm_tx1]
1425 1 root S < 27252 10.6 0 5.9 ./hal
1445 1 root S 4252 1.6 0 4.0 ./fdb
1223 2 root RW< 0 0.0 0 4.0 [bcmLINK.0]
1281 2 root RW 0 0.0 0 3.8 [bcmRX]
1262 2 root DW< 0 0.0 0 2.7 [tocpd_tx]

Is there a way to discover what this tbcm_msm_tx1 process is? After a few hours, maybe less, it returns to normal, meaning it uses no CPU.

Thanks

[jasonlsch]

Forgive my ignorance but what do you think they will be complaining about?

[BradP · Accepted solution]

It depends on your environment. A possible option would be to lower it day-by-day after the spike occurs and monitor for complaints.

[jasonlsch]

I changed "configure iparp max_pending_entries 256" to 64. The tbcm_msm_tx1 process now uses 20% - 30% when it runs, not 70% - 80%. Is it ok to go even lower?
Thanks

[BradP]

It means the CPU of the switch is generating a lot of traffic. Check out these articles:
https://gtacknowledge.extremenetworks.com/articles/Solution/High-CPU-utilization-from-the-tbcm-msm-tx-process
https://gtacknowledge.extremenetworks.com/articles/Solution/High-CPU-utilization-for-the-tbcm-msm-tx1-and-bcmRX-coupled-with-CPU-congestion-on-BD8800-IO-modules

Thanks
Brad