We have two 24 port 3850 switches stacked. Now CPU processing of the stack is very high and it causes the packet loss. Cisco shows fed(PunjectRx and Xcvr) process is mostly consuming the CPU. See the attached process output. Is there any solutions to mitigate this issue?
Hi Really you should TAC it as per the bug if you have support and yes it effects stacks
High CPU utilization due to FED and punjectrx. These are some possible processes, please have TAC confirm this defect as there could be multiple reasons why this occurs.
Conditions: This can be easily seen in the switches/stacks that has long uptime (several months).
if your going to a new release go to one of the recommended versions for stability the star versions , these are new package releases Everest and Denali and need to be install mode , make sure to read the release notes of the release your going to there will be boot loader and micro code upgrades coming from such an older release like 3.6.5 , other choice is jump up in same train to 3.6.9 from last year , either way chose an MD release