This is the Q&A from "Troubleshooting and Upgradation on Cisco LAN switches Session 1"
High CPU related questions:
Q. Is debug platform command CPU intensive?
A.The 'debug platform cpu-queues' command can run even when the CPU is high without any problems.
Q. Does "Debug platform cpu-queues (queus-name)" command uses cpu or not?
A. As this command not CPU intensive, you can run this command even when the CPU is high.
Q.How can we identify the process or interrupt which is causing high cpu?
A.When you issue the “show process cpu sorted” command, you can see the CPU utilization in terms of X/Y. The X value is the total CPU utilization of the system and Y value is the utilization due to interrupts. Naturally, if the Y value is closer to the X value, then the majority of the CPU utilization is due to interrupts.
Devices Crash related questions:
Q.What is parity crash?
A.In switches any data will be parsed as a binary data. Let’s say "HELLO" word will be converted into "1110011”, due to heat on the ASIC, the fast moving electrons binary value gets flipped into something else like 000111000. Supervisor/CPU will identify the integrity of the data lost with CRC checks & checksum. To recover from this situation, the supervisor will try to correct it and if it is unable to correct it, it reloads the switch to recover from this situation. The above behavior is called as Parity crash.
Q.In case of Parity crashes, do we need to replace the hardware?
A.Most of the time it will be one time occurrence. CISCO recommends to monitor the switch for 48 hours. If the same issue happens again within 48 hours then the issue could be related to potential hardware issue.