CSCvd20379 - "fwm" Crash From Missing Core OIFL During ISSU to 7.1(4)N1(1)

[graham-lewis]

Symptom:
Durring a multi-stage ND-ISSU software upgrade a Nexus 5K switch may experience a crash in the FWM process with the following messages:

%USER-2-SYSTEM_MSG: <<%FWM-2-FWM_ASSERT_FAILURE>> ../platform/nuova/forwarding-sw/server/fwm_pss.c.8717 assertion '0'
stack [0x8490407 0x8242249 0x69e6cc8e 0x69e7264b 0x69eabd4b 0x820b818 0x8211033 0x6aebd5c5 0x80aa211 ] - fwm
%SYSMGR-2-SERVICE_CRASHED: Service "fwm" (PID 3660) hasn't caught signal 6 (core will be saved).

 

It looks like we are running into this bug when trying to ISSU our 5596 Nexus switches - could anyone explain how to carry out the workaround ">> Add a dummy core interface and remove it (Safe)"?

 

Thanks

[mojafri]

Hi @graham-lewis,

Does your switch have Fabric path configured? This bug is ONLY applicable to such setup. To answer your question: 

Adding a dummy core port (between two connected switches) in the sense that it is configured with “switchport mode fabricpath” and brought UP. Once the core OIFL is updated, you can remove that configuration.

Please rate if you find it helpful. 

Regards,

MJ

[graham-lewis]

Hi MJ, thanks for your answer. No we don't have Fabric Path configured so it must be something else causing the crash.

[mojafri]

Yes, and if you see any core generated, open a TAC case to get it analyzed.

Don't forget to mark answers as correct if it solves your problem. This helps others find the correct answer if they search for the same problem.

[graham-lewis]

Thanks MJ, will do.