3750X Upgrade to 15.0

[jrmildren]

I upgraded one of our new 3750x's to 15.0(2)SE5 from 12.2 and now I am seeing the following message in the log..

Mar 30 01:28:55.015: platform assert failure: hwidb->snmp_if_index == ifIndex: ../src-hulc/src-common/hpm_idbman.c: 1893: hpm_register_idb_with_snmp

Mar 30 01:28:55.015: -Traceback= 4FFA04z 206281Cz 21F9718z 21FB8DCz 21FBD64z 22056E0z 21DB598z 206490Cz 1C521ACz 1C5236Cz 1C52488z 50DFA4z 50E1E4z 26D8CF8z 26D344Cz
Mar 30 01:28:55.015: platform assert failure: hwidb->snmp_if_index == ifIndex: ../src-hulc/src-common/hpm_idbman.c: 1893: hpm_register_idb_with_snmp
Mar 30 01:28:55.015: -Traceback= 4FFA04z 206281Cz 21F9718z 21FB8DCz 21FBD64z 22056E0z 21DB598z 206490Cz 1C521ACz 1C5236Cz 1C52488z 50DFA4z 50E1E4z 26D8CF8z 26D344Cz
Mar 30 01:28:55.015: platform assert failure: hwidb->snmp_if_index == ifIndex: ../src-hulc/src-common/hpm_idbman.c: 1893: hpm_register_idb_with_snmp
Mar 30 01:28:55.015: -Traceback= 4FFA04z 206281Cz 21F9718z 21FB8DCz 21FBD64z 22056E0z 21DB598z 206490Cz 1C521ACz 1C5236Cz 1C52488z 50DFA4z 50E1E4z 26D8CF8z 26D344Cz

Has anyone seen this message and know if we should be concerned?  I am not sure if it's just something weird with snmp moving to the 15.0 code from 12.2.

[Leo Laohoo]

I upgraded one of our new 3750x's to 15.0(2)SE5 from 12.2 and now I am seeing the following message in the log

My testing of version 15.0(2)SE5 lasted for about 15 minutes.  And the entire time was attempting to ROLL BACK.   When I loaded this version on a 3560, the CPU spiked that I lost connectivity (remotely and console).  I will NEVER recommend any sane person to use 15.0(2)SE version except for 15.0(2)SE4.  

For 3560/G/E/X and 3750/G/E/X, I will recommend 12.2(55)SE9 as one of the most stable codes Cisco's ever published.  For 3560G/E/X and 3750G/E/X (with 32 mb flash), you can also try 15.0(2)SE4.

PS:  Don't be tempted to use 15.2(1)E train either.  It's the worst IOS train I've ever seen.

[Lee Valentin]

Those traceback errors should be addressed immediately.

If you are using Cisco Prime, you could be getting hammered via SNMP so maybe you can temporarily do a 'no snmp-server' and see if the messages go away.

Another issue I encountered was a memory leak caused by Auth manager on this IOS version, and it could be a bug: CSCub85948 (Memory leak caused by CDP, LLDP or DHCP traffic).

The workaround listed on this bug is the following:

Configure the following:
no macro auto monitor
device-sensor filter-spec dhcp exclude all
device-sensor filter-spec lldp exclude all
device-sensor filter-spec cdp exclude all

Not saying that either are the fix but this helped and it is best you get TAC involved to assist.

Good luck