Upadte inventory make a 7204VXR reboot

[mmisonne]

Hello

Apparently with LMS3.2 (RME 4.3.1), each time we do an update inventory on a 7204VXR, it makes the cpu proc increase to 90%.

The IOS version is old, c7200-is-mz.122-2.T4.bin, but the customer do not want to upgrade now.

Last time, the routeur crashes (see sh ver and crashinfo)

With old LMS (2.6) it never crashed.

Is it a way to avoid this pb, apart from doing a "supend device" in RME>device>dev mgt ?

Regards

Michel Misonne

[Joe Clarke]

This looks like CSCdx41392. It is fixed in 12.2(11)T and higher. However, it looks like a view blocking the CISCO-COPS-CLIENT-MIB will prevent this crash.

The COPS-PR MIB is no longer part of the Cisco IOS code as of mid-May 2002.

All subsequent images do not support this unsupported feature. If an

issue is reported against the COPS-PR MIB code, it will not be fixed

as this code is no longer supported.

Results of this caveat (i.e., a router crash during a mibwalk) can be found

with either V1 or V2.

Conditions: This caveat will resolve all COPS-PR MIB issues as that code will

no longer be present for the Cisco 7200 and Cisco 7500 platform with images

listed below:

12.2(08)YW03

12.2(15)BX

12.2(15)ZN

12.2(15)BZ

12.2(10.07)PI05

12.2(10.07)T

All other platforms will have code removed in phase 2 which will be completed

in Cisco IOS Release 12.3(05.05)T. Cisco 2600 Routers Fixes are verified in

Cisco IOS Release 12.2(15)T7 and Release 12.2(17).

Workaround: To disable the components from CISCO-COPS-CLIENT-MIB, define an

SNMP view. Required commands to achieve this are as follows:

no snmp-server community RO

snmp-server view no_cops_client internet included

snmp-server view no_cops_client ciscoCopsClientMIB excluded

snmp-server community view no_cops_client RO

snmp-server community view no_cops_client RW

[mmisonne]

Hello,

We tried to enter all those 5 commands but it is the same pb, i.e. the cpu increase dramatically, when doing a update inventory.

Regards

Michel

[Joe Clarke]

But does the device crash? The stack trace in your crash info corresponded to that bug. If the CPU is rising, you will need to capture the "show stack PID" for the PID of IP SNMP when the CPU is pegged.

[mmisonne]

Hello,

Enclosed the output you asked.

The "sh stack pid"  when the cpu increased.

THe "SNMP engine" increase to 32%, but the "sh cpu history" shows 90%

REgards

Michel Misonne

[Joe Clarke · Accepted solution]

It looks like you're also seeing CSCdz13959 which is fixed in 12.2(14)T.  To workaround this, you would need to exclude the ciscoFlashMIB in addition to the other branches.

[mmisonne]

Thanks, it's OK now !