BUG in 'Detailed Device Report' for cisco 7200 ??

Unanswered Question
Jul 19th, 2007
User Badges:
  • Blue, 1500 points or more

I am currently looking at the 'Detailed Device Report' for a cisco7204VXR device;

it states 'Slot Capacity' = 5 and 'Free Slots' = 1;

But definitly there are 2 available slots for this device.

Looking at CiscoView I see modules in slot 1 and 4, double-check with 'show diag' on the router states the same


It seems the algorithm to define the 'Slot Capacity' is wrong for cisco72xx device type. I checked this for several 72xx and it was truth for all. (Perhaps this is because the 'Slot Number' for the NPE slot has a value of '-1')


For the 'Other Interfaces' section I have got the following question:

why is there no 'Parent Index' for these interfaces, so that these 'logical Interfaces' can easily be mapped to there physical ones? Currently this info must be drawn out from the 'Name' field of 'Interface Description'




  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (3 ratings)
Loading.
Martin Ermel Fri, 07/20/2007 - 02:28
User Badges:
  • Blue, 1500 points or more

could anybody confirm this behaviour for the cisco72xx chassis?


In fact it will be related to the ICServer rather then the report engine, I think.

I also noticed, that CiscoView cannot handle the case if the 'Vendor Type' for a module cannot be determined and is set to 'cevModuleUnknownCard' in DDR; CV is then unable to place a modules in the right slot.

CV dev package is RTR7000 ver 21.0


I added some screenshots from 2 devices to show the problem.




Joe Clarke Fri, 07/20/2007 - 13:19
User Badges:
  • Cisco Employee,
  • Hall of Fame,

    Founding Member

This looks like a pure device bug to me. RME is using the ENTITY-MIB to manage the VXRs, and it seems like the entPhysicalTable is messed up big-time. The total slots in a 7206VXR should be 7 as reported by chassisSlots.0 from the OLD-CISCO-CHASSIS-MIB. Free slots are calculated by walking through the containment information, and determining which slots are occupied.


I have a 7204VXR running 12.2(31)SB3 which shows correct containment information, so I'm fairly certain we're looking at a device problem.


I'll be at Networkers all next week, so my forum attention will be much less. You might want to open a TAC service request for this with a show tech and a walk of the entPhysicalTable from one of the affected devices.

Martin Ermel Fri, 07/20/2007 - 15:06
User Badges:
  • Blue, 1500 points or more

In your case are all modules listed with a 'Vendor Type' (Module Information)? - If one is not, could you please check how CiscoView looks like for this.


in one case I have 6 chassis all with IOS 12.2(15)T13.

perhaps the problem is with the I/O controller. The I/O controller sometimes is listed in the 'Processor Information' section, and sometimes under the 'Module Information'. I have no idea why this is treated different.

If it is listed as a module, the calculation for free slots is wrong.


If there are additionally some unrecognized cards, like a ''2 Port Fast Ethernet/ISL 100BaseTX Port Adapter' ('Module Information'.'Description'), then all is going to be messed up in CiscoView as well


A quick look over the 6 chassis show that every one has at least one of the issues (IO Controller as a module and/or an unrecognized card)


I will try to get info about some other cisco7200 with different IOS.

It's a pitty but I have no option to come to the Networkers next week, it is too far .... ;-)


Joe Clarke Fri, 07/20/2007 - 15:38
User Badges:
  • Cisco Employee,
  • Hall of Fame,

    Founding Member

The different NPE types dictate whether or not the I/O controller will be listed as a module or part of the processor. The NPE400 is listed as a processor module where as the NPE225 has it as a module. The fact that it messes up the free slot calculation is a bug in RME, but I'm not sure how fixable that is.


See, RME relies on the overall free slot count from the device (as I said previously). When it finds the I/O controller as a module, it subtracts one from the free slot count. But this is incorrect. This module should not count against the free slot count since the I/O slot is not a PA module slot.


My VXR has all known card types, but I don't think that is what is causing your problem in CV. I think there is some other ENTITY-MIB issues. That's why the entPhysicalTable walk will be important.

Martin Ermel Thu, 07/26/2007 - 03:23
User Badges:
  • Blue, 1500 points or more

==========IO-controller listed under 'Processor Information'


device1


IOS: 12.2(17)

'Chassis Information'.'Chassis Vendor Type': cevChassis7204Vxr

'Chassis Information'.'Slot Capacitiy': 5

'Chassis Information'.'Free Slots': 3

'Processor Information'.'Vendor Type': cevCpu7200Npe225

'Processor Information'.'Vendor Type': cevC7xxxIo2FE

'Module Information' **no modules installed**



device2


IOS: 12.2(19)

'Chassis Information'.'Chassis Vendor Type': cevChassis7206Vxr

'Chassis Information'.'Slot Capacitiy': 7

'Chassis Information'.'Free Slots': 3

'Processor Information'.'Vendor Type': cevCpu7200Npe400

'Processor Information'.'Vendor Type': cevC7xxxIo2FE

'Module Information'.'Vendor Type': cevPa1feTxIsl

cevPa2feTxI82543



==============IO-controller listed as a module:


device3


IOS: 12.1(19)

'Chassis Information'.'Chassis Vendor Type': c7206

'Chassis Information'.'Slot Capacitiy': 7

'Chassis Information'.'Free Slots': 4

'Processor Information'.'Vendor Type': cpu-7200-npe150

'Module Information'.'Vendor Type': c7200-io

pa-8e


===========================


The problem is not related to the chassis type nor to where the IO-controller is listed (under 'Processor Information' or 'Module Information')

The overall slot count is correct for all cisco72xx types, if you count all slots available at the front side, no matter if they are for PA or for the IO controller.You have to look at this as if there is a slot in which you can insert a moudule or not, regardless of its function. And I think this is correct in regard to NPE-1G and NPE-2G where the IO controller doesn't need to be installed anymore - so any slot should be counted.

But the problem seems to be the fact that the NPE is removable as well.So it also resides in a slot - no matter that this slot is at the rear of the chassis. But there is no 'ENTITY-MIB::entPhysicalClass.n = INTEGER: container(5)' object created for this slot in the MIB but the NPE is listed as a module ('ENTITY-MIB::entPhysicalClass.2 = INTEGER: module(9)'). So the NPE counts for the total number of modules as well whereas the NPE slot does NOT count for the number of available slots for a chassis.

As a result, the total available slot count will always be one less then it is in reality whereas the number of FRU modules is calculated correctly which leads to an incorrect calculation for 'Free Slots'.

Changing the MIB is one option - and I think the worst one, always augment the 'Slot Capacity' for all cisco72xx series routers by '1' could be the way - and would in deed reflect the actual number of slots for these routers...


Or is it just the case that the number of slots is hardcoded in the OLD-CISCO-CHASSIS-MIB::chassisSlots.0 with a wrong value ?



I added a few files from 2 devices, snmp walk of the entPhysicalTable, CV and an excerpt form the DDR report for these devices..



Concerning the CiscoView problem when there is an unknown card (the 'Vendor Type' field stated 'cevModuleUnknownCard') and as a result I got a messed up picture - I will try to get more info on this (MIB walk ...).

What is obvious is that the 'Free Slot' calculation does not affect CV. I've got correct pictures for all Cisc72xx when all modules have a valid value in 'Vendor Type'. I assume that CV cannot handle the 'cevModuleUnknownCard' in 'Vendor Type' - I will see....





Joe Clarke Thu, 07/26/2007 - 12:28
User Badges:
  • Cisco Employee,
  • Hall of Fame,

    Founding Member

I've quickly looked over this output, and it looks like the containment information in the ENTITY-MIB is correct. The slot count will always be incorrect for these devices since the NPE and the I/O module both get counted, when the overall module count only takes into account the slots on the front (PA plus I/O). Sorry if I was not clear on this before.


As for CV, I don't see any problem with these devices. I would need to see the ENTITY-MIB output for a device that is currently having a problem being displayed in CV.

Actions

This Discussion