[c-nsp] Catalyst 6500 Snmp IfIndex Assignment Logic

[Michael Balasko]

Thanks Phil.

I see the ifindex of 1 on the Sup T1/1 etc as well but the line cards have me puzzled. The really really bad part is that although I can peel the ifindex table fill off the converted IOS box I can't put it back as us mortals cannot tftp write to nvram.

I kinda like the flipping indexes to names so I might go down that route. I appreciate the help!




Michael Balasko
CCSP,CCNP,MCSE,SCP
Network Specialist II
City of Henderson
240 Water St.
Henderson, Nv 89015
P:702-267-4337
C:702-373-2730

Coincidence, n.:
  You weren't paying attention to the other half of what was going on.

On Dec 8, 2011, at 11:45 PM, Phil Mayers wrote:

On 12/09/2011 12:47 AM, Michael Balasko wrote:
All,


For the TL:DR crowd- Does anyone know dead nuts how the Catalyst 6500
allocates SNMP ifindex numbers? Specifically the 6513/Sup720-10Gig?

Sort of.

It does it sequentially based on interface creation in my experience. If you take a completely empty box, chances are Te1/1 will have ifindex 1, Te1/2 ifindex 2, and so on.

For example, we have a "been in service long time" box, and the ifindex values are:

Te1/1 = 1
Te1/2 = 2
Te1/3 = 3
Te1/4 = 4
Gi5/1 = 5
Gi5/2 = 6
Gi6/1 = 7
Gi6/2 = 8
Gi8/1 = 9

...and when this chassis was intially powered up, it had a 6704 in slot 1, sup720 in slot 5 & 6, and SFP linecard in slot 8 & 9. The ifindex values climb sequentially through those cards until I reach virtual interfaces:

Vlan1 = 105
EOBC0/0 = 106
Null0 = 107
Loop1 = 108
Loop2 = 109

...and then into SVIs and such. As far as my memory can tell, the ifindex values seem to be sequential for when I created the SVIs; likewise, the newer linecards in other slots have much higher ifindex values.

So, I reckon it's sequential, but I can't explain why you're seeing other behaviour.

You've already indicated you're aware of the ability to persist ifindex values, but it sounds like you want to control the assignment of new ones?


Subtext: We are converting our Core infrastructure switches from
CatOS to IOS(YAY!)  and we have nearly 500 devices plugged into EACH
of these monsters(WOW). The ifindexes of course are going to change
and I would like to be able to predict them (HAHAHAH) to what they
will be and or figure out if I can manipulate them such that I can
have a script ready to realign our monitoring and trending systems.

I'm afraid not. AFAIK you can't specify an ifindex (which seems like it would be a pretty trivial feature - one global command for an "automatic" ifindex range, one per-interface command for a static value, presumably outside of that range).

I would plan to instead use snmp against e.g. ifName before and after, and upgrade your NMS. Tedious I know (especially if you are using something horrible, like Cacti ;o)

The only other suggestion I can make is to try copying the ifindex persist file for CatOS and IOS off - they might be the same format, or if not, it might be possible to convert one to the other, and "force" the IOS box to read the persist file.
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/