> I am running it now. It still has the following glaring bugs:
...
> snmp ifTable still returns cef-layer interfaces.
>I'd argue this is not a bug, but a feature (in the true sense). It
>is the only way you get to know the ordering of these, which is
>vital if you want to make any sense of information that uses
>it (like Netflow export).
The ordering of what?
>Just step over them when you walk the tree. You have to step
>over all sorts of stuff in ATM interfaces etc. anyway, so what's
>the problem?
That is not always easy when they change dynamically and your NMS is looking
at the whole tree during discovery. As far as it being a feature, it is
being yanked in 12.0.14 mainline, and is gone in 12.0.13(1). The lead NM
developer at Cisco took the issue to IETF in Pittsburgh and the consensus
was that any data that needed to latch into an interface data structure
should use the natural ifIndex, not a virtual one. As far as I know, the
BGP Policy MIB uses the cef-layer sub. What does netflow export have to do
with this? The interface table returns no data on the cef-layer. Why not
latch into the interface itself? If you could get CEF switching stats from
it, that may be useful.
chris
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:17 EDT