[c-nsp] CEF fun in SXF

[Brett Frankenberger]

On Thu, Dec 08, 2005 at 10:24:16PM +0100, Gert Doering wrote:
> 
> we've just had lots of fun with one of our Sup720/12.2(18)SXF boxes, and
> I'm curious whether one of you has been hit by it before, and maybe
> even has a bug ID (which would save me some time with first-level TAC) 
> or case number that I could link to...

I've seen what is probably the same problem, although I was unable at
the time to correlate it to the deletion of an SVI.  (It's happened
three times to me, all on relatively non-critical VLANs -- when it
happens to me, not all VLANs experience the packet loss -- so I wasn't
able to correlate the report of the problem (which was sometimes over a
day after it started) with any particular activity on the switch.)

In my case, two of the three instances of problems were cleared by
doing a "shut" / "no shut" on the VLAN having the problem; the third
case was cleared by either "clear arp-cache <cr>" or "clear arp-cache
interface XXXX" where XXX was the SVI for the interface experiencing
the packet loss.  (The person in the NOC who did it didn't remember if
cleared ARP on the whole box or just the interface ... my guess is that
just clearing the interface would have worked.)

> *All* directly connected (!) customers had massive packet loss - about 
> 50-80% - while packets to non-"connected" targets had no problems at all.

When it has happened to me, it has always affected just one VLAN (it's
happened three times, on three different switches; one VLAN each time),
and my packet loss rate is more like 25% - 35%.  (Actually, it may be
that it affected more than one VLAN and I only got reports on one VLAN,
but it certainly didn't affect them all -- most of the VLANs are such
that we'd have known very quickly had they been affected.)

> Does this sound familiar to anyone?

Yep.  Unfortunately, I don't have a bugID, so I can't offer you much
beyond the knowledge that you're not alone.  (I, on the other hand, am
glad to know what the triggering event probably was and thank you for
that information.)

If you get a BugID, I'd be interested in knowing what it is.  (I'll
probably try to reproduce it now that I have some idea where to start;
but I probably won't bother with a TAC case unless I get a reproducible
condition.)

     -- Brett