[c-nsp] Cat6500 odd arp behavior

[Jeff Kell]

On 1/24/2013 3:24 PM, Vinny_Abello at Dell.com wrote:
> Thanks Andrew... I should have elaborated further. The hosts aren't directly connected to the 6500. The 6500 aggregates several TOR switches just doing pure layer 2, no trunking or tagging or anything. The 6500 provides an SVI for each VLAN though to act as a gateway.

I had a somewhat similar issue a few months ago, I can dig up the TAC
case if it helps, but there was no real resolution.

We have a VRF that runs building controls and power monitoring.  There
is a backbone vlan on our core 6509 that carries the VRF for those
buildings with a routed local vlan for this purpose, and another vlan
(same VRF) that was routed out of the core for legacy devices not yet
converted over to routing within the building.

At any rate, the legacy vlan has an SVI in the core and was trunked out
to the remaining legacy buildings, typically to a 3550/3560 EMI.

After a campus-wide power outage that outlasted our building UPS's, a
number of the power meters were "unreachable" from the core.  No ARP
entries, but the mac-address table was populated on the proper vlan.  In
the buildings themselves, the ports were on the proper vlans, and the
mac-address tables populated.  After numerous combinations of clear
mac-address and clear arp and other efforts on both the core 6509 and
building switches, nothing changed.

In desperation, I created an SVI with a secondary address in the
building switches.  It *could* reach the meters and populated the local
ARP table.  The 6509 could not reach the power meters (same symptoms)
but could reach the new SVIs.  The new building SVI could also reach the
other "unreachable" meters.

Since the meters "seemed" to be OK and the only things at fault, we
wrote it off to something peculiar with the meters.  We even changed the
IP of the meter and the core 6509 *could* reach it, until we changed it
back.

Since we were going to redo them to be routed at the building, we went
ahead and did that and just wrote it off as a Siemens problem :)

But it was the strangest thing I've ever seen on a 6509 for what should
have been a CCNA-intro lab exercise (just a flat vlan, nothing fancy
except it was in a VRF).