[nsp] 12.0(22)S2 / 12.0(23)S1

Pete Kruckenberg pete@kruckenberg.com
Wed, 4 Dec 2002 22:11:54 -0700 (MST)


I have been running 12.0(21)S5 on 7500 (two different
7507's) and found what looks like a counter bug on POS
interfaces. It doesn't appear to affect other interfaces.

SNMP polling of the POS (OC-3) interface consistently shows
nothing outbound. Inbound measures normal.

On the RP, a "show interface" shows both in-bound and
out-bound at normal levels.

Running "show interface" on the line-card itself (via the
undocumented command "if-con X" (http://boerland.com/dotu/),
it showed normal in-bound traffic, but nothing out-bound. I
guess SNMP polls directly off the line-card, not the RP.

This seemed to happen after the routers had been up about a 
week.

Anyone else seen this, or seen something different?

12.0(21)S5 also has a distributed multicast bug (on 7500 POS
interfaces) that surfaced in 12.0(21)S3 (CSCdx36692). This
was supposed to be fixed in 12.0(21)S5, but my experience is
that it is exactly as broken as it was in 12.0(21)S3.

Pete.

On Wed, 4 Dec 2002, Gert Doering wrote:

> Date: Wed, 4 Dec 2002 17:01:10 +0100
> From: Gert Doering <gert@greenie.muc.de>
> To: Jared Mauch <jared@puck.nether.net>
> Cc: Yuval Ben-Ari <yuvalba@netvision.net.il>, cisco-nsp@puck.nether.net
> Subject: Re: [nsp] 12.0(22)S2 / 12.0(23)S1
> 
> Hi,
> 
> On Wed, Dec 04, 2002 at 10:56:51AM -0500, Jared Mauch wrote:
> > 	I've run 12.0(23)S on GSR, RSP and 7200 routers without any
> > significant problems - read: crash/service affecting bugs.
> 
> Have you tested the interface counters?
> 
> I'm pleasently surprised that 12.0(21)S5 seems to have finally fixed 
> them for good (even though the results of using "show interface" and
> SNMP counters still differ by a few per cent, which I can't imagine
> how to achieve this)...
> 
> gert
> 
>