We had the same problem, and according to Cisco, the bug was unknown until
we reported it about 3 months ago.
This was definately fixed, tho, in:
IOS (tm) 7200 Software (C7200-JS-M), Experimental Version 12.1(20000914:043649) [amcrae-cosmos_e 467]
whatever the hell that is.
core1.adc uptime is 9 weeks, 6 days, 18 hours, 14 minutes
And this box is pretty busy with lots of BGP peers. About 40, of wich 3
are full 90k+ routes, one is about 40k routes, and the rest in the 0-1000
routes.
CPU utilization for five seconds: 37%/30%; one minute: 38%; five minutes: 39%
The two useful interfaces:
core1.adc>sho in | inc input rate
30 second input rate 30100000 bits/sec, 8364 packets/sec
30 second input rate 37027000 bits/sec, 6708 packets/sec
core1.adc>sho proc c | inc BGP
2 952 20237 47 0.00% 0.00% 0.00% 0 BGP Open
115 23264088 40809425 570 0.08% 0.10% 0.14% 0 BGP Router
116 2782544 11516535 241 0.08% 0.02% 0.00% 0 BGP I/O
117 274376804 1464460 187358 0.00% 3.50% 3.43% 0 BGP Scanner
On Tue, 20 Mar 2001, Alastair Pooley wrote:
> Hiya,
>
> Just to add to the NSE-1 experience have a look at
>
> CSCds64469
>
> We had a lot of hassle with this bug - any BGP changes occurring on our network (e.g. if we rebooted an IBGP peer) led to 100% CPU on the NSE-1. We have a full IBGP mesh with 3 EBGP peers each with a full internet table. The CPU usage led to peer's closing sessions and major hassles for us. Recommended IOS from TAC 121-5c.E8 failed to resolve this. As someone mentioned earlier in this thread it was like the scheduler was broken, despite the CPU being maxed it wouldn't build peering sessions. We found the best IOS was actually 12.1(6) main line release. It reported the NSE-1 as NPE-175 and all of our problems went away!
>
> In the end we pulled out the NSE-1's and we're back to NPE-300's for some stability.
>
> Cheers
> Al
>
>
> ----- Original Message -----
> From: "Christian Reichert" <c.reichert@resolution.de>
> To: "George Robbins" <grr@shandakor.tharsis.com>
> Cc: <sthaug@nethelp.no>; <cisco-nsp@puck.nether.net>
> Sent: Monday, March 19, 2001 10:20 PM
> Subject: Re: [nsp] Experience with NSE-1 and IOS 12.0S?
>
>
> > Hi !
> >
> > CSCdt33911
> >
> > It's also in versions 12.1(1)E, 12.1(2)E, 12.1(3)E, 12.1(4)E, 12.1(5)E
> > basically all we tested ;-)
> >
> > Although if you don't use MPLS, you don't necessarily have problems - in
> > most of our test, we didn't see any problems during 'normal' routing but
> > you could see that CEF is not working because packets wich should be
> > label switched were just routed.
> >
> > Cheers,
> > Chris
> >
> > George Robbins wrote:
> > >
> > > Hmmm... We have a couple running cef+pxf enabled and haven't
> > > seen any CEF bugs, and we've lived thru a lot of CEF bugs in
> > > the past. Not to say there aren't bugs, but I say it's not a
> > > dead-in-the-water kind of thing...
> > >
> > > As far as switching performance, the boost is suposed to come
> > > with the "high touch" stuff where you're looking at more than
> > > the IP address for routing decisions, otherwise you might as
> > > well just get a NPE-300/400.
> > >
> > > George
> > >
> > > > From c.reichert@resolution.de Mon Mar 19 16:26:34 2001
> > > > X-Authentication-Warning: magrathea.resolution.de: Host t1o311p80.teliauk.com [195.12.226.80] claimed to be resolution.de
> > > > Date: Mon, 19 Mar 2001 22:54:19 +0100
> > > > From: Christian Reichert <c.reichert@resolution.de>
> > > > Organization: resolution GmbH
> > > > X-Accept-Language: en
> > > > To: sthaug@nethelp.no
> > > > CC: grr@shandakor.tharsis.com, cisco-nsp@puck.nether.net
> > > > Subject: Re: [nsp] Experience with NSE-1 and IOS 12.0S?
> > > > References: <200103191649.LAA03763@shandakor.tharsis.com> <7189.985028111@verdi.nethelp.no>
> > > >
> > > > Hi !
> > > >
> > > > You can even use it with earlier IOS Versions, they might report
> > > > 'unknown NPE' but it still works - even more stable than some 12.1E
> > > > versions :-)
> > > > You certainly won't have pxf nor any support from Cisco.
> > > >
> > > > One more thing to mention about NSE-1, as long as PXF is turned on CEF
> > > > (Cisco Express forwarding) is broken ! (all versions so far)
> > > > This means if you use the 7200 as a MPLS PE, it won't work ... no pxf
> > > > solves the problem.
> > > >
> > > > Another thing, some of our tests showed that with pxf turned on the
> > > > switching performance is less than without pxf ...
> > > >
> > > > Cheers,
> > > > Chris
> > > >
> > > > sthaug@nethelp.no wrote:
> > > > >
> > > > > > I supsect the PXF technology was developed on a 12.1 base, and isn't
> > > > > > going to be retrofitted to 12.0. Someone (maybe from Cisco) mentioned
> > > > > > that the NSE board would work as if it were an NPE-300 under releases
> > > > > > where the NSE-1/PXF wasn't "supported"...
> > > > >
> > > > > That is indeed our experience with NSE-1 and 12.0(15)S1 - behaves like
> > > > > an NPE-300.
> > > > >
> > > > > Steinar Haug, Nethelp consulting, sthaug@nethelp.no
> > > >
> >
> >
> > ________________________________________________________________________
> > This message has been checked for all known viruses, by Star Internet,
> > delivered through the MessageLabs Virus Control Centre.
> > For further information visit:
> > http://www.star.net.uk/stats.asp
> >
> >
>
>
> ________________________________________________________________________
> This message has been checked for all known viruses, by Star Internet,
> delivered through the MessageLabs Virus Control Centre.
> For further information visit:
> http://www.star.net.uk/stats.asp
>
-- Alex Rubenstein, AR97, K2AHR, alex@nac.net, latency, Al Reuben --
-- Net Access Corporation, 800-NET-ME-36, http://www.nac.net --
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:32 EDT