nv-l
[Top] [All Lists]

Re: NETMON causing high SNMP Util on router

To: nv-l@lists.tivoli.com
Subject: Re: NETMON causing high SNMP Util on router
From: Sean Davidson <sean.davidson@MAIL.PUBLIX.COM>
Date: Wed, 13 Oct 1999 18:07:50 -0400
The router doesn't log who is generating the queries or what variables are
being queried. The only way I can get at that
is via debugs which would more than likely hose the router up. This router
is the backbone of my entire network, there is never a good time to bring
this one down.

IOS is 11.1(20)CA1. Been running that version on this router for several
months and haven't noticed this problem until recently. Using Cisco MIB
queries when the problem occurs, I see the snmp get variable go from 4 or 5
in a 5 sec interval to up to 70. This is what leads me to believe that the
problem is actual SNMP queries.

I've got Netview setup to do config poll. How often does this take place? I
know when NetView goes to pull the ARP cache
off this router, it takes a few minutes and CPU is high. I would only expect
this to happen once per day, but I don't recall how often the config poll
occurs. The arp cache on this router has 7000 or so entries. The route table
contains appx 160 routes.

Well, as I'm getting ready to send this, I may have found the problem. Did a
demandpoll on this guy and after 23 min, closed it. Of that 23 min, netmon
was pulling the arp table for 22 min 48 sec. Due to the way routing is
configured on this guy, he maintains a huge arp cache. It's a user
configuration issue on the router. I need to modify the way routing takes
place to get the ARP table down to a manageable size. This is no small task,
so I'm trying to find the easiest short term solution:

I know I can turn off config polling, but is there a way to tell netmon to
config poll everyone but this guy, or to tell him to config poll him only at
a certain time? What do I lose by disable config polling altogether? Is it
possible to tell netmon not to query the ARP table?

__________________________
Thanks,
Sean Davidson

Sr. Network Systems Engineer
Publix Super Markets, Inc.
P.O. Box 32015
Lakeland, Fl. 33802-2015
Email - sean.davidson@publix.com
Voice - (863) 686-8754 x6889
Fax - (863)616-5895




-----Original Message-----
From: Jorge Jiles [mailto:Jorge.Jiles@UALBERTA.CA]
Sent: Wednesday, October 13, 1999 5:22 PM
To: NV-L@UCSBVM.ucsb.edu
Subject: Re: NETMON causing high SNMP Util on router


What software version are you running on the 7507? We had a similar problem
and it was code related on the router (documented software problem from
Cisco).


At 05:01 PM 10/13/99 -0400, you wrote:
>Having a problem where the IP SNMP process on a Cisco 7507 is using a lot
of
>CPU. CPU on the router will go to 99% for 20 to 30 min. I look at the
>processes on the
>router and the IP SNMP process is getting 70% of the CPU. I stop netmon and
>CPU drops to 25%. Restart netmon and I'm fine for a while, but at some
>point, the same problem happens and the process repeats.
>
>I'm running the following:
>
>AIX v4.2.1
>Tivoli 3.6.1
>NetView/6000 v5.1.1
>
>For what it's worth, the router is a Cisco 7507 w/ RSP2. Because the SNMP
>process and not the IP forwarding or any of the processes related to
>forwarding data, the traffic has to be SNMP destined for this router. I've
>tried turning netmon tracing on for SNMP requests (tracemask 4), but I've
>got 30,000 objects in my database and this creates quite a bit of ouput, so
>I can't tell what's going on. Is there a better way to see what's
happening?
>
>__________________________
>Thanks,
>Sean Davidson
>
>Sr. Network Systems Engineer
>Publix Super Markets, Inc.
>P.O. Box 32015
>Lakeland, Fl. 33802-2015
>Email - sean.davidson@publix.com
>Voice - (863) 686-8754 x6889
>Fax - (863)616-5895
>
.
Jorge A Jiles
Network Analyst
Computing & Network Services
University of Alberta
Edmonton, Alberta
Canada


<Prev in Thread] Current Thread [Next in Thread>

Archive operated by Skills 1st Ltd

See also: The NetView Web