home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!iggy.GW.Vitalink.COM!cs.widener.edu!eff!sol.ctr.columbia.edu!destroyer!ncar!noao!arizona!arizona.edu!telcom.arizona.edu!leonard
- From: leonard@telcom.arizona.edu (Aaron Leonard)
- Newsgroups: comp.dcom.sys.cisco
- Subject: Re: AGS+ pauses for 800ms every 30 seconds
- Message-ID: <1992Sep3.084740.3656@arizona.edu>
- Date: 3 Sep 92 15:47:37 GMT
- References: <184ee4INNimp@moe.ksu.ksu.edu>
- Reply-To: Leonard@Arizona.EDU
- Distribution: world,local
- Organization: University of Arizona Telecommunications
- Lines: 82
- Nntp-Posting-Host: bugs.telcom.arizona.edu
-
- >We've got an AGS+ running Version 8.2(7) which seems to pause every 30
- >seconds. Seems likely this is RIP updates, but it seems unlikely that
- >the router should do absolutely nothing else for that period. Is there
- >a simple configuration change that we've overlooked which will make this
- >all better?
-
- We observed exactly this behavior on our (then) AGS/2s when we a)
- had 600+ RIP-advertised routes floating around our backbone and b)
- had foolishly monkeyed with a CPU scheduler param ... (let me check
- my old mail folders) ... ah, the scheduler-interval parameter.
-
- Our problem was fixed by getting rid of the scheduler-interval change,
- and, even better, filtering out the huge volume of routes from bothering
- our backbone.
-
- Appended is a mail message I sent to cs@cisco.com when I figured out
- what I had done wrong ... maybe this is your problem too.
-
- Aaron
-
- Aaron Leonard (AL104), <Leonard@Arizona.EDU>
- University of Arizona Network Operations, Tucson AZ 85721
- |> If it ain't broke, break it.
-
- Return-path: <LEONARD@Arizona.edu>
- Date: Wed, 21 Aug 1991 12:04 MST
- From: Aaron Leonard <LEONARD@Arizona.edu>
- Subject: RE: Ability of CSC/2s to handle large RIP updates
- To: TLI@CISCO.COM, CS-ROUTING@CISCO.COM, CS@CISCO.COM, LEONARD@Arizona.edu,
- TSF@Arizona.edu
- Message-id: <37BB519046A03605@Arizona.edu>
- X-VMS-To: tli@cisco.com
- X-VMS-Cc: cs-routing@cisco.com, cs@cisco.com, leonard@arizona.edu,
- X-VMS-Cc: tsf@arizona.edu
-
- Greetings, Tony:
-
- On 12-Aug-1991, I wrote,
-
- The University of Arizona has 4 core cisco AGS/2 routers with CSC/2
- processors. Up until recently, they managed to route maximum-size
- packets with few drops (<<.5%) and good latency (less than 1% of
- packets forwarded with > 100ms delay.)
-
- A week or so ago, however, all four began displaying the
- following characteristics:
-
- Packet loss up to 1-3% consistently, around the clock.
- 10% of packets forwarded with > 100ms delay, with perhaps
- 5% taking > 1000ms.
-
- What changed a week ago: the number of nets being RIPped
- onto our backbone increased from 90 to 600, due to the
- fact that NSI began RIPping routes to all NSI nets.
-
- We have found that the packet dropouts/delays happen on 30-sec.
- cycles (aha!), coincident with either the routing updates
- being issued from the NSI router, or being redistributed by
- our AGS/2s.
-
- Since then, I discovered that the problems is due to user
- misconfiguration (i.e., by myself.)
-
- What I did was this:
-
- When one of our routers was suffering 100% CPU saturation
- (due to an as yet unidentified problem), I misguidedly set
- the SCHEDULER-INTERVAL param to 500 on our core routers.
- When I realized that this was a useless step, I *thought*
- I turned off this param. Unfortunately, I didn't.
-
- Last night I did away with the SCHEDULER-INTERVAL param, and
- since then the AGS/2s are back to performing normally, that is,
- with less than .5% packet loss.
-
- By the way, thanks for pointing out the access-list approach
- to controlling rip updates. We'll try it out and see whether
- it causes performance to suffer.
-
- Thanks,
-
- Aaron
-