From nobody@FreeBSD.org Sat Apr 8 14:31:32 2000 Return-Path: Received: from freefall.freebsd.org (freefall.FreeBSD.ORG [204.216.27.21]) by hub.freebsd.org (Postfix) with ESMTP id DE2F137B544 for ; Sat, 8 Apr 2000 14:31:31 -0700 (PDT) (envelope-from nobody@FreeBSD.org) Received: (from nobody@localhost) by freefall.freebsd.org (8.9.3/8.9.2) id OAA77242; Sat, 8 Apr 2000 14:31:30 -0700 (PDT) (envelope-from nobody@FreeBSD.org) Message-Id: <200004082131.OAA77242@freefall.freebsd.org> Date: Sat, 8 Apr 2000 14:31:30 -0700 (PDT) From: simon@lok.net Sender: nobody@FreeBSD.org To: freebsd-gnats-submit@FreeBSD.org Subject: no more buffers / can't allocate llinfo when running gated X-Send-Pr-Version: www-1.0 >Number: 17868 >Category: kern >Synopsis: no more buffers / can't allocate llinfo when running gated >Confidential: no >Severity: critical >Priority: high >Responsible: freebsd-bugs >State: closed >Quarter: >Keywords: >Date-Required: >Class: sw-bug >Submitter-Id: current-users >Arrival-Date: Sat Apr 8 14:40:01 PDT 2000 >Closed-Date: Thu Jun 1 15:44:17 PDT 2000 >Last-Modified: Thu Jun 01 15:44:32 PDT 2000 >Originator: Simon Lok >Release: RELEASE 3.4 >Organization: INW >Environment: FreeBSD digex-gw.name.space 3.4-RELEAE FreeBSD 3.4-RELEASE #1 Sat Apr 8 18:38:09 EDT 2000 root@digex-gw.name.space:/usr/src/sys/compile/ROUTER >Description: This problem occurs when running gated with a full routing table. After 20 minutes or so, when the routing table is just about done downloading (75K routes nowadays), the system begins to complain about "no more buffers." A few minutes after that, the system starts complaining "arpresolve: cannot allocate llinfo". In order to attempt to fix this, I recompiled the kernel with MAXUSERS 256 and NMBCLUSTERS 4096. (The machine has 128 MB of RAM). That didn't help. I looked on newsgroups, and there were some other people reporting this problem, but no solutions. I am wondering if there is some kind of hard limit to the size of the routing table or something like that. >How-To-Repeat: Setup a machine, compile and install gated. (www.gated.org). Set the gated configuration to import all routes... point it at your upstream (tell them to let you download all routes). Run gated. Wait 20 minutes. >Fix: >Release-Note: >Audit-Trail: From: Joe Greco To: freebsd-gnats-submit@freebsd.org, simon@lok.net Cc: Subject: Re: kern/17868: no more buffers / can't allocate llinfo when running gated Date: Sun, 7 May 2000 07:45:38 -0500 (CDT) GateD has nothing to do with this, fyi. The kernel has a limit on the amount of kernel VM that can be allocated to tables such as these. You may do a "vmstat -m" to see. Look at: Memory statistics by type Type Kern Type InUse MemUse HighUse Limit Requests Limit Limit Size(s) routetbl161703 22113K 22127K 40960K 240489 0 0 16,32,64,128,256 Memory Totals: In Use Free Requests 23618K 23K 255740 You will probably find your "Limit" field to be 20K. It's not real obvious, but this limit is derived from hald of VM_KMEM_SIZE. I compiled a kernel with options "VM_KMEM_SIZE=83886080" and fixed it. Alternatively, one could specify this in boot/loader.rc, as I think someone stuck in a way to alter it in that manner. > (75K routes nowadays) # netstat -rn|wc 78347 471979 5489863 Some internal routes probably present, but > 75K routes. > Run gated. Wait 20 minutes. What speed connection do you have, anyways? I wait maybe 20 seconds. :-) > I looked on newsgroups, and there were some other people reporting > this problem, but no solutions. If this fixes the issue for you, please take the time to help them as I've taken the time to help you. Much appreciated. -- ... Joe ------------------------------------------------------------------------------- Joe Greco - Systems Administrator jgreco@ns.sol.net Solaria Public Access UNIX - Milwaukee, WI 414/342-4847 State-Changed-From-To: open->closed State-Changed-By: phk State-Changed-When: Thu Jun 1 15:44:17 PDT 2000 State-Changed-Why: resolved I pressume. http://www.freebsd.org/cgi/query-pr.cgi?pr=17868 >Unformatted: