From owner-freebsd-current Sat Jun 28 06:08:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id GAA19187 for current-outgoing; Sat, 28 Jun 1997 06:08:53 -0700 (PDT) Received: from aleste.lovett.com (aleste.lovett.com [193.195.45.10]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id GAA19182 for ; Sat, 28 Jun 1997 06:08:47 -0700 (PDT) Received: from aleste.lovett.com [193.195.45.10] by aleste.lovett.com with esmtp (Exim 1.62 #1) id 0whxF5-0003Xz-00; Sat, 28 Jun 1997 14:08:43 +0100 To: freebsd-current@freebsd.org Subject: 3.0-current issues (build 97-06-28) Organization: Demon Internet Ltd. Reply-To: ade@demon.net Date: Sat, 28 Jun 1997 14:08:42 +0100 From: Ade Lovett Message-Id: Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi all, this is my first posting here, so please be gentle :) I'm currently experimenting with a relatively large FreeBSD/SMP box as a dedicated nnrp server, with a remote NFS-mounted spool over FDDI. Machine is a PR440FX motherboard, with dual Intel Pentium-Pro 200MHz processors, 512Mb of main memory, Adaptec 3940 SCSI controller, a couple of DC21040 ethernet cards, and a DEFPA FDDI adapter. With kernel options: maxusers 512 options "MAXMEM=(512*1024)" I'm seeing a number of panic: kmem_malloc: kmem_map too small under even moderate network load (maybe 70 or so nnrp daemons, the machine should theoretically cope with at least 400 of them, with our modified news setup here) Looking back through the mailing list archives, the only references I can find to this type of panic have had suggested solutions about raising the NMBCLUSTERS option to 4096 (or so). Looking at the crash dumps of this machine, I see that nmbclusters is set to 8704 (which ties in with the code in conf/param.c) What am I doing wrong here? I've been beating my head against walls for a couple of days now with no success. -aDe -- Ade Lovett, Demon Internet Ltd.