From owner-freebsd-chat Sun Aug 27 22:33:26 1995
Return-Path: chat-owner
Received: (from majordom@localhost)
by freefall.FreeBSD.org (8.6.11/8.6.6) id WAA02054
for chat-outgoing; Sun, 27 Aug 1995 22:33:26 -0700
Received: from aries.ibms.sinica.edu.tw ([140.109.40.248])
by freefall.FreeBSD.org (8.6.11/8.6.6) with ESMTP id WAA02046
for <freebsd-chat@freebsd.org>; Sun, 27 Aug 1995 22:33:17 -0700
Received: (from taob@localhost) by aries.ibms.sinica.edu.tw (8.6.11/8.6.9) id NAA10285; Mon, 28 Aug 1995 13:31:39 +0800
Date: Mon, 28 Aug 1995 13:31:39 +0800
From: Brian Tao <taob@gate.sinica.edu.tw>
Message-Id: <199508280531.NAA10285@aries.ibms.sinica.edu.tw>
To: jgreco@brasil.moneng.mei.com
Subject: Re: INN 1.4unoff2 ready for alpha testing
Newsgroups: news.software.nntp
In-Reply-To: <41id66$jel@brasil.moneng.mei.com>
References: <3ujug1$l3f@dodgson.math.psu.edu> <40603a$4km@panix3.panix.com> <418ev0$qn1@eden.adam.com.au> <8768jobn8l.fsf@interbev.mindspring.com>
Organization: Institute of Biomedical Sciences, Academia Sinica
Cc: freebsd-chat@freebsd.org
Sender: chat-owner@freebsd.org
Precedence: bulk
In article <41id66$jel@brasil.moneng.mei.com> you write:
>
>I am running 23179 groups on a 5GB spool + .6GB NOV database quite reliably
>(one "crash" in the last two months and it was probably due to outside
>factors, which reset the system). The system is a 486DX/33 with 32MB of
>RAM, running FreeBSD 2.0.5R. It is tight on memory, taking about 7 hours to
>complete news.daily.
No readers though, right?
>I have just started running 23207 groups on an 8GB spool + 2GB NOV database
>on a Pentium 100 system with 64MB RAM, running a slightly later version of
>FreeBSD (had to pick up a device driver for the customer's twin channel SCSI
>controller, sweet boards). This system is spec'd to handle 50 simultaneous
>active nnrp clients and eight nntplinks without huffing and puffing.
>Reliably.
Okay, cough it up, Joe. What's your secret? :) We've got a P90
with 128 megs of RAM, spool and overviews for 16000 groups split
across three spindles on a Bt-946C controller, running BSD/OS 2.0 with
8 nntplink feeds and peaking around 150 simultaneous in.nnrpd's. It's
dying under that kind of load, our spool is chronically 36 hours
behind and ps reports each in.nnrpd process takes up over 2 megs
virtual and 1600K resident. I'm tempted to ditch the whole thing, buy
an all-new Triton P100 system (the current one is Saturn-based, I
think), drop a couple of NCR controllers in it and run FreeBSD
instead. *sigh*
--
Brian ("Though this be madness, yet there is method in't") Tao
taob@gate.sinica.edu.tw <-- work ........ play --> taob@io.org