Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 23 Aug 1998 22:44:43 -0400
From:      "Gessner, Matt" <mattg@aiinet.com>
To:        "'hackers'" <freebsd-hackers@FreeBSD.ORG>
Subject:   Q was on questions, now putting it here.
Message-ID:  <7283DE19D141D111AD0E00A0C95B1955CF12A2@mail2.aiinet.com>

next in thread | raw e-mail | index | archive | help
Hi, all,

I posted to comp.unix.bsd.freebsd.misc but nothing's come back... news takes
a while.

I'm running 2.2.7-RELEASE, and I've already changed my kernel a bit to
allow me to do a whole bunch of simultaneous connections, but I've
probably done something wrong, because it ain't working.

Here's what I have:

P5-166 Gateway
32 MB RAM
2 GB disk, 850MB belongs to FreeBSD, of which 100MB is swap.
PCI DEC Ethernet card

Here's how I've mangled my poor kernel:
    maxusers 128
    options CHILD_MAX=256
    options OPEN_MAX=2048
    options NMBCLUSTERS=4096

Here's how I've changed login.conf

    default and root both have unlimited child process and open file
counts.

I've done a little poking around in param.c and don't see immediately
that I've violated any rules by setting these limits the way I have.

When I type limits, I get the following info back:

    maxprocesses 2067
    openfiles 4136

But I can't find anything anywhere that talks about tuning the kernel
for doing this.

What I need to do is be able to run lots of TCP connections, and I know
it can be done with FreeBSD. 

The connections are being made from FreeBSD to a proprietary board and
then to a Solaris machine.

My end goal is to be able to bring up about 1024 TCP connections
outbound and route them back to the same box, for a total of 2048
connections.

Right now, I run 1024 connections to the other system, and what happens
is when the sockets are all connected, and the writes start to occur,
FreeBSD
just reboots the machine -- no messages anywhere.

Have I really goofed up somewhere/something?  Can anyone give me some
insight as to what MIGHT be happening here?  E-mail is fine, so's the
newgroup.

I suppose it's completely possible I'm running out of RAM and swap, but
I'd expect to see something in /var/log/messages to that effect.  So
far, nothing.

Recently I uppped NMBCLUSTERS to 4096, but it didn't seem to fix the
problem.
That was based on some notes in the FAQ.

Even more recently (but haven't tested it yet) is that I upped my swap space
to
256MB for 32MB of RAM.  I killed off that nasty Win95 virus that was
consuming all that disk space.... Nasty, nasty thing.

Thanks in advance,

    Matt Gessner





To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?7283DE19D141D111AD0E00A0C95B1955CF12A2>