OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: [2] how to bind smtp client to specific address?
From: Mailing List Account (mlistsohoweb.net)
Date: Thu Mar 30 2000 - 16:40:34 CST

Sorry for submitting to the list -- the receiver's mail host uses ORBS, so
I can't get mail to him directly. Not going to get into the whole issue of
ORBS abusive practices, I'll just handle it by posting to the list...

At 11:22 AM 3/30/00, you wrote:
>Uh, "Mailing List Account"! What a funny name! :)

It's just so I can keep my list-related traffic separate from my other
traffic. I get hundreds of list mails a day, and need to be able to not
pick them up if I'm in a hurry, or accessing remotely ;-)

>LaTeX source file is readable as plain text -- not very perfect, but still.
>Converting this into something more usable is a my dream :) -- I'm now working
>on this on my (very limited) spare time...

Just didn't go deep enough into the file, then....I hit all the
binary-looking stuff in the top of the .ps file and presumed it took LaTex
to even interpret it at all.

>Oops, I can guess your config! And maybe add some more suggestions
>(but still can't see the points of having all those hostXX addresses).
>
>So:
> eth0 208.184.6.8/255.255.255.128 (or 208.184.6.8/25)
> eth0:0 208.184.6.54/255.255.255.0 (or 208.184.6.54/24)
>
> > Destination Gateway Genmask Flags Metric Ref Use
> Iface
> > 208.184.6.8 * 255.255.255.255 UH 0 0 0
> eth0
> > 208.184.6.0 * 255.255.255.128 U 0 0 0
> eth0
> > 208.184.6.0 * 255.255.255.0 U 0 0 0
> eth0
> > 127.0.0.0 * 255.0.0.0 U 0 0 0 lo
> > default 208.184.6.gate. 0.0.0.0 UG 0 0 0
> eth0
>Note that you have two entries for 208.184.6.0 that is useless.

Both my Redhat boxes do that. I'm sure it has something to do with setting
up the aliases initially using Linuxconf.

>What is 208.184.6.gate.? Is it 208.184.6.1?

It is.

>And -- what address you want to use for outgoing connections -- 6.8 or
>6.54 (or 6.70 ;)?
>(Thinking that .5. in 208.184.5.54-70 was a typo).

It was. Should be 208.184.6.54 to 70. (The typo was a side effect of 6 days
in a row with 4 hours of sleep a night)

>Here, you can do:
>- setup "primary" address (that you want to use for out conns) on eth0
>(either with
> ifconfig or ip) with netmask /24 (255.255.255.0).
>- configure default route to 208.184.6.1 (route add default gw 208.184.6.1, or
> ip route add default via 208.184.6.1)
>- setup all the other "secondaries" with netmask /32 (all-ones,
>255.255.255.255),
> again, does not matter -- ifconfig/ip addr
> (or, use the same netmask as "main" address, i.e. /24)
>
>Guess that those strange addresses are for virtual stuff?
>Maybe better to use one whole subnet (say, 32 or 64 hosts) for this?
>(I.e. 32..63, 64..127 etc)?

Well, there's a story behind that having to do with how I was allocated IPs
by my colo center. Basically, I have a 64-IP range, plus a few. (full range
is .7 - .70) Everything below .54 is allocated to another machine right now.

>But with your situation, all programs that ise INADDR_ANY (theoretically)
>should be given 208.184.6.8 source address.

I can see that. My understanding was that if I got the IP stack
communicating correctly (I presumed by using iproute and the "src"
parameter) that postfix would send mail off the virtual host on .54 using
.54 instead of .8 -- that's what I want ;-)

Best,
Scott Temaat, President/CEO
SOHOWeb Technologies, LLC (http://www.sohoweb.net)
Sales/Admin: 888-421-SOHO Customer Service: 888-421-7981

Coming Soon: Web Presence services custom-fitted to your small business needs
and budget. No more one size fits all! Details at http://www.sohoweb.net