SUMMARY: Incoming Timeouts

From: Mike Peppard <>
Date: Tue Jun 12 2001 - 13:55:46 EDT
Thanks in no particular order to:
Dan Anderson
Sergio Gelato
Martin Akesson
Khann Tran

Moral of the story is to use 192.168 when you create an
internal network.  Otherwise someone will have to eventually
change your TNS, mail, DNS, NT, AAARRRGGG, right down to my
fancy ghosted images.  Use route add as a Band-Aid.

Stop here... story follows:

For everyone who said the nameserver was called  (Hanging my head) Yes. I set it up too.

For those who said the nameserver was messed up...
nope it's fine.  When the world becomes a warm fuzzy
place I'll open my server for you to use my bandwidth,
but for now I'll stick to tinydns and ipchains.  Maybe
someday I'll put bind back out there.  Maybe not.

What confused me about the problem was that the
client's server was timing out occasionally (>150Ms)
and the messages that were forwarded to me said that the
sales people at the client side weren't able to attach
to their INTERNAL mail server, not mine!  Several different
problems entirely confusing the issue.

Anyway, Khann helped me snoop with the clients server and
I really looked at the IP of the client mailserver for the
first time.  I'd inherited a class B internal network that
is NOT the range.  The client was in my internal
IP range (they own practically the whole range of IPs as my
internal network) so I was treating their traffic as internal
and dropping it.

A few route adds out to the external gateway and I'm fine.
Now to tackle the root problem... aarrrggggg.  First the
gaping security holes and kiddy hackers, now this.
HEY! I'm supposed to be useless management here.

Thanks again Khann!


>I've been hearing that one specific client of ours can't send email to
>our server.  They have been timing out or having an interrupted system
>call.  We send fine to them.  It's been going on for a while apparently.
>Has anyone an idea where to start?  If you do I will forward the delayed
>message from their server and anything else you ask for... You'll
>understand when you get their IP address.
>My MX/gateway is a Sparc station running 2.6 (sendmail as the relay.) Top
>put my MX/gateway at 98.3% idle, my MX/gateway had a throughput just now of
>154K/second from my desk through to the Internet.  I couldn't find anything
>in my gateway logs for the remote IP or their MX to indicate an issue or
>even that they hit the MX.  The MX/gateway log shows one person reading the
>newyorker on the web at the time of the interrupted system call.  Not
>something that would stress my MX/gateway.
>One interesting note is that the MX that had the problem is listed fairly
>low (15) in their NS records.  I'd appreciate any assistance.
>If you wanted to send a test through here.
>My DNS is
>My MX is
Received on Tue Jun 12 18:55:46 2001

This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:24:56 EDT