Re: [Exim] Timeout at data close

> Ok, I'm stuck. I've jumped through the hoops. I've waved the rubber chicken.
>
> Symptom: With certain sites, and certain messages hang at end of data.
> I've seen this problem mentioned in the Archives, and have tried
> every fix I've found there.
>
> Status in msglog/
>
> 2003-05-01 15:44:51 SMTP timeout while connected to mail.roadrunner.nf.net [192.75.13.132] after end of data (6535 bytes written): Operation timed out
> 2003-05-01 15:44:51 fannyt@??? R=dnslookup T=remote_smtp defer (60): Operation timed out: SMTP timeout while connected to mail.roadrunner.nf.net [192.75.13.132] after end of data (6535 bytes written)
>
> Situation:
>
> Mail server is named postie. He's running exim-4.14, compiled with
> DELIVER_OUT_BUFFER_SIZE=512
>
> Firewall is win2k running sygate office network for NAT translation.
> The firewall is set to use an MTU of 1400.
>
> Link is bidirectional: Uplink is 28,800 modem connection to Telus.net
> Downlink is DirecPC satellite link. (Which is why the firewall runs windows.
> Sigh.)
>
> The same problem occurs with even greater frequency (more sites)
> with Midpoint firewall.
>
> What do I try next?

Interesting you meantion a firewall because where I work, we had the exact
same problem. At that time, I used 3.36. I couldn't figure out what was
going on.

I got fedup with it and decided to try a dialup to a local ISP. I routed
the mailserver through the dialup and the message that was stuck got
delivered successfully.

If you can try this, I'd recommend trying it.

They won't allow this machine on the net w/o the firewall so I decided that
anything immediately undeliverable would be sent to our ISP's mailserver.
Look in the SMTP transport for the option that does this. I think it's
fallback_hosts.

I'm not sure if it's a TCP/IP bug in the firewall or the remote machine, but
the combination of the 2 trigger this exact same thing.

FYI: We use sonic wall hardware firewall at work.

> I'm including tcpdump -vvv and exim -d logs.

<cut>

--
Lab tests show that use of micro$oft causes cancer in lab animals