Re: [exim] Error relaying large emails over two exim hosts

OPTITAS Info wrote:
> Environment (all servers running Debian etch w exim4 and latest patches applied).
> Both Linux-Servers are running FWBuilder iptable firewalls.
>
> Scenario sending a mail from the WXP PC:
> * WindowsXP-PC w Thunderbird (eMail source)
> -> MailServer1 (gateway, smart router, connected via DSL Router)
> -> MailsServer2 (relay, outgoing eMail) sending emails to the final destination.
> Anything works fine until large eMails (or eMail w large attachments / approx > 1MB) are sent.
> Whenever large email are sent, the connection once is broken and the sending host (Server1) retries sending the emails on next
> queue-run.
>>From the exim log of Server1:
> "R=smart_route T=remote_smtp defer (-46): SMTP error from remote mail server after end of data: host xx.xx.xx.xx [xx.xx.xx.xx]: 421
> adomain.eu SMTP incoming data timeout - closing connection"
>
> I did some investigation a few weeks ago and found out that after a while fragmented packets are sent over the Internet line. I
> tried some "optimized" MTU settings but could not solve the problem. I do not think it's a problem with the MTU.
>
> Regardless of the problems sending large mail from Server1 to the internet, receiving large eMails is not a problem.
>
> Any help appreciated.
>
>

Sounds as if the two LinBoxen involved are under your control.

If so, what happens when you scp or ftp comparable sized files between
them over the same links & via the same firewalls (if any)?

My guess is that it is trouble-free.

I suspect you need to be looking at MUA <-> MSA <-> MTA handshakes and
timeouts more than MTU sizes.

JFWIW, we limit to 800 MB - enough to handle a full CD plus transmittal
notes.

Initially, we did have to adjust time-outs ELSE sender's MUA was unaware
it had suceeded, made multiple successive deliveries.

Attachments of several MB are common, and no trouble.

CAVEATS: *BSD servers, ABM desktops, SeaMonkey MUA.


HTH,

Bill Hacker