[Mimedefang] Truncated connections causing multiple resends onincoming mail

[Kimmo Jaskari]

On Fri, 2007-05-04 at 08:43 -0400, David F. Skoll wrote:
> Kimmo Jaskari wrote:
> 
> > I did some testing with a mail client and tried to send mail directly
> > through the mail server to a recipient; the mail client had no problem
> > uploading 100% of the data to the server, but at that point it just
> > waited for about two minutes before giving an error about how the mail
> > server wasn't responding.
> 
> That's the dreaded post-DATA race condition.  RFC 2821 says a client SHOULD
> wait 10 minutes for a response to the final ".".  However, if indeed the
> problem is being caused by a client timeout, the only way you'll fix it is
> to have your server respond more quickly.  If Clam is taking too long to scan,
> you are probably CPU-bound.  I think the only solution to this problem
> is more hardware. :-(

Does this then mean that anything that isn't scanned and processed by
the server in about 2 minutes or less causes an error? What am I missing
here? 

That would imply that I'd need "Gods own server" if someone decided to
send me 20 separate mails that hovered on the size limit I enforce (at
the moment 20MB); at the moment, if I send two mails with 6MB
attachments they bump up against the timeout and "fail" but get sent on
anyway, but if there is nothing else going on on the server sometimes
one solitary 6MB attachment may squeak in under the time limit and the
client gets informed that the mail got sent.

The weird thing is that the same hardware on an earlier OS version,
earlier MD version and scanning with uvscan commandline as opposed to
Clamd apparently didn't have these issues. Granted we're only talking
about a 440mhz UltraSPARC IIi here with half a gig of memory that I
probably should retire, but even so - shouldn't that be able to handle
scanning a couple of measly 6MB attachments without choking?

/Kimmo