waiting for the final acknowledgement. The recommended timeout here
according to RFC 2821 is 10 minutes, so adding under a second of
content-scanning shouldn't be a problem.
The problem is that this phase is the one vulnerable to race conditions:
If the connection breaks between accepting the message and getting the
response through to the sender, a dupe will be created. So it's
desireable to keep that period of time as short as possible.
10 minutes is many eternities for any scanning that does not involve
asking outside systems, and is several eternities even for some
outside checking.
Have there been any reports of increased dups associted with the use
of Sendmail milter filters other than the DCC? I've certainly heard
of nothing like that from users of DCC clients.
Vernon Schryver vjs(_at_)rhyolite(_dot_)com
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg