Robert Lopez:
> >> "reject_unauth_pipelining
> >> ? ? Reject the request when the client sends SMTP commands ahead of
> >> ? ? time WHERE IT IS NOT ALLOWED, or when the client sends SMTP
> >> ? ? commands ahead of time WITHOUT KNOWING THAT POSTFIX ACTUALLY
> >> ? ? SUPPORTS ESMTP COMMAND PIPELINING.
> >
> > a) the system sends commands together where it is not allowed by
> > RFC 2920, even after prior negotiation, or b) the system sends
> > commands together without prior negotiation as per RFC 2920.
> >
> > ? ? ? ?Wietse
> >
>
> Thank you. Prior to reading RFC 2920 I was assuming that pipelining was
> a bad thing done by spammers. I never comprehended it could be a good thing
> if managed by both ends correctly.
>
> At a web meeting today I was told they will use 40 concurrent connections.
> With the default max connection limit (given no other server resource limits)
> I suppose that is not blasting an insane number of SMTP connections.
>
> Would this situation be better if I moved reject_unauth_pipelining from
> smtpd_client_restrictions to smtpd_data_restrictions, taking it out
> completely,
> or leaving it as it is?
A properly-implemented sender will implement SMTP command pipelining
in a manner that respects RFC 2920.
I highly recommend that you run some tests before you make any
promises that the device will "work" with your mail system.
I think that many sites would object to 40 simultaneous connections
without some prior agreement.
Wietse
