From: Noel Jones on
On 6/1/2010 9:01 AM, Kaleb Hosie wrote:
> Hey All,
> I have a user trying to send an email however it was bounced back. After
> checking the maillog, I found that it was timing out when sending "RCPT TO".
>
> I telnet'ed into the recipients server and started sending the series of
> commands to send mail and it seemed very quick up until I sent RCPT TO; it
> proceeded to hang there for 50 seconds before responding with a 250 SMTP
> code. I was able to get the email to send by increasing the timeout period
> to 60 seconds for "RCPT TO".
>
> My question is what is the purpose in the delaying that SMTP code? Would it
> be advantageous to implement that into my Postfix configuration?

There is no benefit to adding a delay after RCPT TO. It would
not be helpful -- and likely harmful -- to add a delay at that
point to postfix.

My guess is that the receiving system is doing some sort of
anti-spam analysis during this delay, such as real-time
address verification probes or excessive RBL lookups or maybe
they are checking some dead RBLs that time out. If such tests
take 50 seconds something's broken.

But ultimately, speculation is pointless. You'll need to
check with the postmaster of the remote system to find out why
their RCPT TO has excessive delays.

-- Noel Jones

From: Bastian Blank on
On Tue, Jun 01, 2010 at 10:01:58AM -0400, Kaleb Hosie wrote:
> I telnet'ed into the recipients server and started sending the series of
> commands to send mail and it seemed very quick up until I sent RCPT TO; it
> proceeded to hang there for 50 seconds before responding with a 250 SMTP
> code. I was able to get the email to send by increasing the timeout period
> to 60 seconds for "RCPT TO".

The minimum timeout for RCPT is 5 minutes according to RFC 5321. This
also matches the default postfix config. If you changed this value to
smaller values, you are on your own.

| $ /usr/sbin/postconf -d smtp_rcpt_timeout
| smtp_rcpt_timeout = 300s

Bastian

--
Punishment becomes ineffective after a certain point. Men become insensitive.
-- Eneg, "Patterns of Force", stardate 2534.7