On Fri, Nov 26, 2010 at 09:59:08PM -0800, Michael J Wise wrote:
> That having been said....
>
> On Nov 26, 2010, at 1:36 PM, Terry Gilsenan wrote:
Yes, Terry was posturing. No need to take it too seriously. Putting the
posturing aside, there is an underlying interoperability issue. The new
software behind mail.global.frontbridge.com does not correctly implement
RFC 2920. Only the first command in a group is processed when the group
ends in QUIT.
With a bit of luck (and likely some time) this will be fixed. I am
reposting what it takes to reproduce the problem. That's the substance
of this thread, the rest is idle chatter...
RFC 2920 Failure:
Client IP and envelope sender/recipient masked as they are not germane.
S> 220 VA3EHSMHS012.bigfish.com Microsoft ESMTP MAIL Service ready at Sat,
27 Nov 2010 06:47:57 +0000
C> EHLO amnesiac.example.com
S> 250-VA3EHSMHS012.bigfish.com Hello [192.0.2.1]
S> 250-SIZE 157286400
S> 250-PIPELINING
S> 250-ENHANCEDSTATUSCODES
S> 250-STARTTLS
S> 250-AUTH
S> 250-8BITMIME
S> 250-BINARYMIME
S> 250 CHUNKING
C> MAIL FROM:<sen...@example.com>
C> RCPT TO:<recipi...@example.com>
C> RSET
C> QUIT
S> 250 2.1.0 Sender OK
S> 221 2.0.0 Service closing transmission channel
<client sees premature EOF>
Delaying the QUIT results in an RFC-compliant interaction:
S> 220 VA3EHSMHS012.bigfish.com Microsoft ESMTP MAIL Service ready at Sat,
27 Nov 2010 06:47:57 +0000
C> EHLO amnesiac.example.com
S> 250-VA3EHSMHS012.bigfish.com Hello [192.0.2.1]
S> 250-SIZE 157286400
S> 250-PIPELINING
S> 250-ENHANCEDSTATUSCODES
S> 250-STARTTLS
S> 250-AUTH
S> 250-8BITMIME
S> 250-BINARYMIME
S> 250 CHUNKING
C> MAIL FROM:<sen...@example.com>
C> RCPT TO:<recipi...@example.com>
C> RSET
S> 250 2.1.0 Sender OK
S> 250 2.1.5 Recipient OK
S> 250 2.0.0 Resetting
C> QUIT
S> 221 2.0.0 Service closing transmission channel
Because Postfix does not typically wait for a "QUIT" response, it would
cost Postfix little to delay "QUIT", but this would tie up resources on
the receiving server, which is now tied up for another RTT.
So the pipelined QUIT is Postfix being a good citizen, and it would be
best if the problem were fixed at the server which advertises RFC 2920
compliance, but does not fully implement the specification.
--
Viktor.
