Reinaldo de Carvalho:
> On Wed, Sep 23, 2009 at 4:29 PM, Wietse Venema <wie...@porcupine.org> wrote:
> > Reinaldo de Carvalho:
> >> Hi,
> >>
> >> Someone can tell me why EHLO reply send 2 packets?
> >>
> >> ~# postconf  mail_version
> >> mail_version = 2.5.5
> >>
> >>
> >> ## tcpdump ##
> >>
> >> ^[OM^[OM15:41:23.546763 IP 10.15.10.9.49898 > 10.15.1.65.25: P 1:10(9)
> >> ack 34 win 46 <nop,nop,timestamp 927636169 259345475>
> >> e.....@.@.M.
> >> .
> >>
> >> ..A....`.[TU.s............
> >> 7J...uLCEHLO myhost.example.com
> >
> > Above is "EHLO myhost.example.com" from the SMTP client to the SMTP server.
> >
> >> 15:41:23.546924 IP 10.15.1.65.25 > 10.15.10.9.49898: . ack 10 win 91
> >> <nop,nop,timestamp 259346474 927636169>
> >> e.....@.?..9
> >> ..A
> >> .
> >>         ....U.s.`.[]...[.......
> >> .uP*7J..
> >
> > Above is a TCP acknowledgment from server kernel to client kernel.
> >
> 
> Verify "MAIL FROM packet" from my the first, or any other SMTP stage,
> postfix reply only one packet.

TCP Acknowledgments are generated by the KERNEL TCP/IP implementation.
They are TCP packets without data content.

In some cases the KERNEL TCP/IP implementation will not send an
ACK, for example when it knows that it will send a real data packet.
Delayed ACKs and a lot more are discussed in Internet RFCs.

This list discusses TCP/IP behavior only when it prevents the
delivery of email.  I don't think that is the issue here. For other
questions about KERNEL TCP/IP implementations I suggest that you
visit the corresponding mailing list.

        Wietse

Reply via email to