On 11/24/2021, Michael Peddemors via mailop wrote:
CONN: 40.107.96.87 -> 25 GeoIP = [US] PTR =
mail-sn1anam02on2087.outbound.protection.outlook.com OS = Windows NT kernel
Returning 250 ok [qp 3539411] for data
QUIT command received, args:
And then it terminates the connection, SSL collapses, without waiting
for the remote mail server to acknowledge the QUIT.
I get it that they 'might' think that closing when 'they are done' and
we gave a 250ok on the DATA, and they sent the QUIT.. I don't know,
thinking to shave a few milliseconds off of the connection, but the RFC
is pretty clear that a QUIT .. AND .. and acknowlegement is part of RFC.
Comments? (And no, there is zero lag before they terminate)
If I understand you correctly;
When sending mail, *.outbound.protection.outlook.com appears to send
DATA, terminate via <CRLF>.<CRLF>, wait for a response, then issue QUIT
immediately followed by a TCP FIN packet.
Operationally, I don't think it makes a difference. RFC pedantics?
Maybe. Does the receiving system advertise the Pipelining SMTP Service
Extension?.. Does setting/un-setting that response to EHLO make a
difference to outlook.com's behavior? If so, does and/or should RFC2920
allow for QUIT->TCP FIN?
_______________________________________________
mailop mailing list
mailop@mailop.org
https://list.mailop.org/listinfo/mailop
