spf-discuss
[Top] [All Lists]

Re: sendmail-milter question

2004-03-25 14:21:07
Paul Howarth wrote:

Graham Murray wrote:

Please can it also fix the problem with (for example) Sourceforge
probes which send an RSET between RCPT to the sender and doing a
postmaster check? SPF-milter 1.3 complains that it needs a HELO, but
the sendmail Privacy options requiring HELO do not complain when the
milter is not running. RFC2821 4.1.1.5 states that issuing an RSET is
more efficient than resending the EHLO, so (as I read the RFC and by
existing sendmail behaviour) the milter should not be complaining
that it needs another HELO after an RSET has been sent.

Are you sure this is spf-milter? Are you running any other milters?
This problem is supposed to have been fixed in version 1.30 and in
fact it *did* fix it for me as I was having a similar problem with a
callback scheme at a UK university.

spf-milter 1.30 has an abort_callback procedure that handles RSET.
The example milter supplied with Sendmail::Milter, upon which many
milters are built, does not have this procedure and may result in
problems with callbacks such as you are experiencing.

Exactly; spf-milter no longer requires a new HELO after a RSET. From the
spf-milter annotation:

# On RSET, forget everything except the HELO name. Noted by Paul Howarth
#
# (note by me: we also need to preserve the hostname of the sender,
# our own hostname, and the IP address of the sender! Best, therefore, to
# use a negative logic, and just delete the things that need to go)

Cheers,

- Mark

        System Administrator Asarian-host.org

---
"If you were supposed to understand it,
we wouldn't call it code." - FedEx


<Prev in Thread] Current Thread [Next in Thread>