At 11:50 23.10.04 -0400, you wrote:
>On Oct 23, 2004, at 11:22 AM, Jeremy Kitchen wrote:
>> On Fri, 2004-10-22 at 18:57 -0300, Walter Souto R. Junior wrote:
>>> Thanks Tom and Jeremy,
>>> I do fix my run file for smtp, but now I have a "501 malformed auth
>>> (#5.5.4)" using telnet. I'm also trying with Opera with plain, auth
>>> cram-md5 without success. My run file looks like:
>>> LOCAL=`head -1 /var/qmail/control/me`
>>> QMAILDUID=`id -u qmaild`
>>> NOFILESGID=`id -g qmaild`
>> Can the qmaild user read your vpopmail information? I certainly hope
>> not. This is why you are unabled to authenticate. Now, go fix this
>> disable your open relay.
>I agree with jeremy. I took the patch that vpopmail provides in its
>contrib dir, and made it one big rolled in one patch file, instead of
>seperate files. This patch requires no hostname in the qmail-smtpd run
>in your qmail source dir just do the following:
>(It is used in a guide of mine, but it is for stock qmail)
>patch < qmail-smtpd-auth.patch
This is a very old story.
Pls. visit: http://www.fehcom.de/qmail/smtpauth.html
Essentially, with all the disturbed discussion, I'ld like to conclude:
- Putting the hostname in the call of qmail-smtpd for SMTP Authentication
conforms to the use of qmail-popup/qmail-pop3f, however
- forgetting the hostname in there leaves the SMTP server "open", wheres
- dropping the requirement for the hostname yields (in worst condition) an
error message to the client.
Thus, my implementation of the SMTP Authentication for Qmail (apart from
the many bug-fixes) is fail save. As outlined, technically there is *NO*
need to include a hostname in the call of the PAM; even not for qmail-popup
- it's simply historic.
Dr. Erwin Hoffmann | FEHCom | http://www.fehcom.de/
Wiener Weg 8, 50858 Cologne | T: +49 221 484 4923 | F: ...24