Hi Claudio!
Thanks for clarification.
In my opinion spamchecks should be done for incoming Mails only.
But what I try to set up in detail:
I'm running simscan-cvs on qmail-ldap (latest patch) and I want to have
spamassassin with per-user preference files, per-user bayes and AWL.
I start spamd this way:
"exec /usr/sbin/spamd -x -s stderr ...
--virtual-config-dir=/home/spamassassin-prefs/%l/spamassassin/ ...
The virtual-config-directory is named with the 'local part' of the
current users email-address (%l), an this directory is being created if
it does not exist (no chance to change that).
The funny thing is, that if I send an (outgoing) message (after
smtp-auth) to [EMAIL PROTECTED] the spamassassin log tells me: "Using
default config for [EMAIL PROTECTED]:
/home/spamassassin-prefs/test/spamassassin//user_prefs" and the
virtual-config-dir gets created automatically.
That would be lots of directories after a short time.
I compiled simscan with --enable-spamc-user=y and really every
email-address of the first "rcpt to" is sent to spamassassin. (no
difference if it's local or not).
Sure I could have maildrop running spamassassin, but that's not my first
choice.
Thanks
Manfred
Claudio Jeker wrote:
On Sat, Sep 24, 2005 at 11:18:09PM +0200, Manfred Langthaller wrote:
Hi!
I set up POPBEFORESMTP for testing purposes.
Everything works fine, env RELAYCLIENT seems to be set, no spamchecks
are done.
My conclusion so far:
smtp-auth does not set RELAYCLIENT at all, right?
auth-smtp does not set the RELAYCLIENT env var. But sets a similar flag in
qmail-smtpd. It would be possible to tell qmail-smtpd to set RELAYCLIENT
after a succesful authentication. Until now I don't see the need for it.
