Matt Kettler <[EMAIL PROTECTED]> writes:
>At 06:29 PM 3/24/04 -0600, Kevin Lewis wrote:
>>.   When a spam comes through the normal
>>channel it is getting scanned but it looks as if not all the rules are
>>getting triggered, no bayes and the white-list-from domains are not
>>getting applied.   For example, here's a header of an spam that came
>>through the normal channels.
>
>Kevin, How are you calling SA for your "normal channels"?
>
>Did you pass spamd a -u parameter?
>
>When you ran spamassassin -t what user were you logged in as?
>
>My guess is that you're starting spamd as root, and calling spamc as
>root.. 
>this will force SA to assume the identity of "nobody" prior to scanning
>mail.
>
>Create yourself an ordinary unprivileged account and use the -u parameter 
>to run spamd as that user.. su to that user when you do training and 
>spamassassin -t tests.

I'm starting spamd with these options -d -a -m8 -H -u nobody

The example spamassassin -t example was run as the user nobody.

I have a user account for nobody.

None of these option have changed prior to or since this problem has
started.

Checking in on the system this morning it looks as if everything has start
to magically work again.   Bayes and auto-white-list from are hitting as
before.  The only thing I've done is restart sendmail, milter and spamd
again for the 100th time.....

This leaves me with the question.   Does changing rule sets have an effect
on Bayes?

Geez,

Kevin Lewis
AETN
Dir. of Information Technology

Reply via email to