You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Dennis Clark <De...@patten.edu> on 2006/05/19 21:41:52 UTC

autolearn=failed

Using FC5, SA 3.1.0, calling SA with spampd. 
Every message that meets the autolearn threshold (spaminess>~30 <1)
results in an autolearn=failed result.  Checked permissions and made
sure bayesian and whitelist were r/w for user mail.  Log shows locking
errors on whitelist.  using  -D --lint reports all good.

Re: autolearn=failed

Posted by Matt Kettler <mk...@evi-inc.com>.
Dennis Clark wrote:
> Using FC5, SA 3.1.0, calling SA with spampd.
>  
> Every message that meets the autolearn threshold (spaminess>~30
> <1) results in an autolearn=failed result.  Checked permissions and made
> sure bayesian and whitelist were r/w for user mail.  Log shows locking
> errors on whitelist.  

1) Are you calling spamc as a non-root user with a real login?
(note: if you call from /etc/procmailrc without DROPPRIVS, this will get run as
"root" or "mail" so your answer is no)

2) Are you passing -u to spamd or spamc?

3) Do you have a bayes_path AND bayes_file_mode?

if no to all 3, then spamd will be attempting to access bayes in a user with no
r/w home dir. (note that if spamd goes to scan mail and finds it is running as
root, it will immediately setuid to "nobody" to prevent any security breaches).

> using  -D --lint reports all good.

What user did you do this as? Is it the same user from 1)?