[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: bayesian spamassassin throwing error



begin Nori Heikkinen quote from Sat, Oct 11, 2003 at 08:19:39AM -0400
> hey all,
> 
> recently, i've been noticing that the Bayesian filter in spamassassin
> has been acting up.  whenever i try to pipe a message to sa-learn
> --ham or something, it tells me:
> 
> Cannot open bayes_path /home/nori/.spamassassin/bayes R/O: 
> Cannot open bayes_path /home/nori/.spamassassin/bayes R/W: File exists
> Learned from 0 messages.
> Cannot open bayes_path /home/nori/.spamassassin/bayes R/W: File exists
> Cannot open bayes_path /home/nori/.spamassassin/bayes R/W: File exists
> Cannot open bayes_path /home/nori/.spamassassin/bayes R/O: 
> Cannot open bayes_path /home/nori/.spamassassin/bayes R/W: File exists
> 
> also, my .procmail/log is full of lines like this -- i assume it's
> trying to auto-learn, as most of my spam that comes in is so far above
> the threshold that it would try to auto-learn it.
> 
> has anyone else seen this behavior?  i should mention that there's no
> file called simply "bayes" in ~/.spamassassin -- all that's in that
> dir are
> 
> bayes_journal  bayes_msgcount  bayes_seen  bayes_toks  user_prefs

This is all I have, too.  What do the permissions on them all look like?
Nothing in the BTS? Does feeding mail to "spamassassin" itself work?
How about to sa-learn?  Is your homedir mounted over NFS?  BerkeleyDB
(the db library used for those files) has "issues" with NFS.  If all
else fails, and you can reproduce this with sa-learn or spamassassin try
strace'ing it to see exactly where it's getting confused.  Oh, and try
enabling the debug option for sa-learn, perhaps that will show you
something interesting.

-- 
Rob Weir <rweir@ertius.org> | mlspam@ertius.org  |  Do I look like I want a CC?
Words of the day:  Axis Of Evil Ft. Bragg morse Serbian red noise colonel spies
Hi, VeriSign!                          bob@715025659f1c5f150829e6e9f6001796.com

Attachment: signature.asc
Description: Digital signature


Reply to: