I am a bit lost. Here is my basic local.cf file: required_hits 5 skip_rbl_checks 0 use_bayes 1 bayes_file_mode 0700 bayes_path /var/spool/bayes/bayes
Am I missing anything here. I want to try bayes and I am still feeding sa-learn spam and ham. My /var/spool/bayes/ directory is being updated each night when I run sa-learn and sa-learn --dump outputs lots of fun stuff which make me believe that it's working correctly. My spamassassin --lint -D output seems to be looking in /Users/admin/.spamassassin for pref files although I am not running SA as this user (I am running it as the mta user that my mail server uses). Dan > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of > David B Funk > Sent: Thursday, December 04, 2003 2:59 PM > To: Pete Henshall > Cc: [EMAIL PROTECTED]; [EMAIL PROTECTED] > Subject: RE: [SAtalk] spamds that don't finish > > > On Thu, 4 Dec 2003, Pete Henshall wrote: > > > Hi dan, list, > > > > > I think it's simply a function of load. The first system gets the bulk of > > the mail thoughput. You can see that the > erratic loads > > > tail off over the weekend. It's wierd. I have tried disabling RBL, bayes > > and even removing all my third party > > > rules. No dice. > > > > If it is still leaving spamds lying around with bayes disabled then I don't > > know.... I have just set bayes_learn_to_journal 0 (thanks David Funk) and my > > problem seems to have stopped.... maybe. > > I'm sorry if I gave you the wrong impression, if you are using Bayes with > auto_learn (auto_learn 1), then you most likely -do- want > bayes_learn_to_journal set to 1. (enabled). > > If you use auto_learn and disable journaling, then each spamd tries to > update the Bayes database with each new message (thus increasing the > probablilty of lock contention problems). > > If you enable journaling then each spamd just appends to the end of the > journal file (no locking needed for a simple text append). Then the > database will perocially get rebuilt and incorporated in the database. > So only that occasional rebuild needs to lock the database. > > > > > As far as I am concerned spamd should NEVER have rouge spamd's coming off it > > that don't have a matching spamc. (is that right??) > > I'm not so sure about this. If you have bayes_learn_to_journal enabled > then a spamd child will need to be run when ever the journal file gets > "full" (size > bayes_journal_max_size) or it's been around for more > than one day. Also, unless you've explicitly disabled it, a db expire > is done daily (which would be another spamd child). > > So unless you disable all automatic Bayes maintanence operations > (learn, expire, etc), then there will be the possibility of spamd > children and potential lock contention. > > Dave > > -- > Dave Funk University of Iowa > <dbfunk (at) engineering.uiowa.edu> College of Engineering > 319/335-5751 FAX: 319/384-0549 1256 Seamans Center > Sys_admin/Postmaster/cell_admin Iowa City, IA 52242-1527 > #include <std_disclaimer.h> > Better is not better, 'standard' is better. B{ > ------------------------------------------------------- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
