You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Rocky Olsen <ro...@mindphone.org> on 2005/05/19 23:15:49 UTC

Bayes_journal and locking

on our mailcluster we have spamassassin running with bayes turned on, but
auto learning turned off, we then hand train the bayesdb and push it out to
all the machines.

However weven with bayes_auto_learn turned off, spamd is still creating the
bayes_journal files, which isn't a problem until spamd decides to flush the
journal file to no where it still locks the bayesdb file causing concurrent
spamd processes to be unable to use the db resulting in inconsistent scores
every few seconds.  The quick and dirty solution was to create a +i
bayes_journal file, spamd never writes to the journal, so it never tries to
flush it...except it now throws an error everytime it tries to write to
bayes_journal.


Just wondering if anyone out there has run into this problem before? Is
there a more elegant solution?

Here is the pertinent section of local.cf:

use_bayes							1
bayes_path							/etc/mail/bayes/bayes
bayes_auto_learn					0
bayes_auto_expire					0
bayes_use_hapaxes					1
bayes_use_chi2_combining			1
bayes_learn_to_journal				0


Thx in advance for any help.

-Rocky

-- 
______________________________________________________________________


what's with today, today?

Email:	rocky@mindphone.org
PGP:	http://rocky.mindphone.org/rocky_mindphone.org.gpg