prob w/ bayes database.. Mailscanner killed lock ???

[shanna leonard]

hi.

running Mailscanner 4.28.6, on solaris 9

in spam.assassin.prefs.conf I have bayes autolearn enabled

bayes_auto_learn        1
bayes_auto_learn_threshold_nonspam -1.0 #(default: 0.1)
bayes_auto_learn_threshold_spam 17  #(default 12)

I have been using sa-learn to feed spam and ham to my bayes filter
periodically throughout the day.

at one point while I was doing this today, it failed and I got about 4
messages like:

Cannot open bayes databases /root/.spamassassin/bayes_* R/O: tie failed:
Invalid argument

I noticed whilst grepping the mail log file that MailScanner had killed
a lock a few minutes before that

Jun 30 13:02:43 peyote.test.ahsl.arizona.edu MailScanner[25688]: Delete
bayes lockfile for 4222

(while I had been sa-learning another message)

I pretty much have had to kill out and rebuild my bayes database. I am
not sure but suspect the database became corrupt because Mailscanner was
updating it with autolearn.

so I am wondering how I can prevent this happenning again.  will it be
sufficient to turn autolearn off and reload mailscanner before running
sa-learn?  Or is there a way to tell Mailscanner to skip trying to
autolearn if there is a lock? can I prevent mailscanner from killing off
a lock file, and have it just skip the bayesan tests? Is MS unable to
use bayes, if sa-learn is running & has a lock?



--
----
MHO
---
shanna leonard
arizona health sciences library
626-2923
----------------------------------

-------------------------- MailScanner list ----------------------
To leave, send    leave mailscanner    to jiscmail at jiscmail.ac.uk
Before posting, please see the Most Asked Questions at
http://www.mailscanner.biz/maq/     and the archives at
http://www.jiscmail.ac.uk/lists/mailscanner.html