You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Robert S <ro...@gmail.com> on 2006/12/01 22:38:25 UTC

bayes: expire_old_tokens: child processing timeout at /usr/sbin/spamd line 1086

There has been some correspondence on this matter recently but I'm
still having problems.  I'm running SA 3.1.3 from debian backports on
an AMD K6.  I'm running the spamd daemon and launching spamc from
procmail.  I've been getting the following message:

spamd[3775]: bayes: expire_old_tokens: child processing timeout at
/usr/sbin/spamd line 1086.

After a while SA seems to pack up completely and spammy messages stop
getting filtered.

I tried to fix up this by creating a daily cron job that runs the following:

sa-learn --force-expire --sync

I still notice that I am getting the message towards the end of the 24
hour period before the cron job is run again.

Is it reasonable to set up a cron job that will run "sa-learn" more
frequently than every 24 hours (eg 6 hourly), or is there another
solution to this (short of upgrading my ancient hardware)?

Re: bayes: expire_old_tokens: child processing timeout at /usr/sbin/spamd line 1086

Posted by Robert S <ro...@gmail.com>.
> Sure.  Run it as often as needed.  It may block bayes access while it is
> running, so if you have a really busy system (and it sounds like you do) you
> want to run it often enough to keep the processing time for each shot down
> to something reasonable.

Strange thing is that its not a very busy system.  Its a small
business mailserver with about half a dozen users (most of whom don't
get much spam - we get about 30 spams per day).  I use FuzzyOcrPlugin,
which probably slows things down a fair bit.

Re: bayes: expire_old_tokens: child processing timeout at /usr/sbin/spamd line 1086

Posted by Loren Wilton <lw...@earthlink.net>.
> Is it reasonable to set up a cron job that will run "sa-learn" more
> frequently than every 24 hours (eg 6 hourly), or is there another
> solution to this (short of upgrading my ancient hardware)?

Sure.  Run it as often as needed.  It may block bayes access while it is 
running, so if you have a really busy system (and it sounds like you do) you 
want to run it often enough to keep the processing time for each shot down 
to something reasonable.

        Loren