On Thu, 1 Jun 2006 11:24:00 -0400, Theo Van Dinter <[EMAIL PROTECTED]> wrote:
>On Thu, Jun 01, 2006 at 11:16:13AM -0400, Patrick Sherrill wrote: >> We run three sendmail servers all w/spamassassin amavis-new, clamav and >> mcaffee. We've been using sa-learn to add to the Bayesian database. Is >> there a preferred method of synchronizing the Bayesian db among the three >> servers? > >Generally you can't sync DBs between servers using the file methods. You can >switch to using one of the SQL back-ends (MySQL, etc,) and then access that DB >from your servers. If you find way of load balancing the MySQL db's I'd be very interested in hearing how. To date the only semi-viable option I've seen is screwing with the DBI. We run 5 SA servers here and a single MySQL db - which is still a point of contention since it's a potential single point of failure. Trialing load balanced MySQL has proven unworkable here. I'm still very interested in fixing this single point failure. KF Nigel