Bayesian SpamDB Rebuild FAQ

From ASSPSMTP

Recommended ASSP settings

In the web interface under "Other Settings", check the values for for MaxFiles & MaxBytes

• MaxFiles should be set to 14000
• MaxBytes should be set to 4000

If your values are larger you should change them to the recommended settings above.

Note: In version 1.3.3.8 and later, these options are located in the "Collecting" section of the web interface.

Bayesian Normality/norm

See Bayesian normality for a detailed explanation.

Server Hardware

• How much memory does my server need?

The recommended amount of memory for a small to medium install is 512MB.

For large installs 1GB is recommended, especially if the MTA is running on the same hardware.

• How much free memory does my server need to rebuild the database?

The rebuildspamdb.pl script usually needs 50MB+ for 5000 spam/ham messages using the above settings.

Servers with a full corpus of 18000 files will usually require from 230MB up to 330MB of free memory.

• How fast should my servers processor be?

1.2Ghz is the minimum recommended for medium volume servers.

• How much cpu is used when running the rebuild script?

The rebuild process is highly CPU intensive, usage from 80-100% is not uncommon.

• How long does it take to rebuild the database?

On the following specs with 18000 spam and 16800 ham messages it takes 15 minutes to rebuild the database.

• 2.0 GHz AMD Athlon 64 3000+, 512MB memory.
• The OS is Server 2003 Standard running inside VMware.