This article details how to configure how SecurityGateway scores messages coming into and out of the server, according to the Apace SpamAssassin engine, to determine if they are spam, and the Bayesian-learning database.
Use heuristic rules and Bayesian classification to analyze messages
This option determines whether SecurityGateway will do any processing of messages for possible spam content using the SpamAssassin engine and the Bayesian-classification system. Disabling this will make all other options under Configuration and Exclusions disallowed from changing. This option is enabled by default.
Add score returned by SpamAssassin to message score
This option will take whatever score is determined by SpamAssassin for the message and add it to its general score, which determines if SecurityGateway quarantines the message or allows it to be sent onto the domain email server. By default, this is enabled.
Reject message if SpamAssassin score greater or equal to...
This option will cause SecurityGateway to reject a message during the incoming SMTP session if it scores equal or higher than the listed value, which is 12.0 by default, and not do further testing on message. This is disabled by default, with the score from the message added to the general score used by SecurityGateway.
Quarantine message if SpamAssassin score greater or equal to...
Similar to the one above, this option will automatically quarantine the message if it scores equal to or higher than the listed amount, which is 5.0 by default, and not do further testing on message. This is also disabled by default.
Exclude messages larger than ... KB
This option will cause SecurityGateway to skip over any message that is larger than the target amount, by default 100 KB. As most spam messages are small in size to aid in quick delivery, it can be assumed that larger files should not be spam.
Exclude messages from whitelisted senders
If a sending address is on either the server whitelist, or the whitelist of the recipient, it will skip being scanned by SpamAssassin. This option is enabled by default.
Exclude messages from authenticated sessions
If a sending client authenticates their session after connecting to the SecurityGateway server with a valid username and password, the sent message will not be scanned by SpamAssassin. This option is enabled by default.
Exclude messages from domain mail servers
Enabling this will allow any message sent out from the SecurityGateway server to skip being scanned by SpamAssassin. This is disabled by default, and is only recommended after verifying SecurityGateway can only be used by valid users, to prevent being used as an open-relay.
Use built-in local SpamAssassin (SGSpamD)
This option will use SecurityGateway's local SGSpamD process on the same server to process messages for potential spam. This option is enabled by default, and can be further configured by clicking on the link under the option. More information on configuring SGSpamD can be found in the related article below.
Use a remote SpamAsassin daemon (SpamD)
If you have another server running SpamAssassin, for perhaps performance reasons or for running multiple email servers, you may use this option to point SecurityGateway to that server's address and the specific port to connect on.
Heuristic settings can be configured for either the entire server, or for specific domains, as chosen in the upper-right corner.
KBA-01885 Configuring SGSpamD updates in SecurityGateway
KBA-01886 Configuring Bayesian Learning updates in SecurityGateway