The training bypass feature allows you to define cases where GBUdb should not be trained. For example, you may have an abuse reporting address that could potentially teach GBUdb to distrust the IPs of your customers when they forward spam to that address. You might also accept traffic from web servers that produce email from forms that may occasionally be abused -- and you would want to prevent GBUdb from learning to distrust those IPs.

The bypass section is a container for bypass header directives. If one of these directives is activated then GBUdb training is disabled for that message.

Note that the pattern scanner is still active when GBUdb training is bypassed. A message that matches a bypass header directive can still be marked as spam by the SNF pattern matching engine. If a bypass header directive is active then the effect is that no IP statistics will be collected for that scan.

Please email support@armresearch.com with any questions.