We have a custom mail server which sit on top of Postfix. The spam scanner is done in Postfix, at that time we lost a certain information such as incoming IP that rspamd see is our IP address. We also signed the email b y the time it reached postfix.
We eventually integrate with rspamd using its HTTP Protocol at higher layer so we can tell rspamd the incoming IP address and the raw email. BY doing that, rspamd can work more effectively and by the time email reached Postfix, the email is legitimate to send out and Postfix only need to send email out.