Here is a setup:
domain.com MX 10 relay1.domain.com MX 20 relay2.domain.com
relay1 & relay2 have booth a SpamAssassing running.
Problem: when a mail is received on relay2, it's being check by SA and tagged as spam. Later, relay2 sent it to relay1. relay1 don't mark it as spam??? How to re-use the tags added by relay2?
Depends on how you are set up to call SA, which you left unspecified.
If you're using something like procmail, you can merely configure a procmail rule on relay1 to bypass any email which has "X-Spam-Status: Yes". Be sure to not bypass anything with "X-Spam-Status: No" as this would allow a spammer to bypass your scanning by faking a scan.
However, I would also consider checking the root cause of the problem. Why did one tag it, but not the other. Look at your maillogs (use the SMTP id from the Received: headers for hints) and see what the hits of the message was on each server. What rules matched on server2 but not server1?
If your difference is that server1 doesn't see dialup lists, adjust your trusted_networks setting to include <ip of relay2>/32 and <ip of relay1>/32.
If your difference is bayes, perhaps you should revisit your training a bit.
