I'm using postfix with amavis and spamassassin to filter out spam before relaying everything to an exchange server. Regularly I got mails which weren't catched by spamassassin, even worse, they were autolearned as ham.
So I started to collect them and exported them in plaintext and want to retrain those mails, marking them as spam.
From what I read so far is, that you don't have to specify anything that you want to "re"train a mail. You should just run with --spam
and it should forget the mail as ham and train it as spam.
But observing sa-learn --dump magic
shows, that it keeps the autolearned ham entry AND adds it to spam.
My Idea is, that Exchange somehow edited the mails so that, when exporting and putting it back to the postfix server, it isn't the same mail than before. Could that be the case? And if so, is there a way to do what i planned to do, for example by removing some headers?