we are currently testing Zimbra Network Edition 5 in a multi-server environment (2x mta, store, ldap, proxy).
We just found out, that outgoing Mails do also go through the spam-filter. And apperently, all our outgoing mail gets tagged as SPAM by our OWN spam-filter.
We have tested internal mail (zimbra -> zimbra), external (zimbra -> other mailserver) sending via webinterface and mail-client.
It mostly results in a X-Spam-Status like this one:
X-Spam-Status: Yes, score=5.065 tagged_above=-10 required=5
tests=[AWL=-0.317, BAYES_99=3.5, RCVD_IN_PBL=0.905,
Thats quite weird in my opinion. Whats the problem here? I've even tested it with different domains, assuming there maybe some dns misconfiguration - same result.
We setup MTA normally and installed SPF on both mta and did some training via the UI. Besides no config changes...
First of all, why does spamassassin also filter outgoing mails (which got delivered by an authenticated user)?
Second, why does spamassassin tag my normal, daily, non-suspicious mail with BAYES_99, which IMHO is only based on textual recognition?