MailScanner Unable to Call SpamAssassin

[Sarah]

According to the output, sa-learn is still working: "Learned tokens from 1 message(s) (1 message(s) examined)" - that indicates success.

[Serra]

One thing I did notice is that the ALL_TRUSTED flag has changed from 0 to -1. I was using ALL_TRUSTED to identify local mail, but now all mail from SMTP shows up under ALL_TRUSTED. This seems to make more spam go through since it is hitting most spam with a -1.

[duncang]

You will need to roll back NetAddr::IP by one version as there seems to be some sort of incompatibility with SA, a similar issue was noted this time last year. From what I've seen SA does not function correctly and it will let very high levels of spam through incorrect marking.