Hello everyone. I just switched to MIAB, and what an amazing setup. Thanks to Josh for creating an amazing package.
So my SA-LEARN shows clearly that when I move messages from inbox to spam, they are adding up as nspam. I’ve been training my Spam Assassin database now for a week, and every day i’m getting about 20 messages to 50 messages in my INBOX which I’d like to be marked as SPAM.
IMPORTANT: These are messages from variousl newsletters and mailing lists. So it is the same news source every morning.
So I’ve been dragging them from Inbox to Junk, and I see the numbers on the Spam Database (sa-learn --dump magic) go up gradually. Which shows the plugins are working great.
Now my question is, why is SPAM ASSASSIN not marking these messages with a bigger SPAM score? I see in the headers that there is a SPAM score on them, but since I’ve been training the SPAM database, when the similar message arrives from that same news source, why would it not automatically go to Junk Mail?
The headers show that the SPAM WEIGHT isn’t high enough (1.8 vs 5.0 or above). Shouldn’t that training help the Spam Score go up? Or shoudn’t that training somehow tell Spam Assassin, hey this guy moved this message from INBOX to SPAM 17 times, so it must be SPAM, hence therefore let’s automatically help him, and move this message to SPAM?
What am I doing wrong? Or this is the expected behaviour and will it take forever to learn?
Any ideas? Thanks.
Are these really ‘spam’ or just things that you do not want to receive any longer? If the latter you’d be doing a horrible disservice to the sender. I sincerely hope that you utilize the unsubscribe function.
Lol :)… Hi Alento, no they are really spam. And no, this is in no way a “horrible dis-service”, as this is my private mail server testing. Whatever I decide to mark as Spam, I should be able to.
But in any case, that is NOT the original quesiton. Do you have the answer to my query?
No, I am sorry but I don’t. I have learned most of the MiaB system in and out, but spam filtering is still a mystery to me. I am looking forward to responses to expand my knowledge. Have you gone through the project’s GitHub ? I am sure that SA has been discussed there.
Ahh ok, well hopefully we can both hear from someone with experience here.
Yes, that and also looked at detailed SpamAssassin documentation. All tests clearly show it is working (the movement of messages takes up the nspam number in the database and the tokens are there to show it).
Just that in actual practice, I don’t see that happening.
So I noticed an interesting behavior (test server) and wanted to validate if that is expected and correct.
If not, then that’s what might be causing the issue described above.
Email, which should be spam, comes into my inbox.
The “nHam” counter on the database goes up, but “nSpam” does not.
I move the email from INBOX to SPAM folder.
I see the “nSpam” counter go up now, BUT BUT… the “nHam” counter does not go down !!!
Question, is that expected? Or should the “nHam” counter go down when that specific email, that took it up, is moved over to Spam folder?
If this is expected behavior, then I suppose I’m back to trying to figure out why my Spam learning is not working?
If this is not expectted behavior, then what should I do to fix this?
Ok, no response here yet, BUT I’ve been working on my server diligently . Happy to report that it is beginning to work, and placing some of my daily “junk” mail automatically into the junk folder. Some of my notes and thoughts on the matter:
The SA-LEARN takes time. My experience shows not just loading it with bunch of “nHam” and “nSpam” type messages, but actual time. It almost seems like it is training and seeing a pet grow and understand. Hmm… I digress, but the point here being it takes time.
The behavior described in my last post, about the counters, is very much normal.
I know it is currently not in scope for Josh to be working on this, but as an improvement to MIAB, I would love to see an interface of Spam handling on the Admin Portal. Maybe some of the Gurus here can work out a magic and create a Fork? (Just like the one for quotas, which I’m using and it works GREAT. Thank you to the gentleman who created it).
Sending my warmest regards and best wishes to all who have been working hard to create this wonderful solution.
