White List Multiplying Like Rabbits

[mcgurme]

I stopped using SpamSieve earlier this year because it wasn’t working well - but the other things I tried after weren’t either.

So I reinstituted SpamSieve a few days ago. It was missing everything.

I reset the corpus, and I reset the whitelist yesterday (the whitelist had >20k rules, most of which were active).

It has been doing much better today, however: in less than a day, my whitelist has grown to over 7,480 rules, most of which are active.

Now, though I receive lots of emails, I haven’t trained anywhere near that number yet.

So it appears that SpamSieve is going through my mailboxes and auto training on old mail!?!? Or what else could be happening? I haven’t cleared out all the spam from the >10,000 messages that are still around in my inbox and various mailboxes. If SpamSieve is using these (rather than only NEW mail), I’m sure it will go back to poor behavior quickly.

What can I do to prevent my whitelist from blowing up like that? What exactly is SpamSieve doing to determine that an address belongs on the whitelist (apart from me manually marking it as Good)?

Thanks

[Michael_Tsai]

@mcgurme

I stopped using SpamSieve earlier this year because it wasn’t working well

Should that happen again, please send in a report.

@mcgurme

I reset the corpus, and I reset the whitelist yesterday (the whitelist had >20k rules, most of which were active).

Usually the only reason to reset the whitelist is if messages were incorrectly trained or mistakes were not corrected.

@mcgurme

So it appears that SpamSieve is going through my mailboxes and auto training on old mail!?!?

SpamSieve only auto-trains on new messages that are received. It does never goes through old messages for any reason. However, if you are using Apple Mail and manually apply the SpamSieve rule (or click Apply after editing the rules and Mail asks if you want to apply them to the current messages) then it would see those messages as newly received.

If the latter is happening, you probably have lots more problems than the whitelist. The corpus would be messed up, too, unless you went through and corrected the mistakes from those messages.

@mcgurme

What exactly is SpamSieve doing to determine that an address belongs on the whitelist (apart from me manually marking it as Good)?

SpamSieve auto-trains the whitelist for all new messages that it thinks are good. If you later train one of those messages as spam, it disables the whitelist rule.[QUOTE][/QUOTE]