Elizabeth Schwartz wrote:
> JADP - the rfc-ignorant rules lost us some important email today. The
> customer was throwing away all mail tagged as SPAM after many months of
> no false positives. I've turned those rules off on my site, and continue
> as always to encourage my users to check their tagged mail before
> tossing it.

*nod*  The *ONLY* mail I autodiscard immediately is mail tagged as a
virus - and even there I'm cautious.  Spam is tagged, and filed in a
separate folder.  I *do* have autoexpiry processes set up to keep the
spam folders from growing out of control, but the shortest period they
run on is 7 days.

> IMHO if a rule is getting legit email tagged as SPAM it should be toned
> down.

And that's one of the real benfits of SpamAssassin;  you *can* do just
that.  If one rule is persistently misfiring on your particular mail
flow, you can score it down or zero it out completely.  If you're
getting really poor scoring across the board, you could even go to the
effort of completely rerunning the entire scoreset to customize it to
your mail flow.  I don't know of many commercial products you could do
that with.

A few years ago, I was seeing FPs on whitelist_from_rcvd for PayPal
(IIRC - it's been a while.)  So I temporarily overrode that whitelist entry.

 Obeying the RFC's is a good thing, but I am trying to tune our
> spam filter to filter spam, not to be a netcop. Our particular contact
> seems to have gotten onto rfc-ignorant's list because it is rejecting
> mail from <>

... which is a *REALLY* bad thing for a mail server to do.

I don't reject mail outright from many systems, but rejecting legitimate
postmaster notices (which are, by definition, generated with the null
sender "<>") is high on my trigger list once a system has been seen
doing other unsavoury things to their mail flow.  I wonder if they ever
read mail sent *to* [EMAIL PROTECTED]

I regularly see remote systems refusing mail delivery notices
(autogenerated by the server that delivers the message to the
appropriate inbox **if requested by the sender** - IIRC Outlook can
request this notice).

On the other side of this debate, I refuse to let client MUAs use the
null sender;  I can't think of any reasons they should ever do so.  Read
receipts should be sent using the usual sender address, so that if the
recipient(s server) bounces it, it goes back where it belongs (ie,
**NOT** in my postmaster mailbox).

-kgd

Reply via email to