Orientation about filtering mail

[cbcentos]

Hi you all guys.

I need some orientation about mail servers. My head is burning because of reading tutorials, manuals and more, and at this moment I'm very confused.
I have a small server with CentOS 5 which is providing internet access to some workstations. Also have a hosting service which provides me a few pop3 mail accounts.
Each workstation has an account and use Thunderbird ford receive and send mails.
I want to filter spam and prevent it from coming to the accounts, so when users open Thunderbird, has no spam.
What I want is the easiest way to do this.
Actually I don't know if I need to retrieve mails first to the server (with fetchmail by example) and then configure Thunderbird to handle messages from local accounts or if I can filter incoming mails directly from ports.
I only need to filter incoming mail, not outgoing.
I'd be very grateful for any information about this topic. I'm so confused and need to find the way. :-)

Thanks a lot.

[michaelnel]

I use spamassassin on our mail servers to analyze incoming mail and mark spam as spam. It does not delete the spam or reroute it, it simply adds "[SPAM]" to the "Subject:" field of the message.

Then it is up to the users to set up filters to do whatever they want with mail that is so marked.

I *could* just send spam to /dev/null if I wanted to, but since there is a very small percentage of mail that is falsely marked as [SPAM], I don't want to trash it, and prefer to let the users deal with it.

Setting up filters to deal with marked spam is trivial in Thunderbird.

[cbcentos]

Hi [b]michaelnel[/b].

What I want is to prevent mail to arrive to Thunderbird. I want to filter spam in my server and only deliver correct mail.
My hosting and mail accounts provider already has a filter that marks spam. Maybe I could use this for filtering, but always in my server, I don't want to filter it in Thunderbird.

Thanks.

[michaelnel]

Then you can set up procmail as the local delivery agent on your server to do whatever you want with the mail that has been marked as spam.

It's not particularly easy to set up procmail the first time you encounter it, but it's very powerful.

[michaelnel]

BTW the problem with disposing of spam upstream of the user is that nobody but the user can really tell what is a false positive. I prefer to err on the side of caution and not delete ANYONE's mail, and leave it to the user to decide.

On the other hand, IME most users are going to simply let it build up in the SPAM folder without ever looking at it. Heck, most of them have 10,000 messages in their inbox, what's a little spam or the occasional false positive to people that clueless?