Ah, I didnt see that filter criteria. excellent thanks.

Ok thanks Ian.

If there are any other headers that are added and are specific to RDNS / XBL detections I'd like to know, as I want to do something different with them than I do with spam detected by Bayesian methods.

Cheers!

I don't think it's as simple as restoring a dictionary, the think the 'new' bay training data is held in memory until the MTA service is restarted. Only then is it written to the TAB file. I found that sometimes on restart of the service only, the ditionary file gets updated, but on a complete OS re...

It's happening to us again now... I restarted the services yesterday and it dropped from about 460/460 to 108/112. joy!

Hi, When I enable DNS Blacklisting and set the action to "Mark the email as spam", can anyone tell me how the messages are marked? If it's with a header line, can you please tell me what the string is so that I can create a filter to handle it. I have searched through the headers of loads of email, ...

I found no cure to this. I did manual builds for a while but got bored of building all the time, so now I've switched back to Auto Training and it seems to be working fine. I think what took a while to get right was the ham/spam collection method. Right now its getting spam from a honeypot mailbox w...

Hi Smurf, when you say you 'prune' can I ask what you mean? Is that the merge option? I've not yet worked out how or what the merge option does so I just delete the TAB file and rebuild the dictionary every few days manually. I'll script it at some point, but I'm thinking that a 'prune' might be exa...

Ok thanks for your reply. I'll have to keep collecting. This morning I gathered the weekends captured ham (authe'd stuff) and an equal amount of spam then built a dictionary based on 1112 of each. Here's the results: Spam Calc Method - 'Recommended': Percentile: 90% 399 [100*(399/1112) = 35.88% ] Sp...

I have a similar setup but in my whitelist filter theres the following: Criteria : Where the From header line contains specific words - Within that criteria, 'Enabled' is ticked and 'Use external list', and the list is in this format: *@gooddomain.co.uk *trusted-domain-ending.gov.uk allowed@i-love-t...

Hi, I'm wondering how big (in token count) your dictionaries needed to be before they became really effective. At the moment I manually build mine and have, right now, 915 spam and ham sitting in the NoSpam and Spam folders. Using 'MESPAMCMD.EXE -v' and running it against the Spam folder I get: Perc...

Well I couldn't find a fix so I've given up on auto training.... Uh... which isn't too difficult.

Just thought I'd update you all as I know you're so interested

Hi, when you look at the headers, do you see any reference to a Bayesian score?
How are you building the dictionary? With auto-training or manually?

Hi, I have Bayesian set to autotrain. As there is always WAY more spam than ham the training is slow, but the token counts do increment slowly. At the moment the spam/ham count in System Overview is 107/113. Everytime I restart MTA or have to reboot the server the figures always return to these. Thi...

Ah right, thanks a lot. I've got that setup now. I'd seen the header in some of the mails but was looking at the logs for proof of the detection. I presume that the same header is used to mark the XBL/URL BL detections also yes? Do you have any idea why I can't see these detections in the logs? I'd ...

Hi, I have same sort of question and decided to carry on this thread. Pro v3.05, W2k3. I have RDNS setup the same way as the OP and can't find any mention of messages being marked as spam in the Filter logs, or any of the SMTP logs. - What should I look at next to troubleshoot? - When a message is '...