Page 1 of 1

Training Magic Spam

Posted: Tue Aug 04, 2015 3:15 pm
by RCC
Is there a way to train magic spam manually what messages are spam and what are ham? I would like to be able to help Magic spam learn faster.

Re: Training Magic Spam

Posted: Wed Aug 05, 2015 9:26 am
by magicspam
Hello RCC,

Thank you for your post and for your question. It sounds like you may be used to software which employs Bayesian methods to "learn" what spam is (eg. SpamAssassin).

MagicSpam works in a very different way though and in fact doesn't perform any content filtering at all. This is how MagicSpam is able to operate in a very fast and efficient manner; content filtering is very resource intensive so we try to avoid it. You may have read elsewhere on this forum, but MagicSpam operates by leveraging IP Reputation lists and enforcing SMTP server best practices, all within the SMTP layer before a message is even downloaded to your server.

Regarding your uncaught spam issue; MagicSpam default recommended settings are relatively safe and it is likely that you just need to enable an extra rule or an extra list to bring MagicSpam's performance up to a level where you are happy with it. If you could, please post some MagicSpam log examples of messages that were marked as "HAM" but should have been blocked (feel free to obfuscate any sensitive email addresses). With these examples we can provide you with a suggestion to stop them moving ahead.

Thanks!
-- MagicSpam Support Team --

Re: Training Magic Spam

Posted: Sat Aug 29, 2015 11:01 am
by kara
Not RCC, but I'll step in! Just set up MagicSpam and have it set to default settings only. I am still receiving all sorts of emails on Herpes, Rachel Ray, Russian Brides and etc.

Mua IP Address Hostname Helo Sender
no 104.171.118.183 asd183.harmokol.com barajas.xyz foxnews@barajas.xyz
no 216.169.119.178 dynamical.bowma.uno dynamical.bowma.uno margery@bowma.uno
no 185.109.147.129 f9jbzh.zogika.racing f9jbzh.zogika.racing GetAbsQuick@zogika.racing
no 216.169.123.92 bordered.cistr.uno bordered.cistr.uno nicolas@cistr.uno
no 185.109.145.20 wpwbv.aboslec.review wpwbv.aboslec.review 1ink@aboslec.review

Re: Training Magic Spam

Posted: Mon Aug 31, 2015 8:50 am
by magicspam
Hello kara,

Thank you for your post.

It appears as if those email messages are flagged in the MIPSpace-Poor list. Simply go to 'Settings' > 'IP reputation' and enable the MIPSpace-Poor list. This should inevitably reject those unwanted message. However, if you face any other issues, please let us know.

Thank you!

-- MagicSpam Support Team --