Anti-Spam

[akertis]

OK. It seems that more and more spam is getting through on our server. I'm assuming the spammers are getting better. It seems many emails are just text with a hyperlink now and have a score under 1.

I'm not sure how much training spam does. How long does this take to really have the zimbra server learn ?

Does everyone just use the builtin clamav for anti-spam? It seems in the last 3 weeks the amount of spam getting through has gone up. Has anyone else seen this? The only other thing I can think of doing is to lower the percentages for the anti-spam options.

Anyone have any ideas?

Thanks!

[raj]

you can read things in wiki which can improve your SPAM trapping.

http://wiki.zimbra.com/index.php?tit...ti-spam_system


Raj

[ewilen]

I would make use of some DNSBLs at the MTA level, as there are a number of them that are discriminating enough to use as straight blocklists. b.barracudacentral.org as a start. Some of them will overlap with the default spamassassin rules, but that shouldn't matter. (I'm thinking it might be worth it to deactivate those rules, though, to cut down on the number of DNS lookups.)

And in addition to the wiki page, this thread may be worth a look: [SOLVED] SaneSecurity ClamAV or FuzzyOCR SpamAssassin Plugins

[phoenix]

Originally Posted by akertis

The only other thing I can think of doing is to lower the percentages for the anti-spam options.!

What are the current settings for the Kill/Tag percentages.

[uxbod]

Using the Barracuda RBL will reduce your SPAM requests quite a bit plus the use of SaneSecurity signatures aswell.