SpamAssassin Question

[Sekka]

Hi,

I am currently running on a define as spam any email that scores above "4.5" system with about 330 emails trained.

Now, SpamAssassin is still missing quite a few emails.

At what point does all the training kick in? And is there any way to find out what score these emails got that passed the spam filter?

I looked at the source code of the email in outlook and there is no tag defining its score.

Thank you.

Edit: Also, I use Outlook which removes emails from the server. Does this remove the training?

 

[lpittman]

How are you training the email?

Removing them from the server will not reverse the training - it is stored in the bayes database (sql or files on the server).

4.5 is pretty low - you'll probably get a lot of false-positives at that setting.

What else are you doing to prevent spam? Do you have dcc, razor and pyzor installed? SPF? RBL? RulesDuJour?

There are a number of steps you must take to limit spam, not just rely on SpamAssassin alone.

Let me know else you have and what server admin experience you have and I'll try to help.

Luke
www.catch22media.com

 

[Sekka]

Thanks for your reply.

The email addresses we host are mostly business addresses and as such either recieve personal emails or spam. But we will be hosting personal emails at some point so I want to get this right

I am running my address at 3.5 now and it is catching all spam. No false positives yet, but this is only 1 day.

To train it, if an email shows up in my inbox and it is spam and not marked spam, I just go into the SpamAssassin panel, select it, and click "Spam". Doesn't look there is anything else I can do. Same for false positives, just tick "Not Spam".

I'm only training it on emails it gets wrong, and this seems to be having a better effect.

I am a web developer and I manage our dedicated server at work. As for server knowledge, I'd class myself as beginner. I know SSH and the odd unix command and I have a general idea of how a server works, so fire away. Good news being I am a fast learner. More Spam filters the better.

 

[lpittman]

What OS are you running?

I personally use a server-wide setup of spamassassin, as I find it far easier to maintain and gives greater overall results. I can help you with setting this up if interested.

The cons are: you have to do all your training, configs, etc from the command line and your users will not have individual settings.

Let me know your OS and we'll continue ... I have only worked with FC and CentOS, so if your using something else I may not be able to help.

Luke

 

[Sekka]

I didn't know you could do the training serverwide. Interesting...

Yeah, I run CentOS.

 

[lpittman]

Great

Not sure where to start here ... I've been working on my spam filtering for some time and haven't necessarily done everything in the right 'order' although its all working now.

Do you have ARTs yum repositories setup?

Luke

 

[Sekka]

Beginner knowledge here. You've lost me already.

The dedicated server was set up by our hosting company so I have no idea what is it on it.

If you can point me to an online tutorial or something I should be ok from there, unless you've got the time to spare.

I'll get any information you need, just tell me how to get it.

 

[lpittman]

I have a bit of time, I'll try and help.

Yum is a utility to help manage packages installed on CentOS (or other linux distro). Yum works off of "repositories" of packages (packages could be mail programs, control panels, kernel updates, etc). Because we will be installing some packages that aren't supplied on the default Plesk repositories, we have to add our own. In this case we are going to use AtomicRocketTurtles repository. He is a well known, very friendly linux guy who was a founder of Plesk. He generously continues to work on updates and packages for Plesk that help out all of us.

Check out his site here: www.atomicrocketturtle.com

His repository info is here:
http://3es.atomicrocketturtle.com/

(scroll down for the configuration info)

The stuff your getting into here is not exactly basic, but not too advanced at the same time.

Have a look and read through it all ... if it makes sense, go ahead and add his GPG Key and repository.

Let me know how this much goes...

On a side note - if your simply looking to have your spamassassin setup properly and don't care to learn, let me know and we can work something out where I just login and do the basic updates for you.

Luke

 

[Sekka]

Ok, I added the key,

rpm --import RPM-GPG-KEY.art.txt

What now? The notes below say something about some "Yum Atomic Archives". Sorry if I am being very simple, I don't want to break anything.

 

[lpittman]

Now you need to add ARTs repo...

You should have a directory as follows:

/etc/yum.repos.d/

Correct?

If so, you should see the following:

CentOS-Base.repo
CentOS-Media.repo

As far as I understand, you simply want to add one for ART. I did the following:

touch art.repo
vi art.repo (loads your basic text editor to edit art.repo, you should become very familiar with vi if you want to do any kind of server admin)

once vi loads, press "a" to enter "Insert" mode and you'll want to put in the info that Scott (mr. turtle himself) outlines in his page ...

I currently have

[atomic]
name=Atomic Rocket Turtle - $releasever - Atomic PSA-Compatible RPMS
baseurl=http://3es.atomicrocketturtle.com/atomic/art/$releaseverES/

[atomic-app-vault]
name=Atomic Rocket Turtle - $releasever - Atomic PSA App Vault RPMS
baseurl=http://3es.atomicrocketturtle.com/atomic/app-vault/

[psa-8.0]
name=Atomic Rocket Turtle - $releasever - SW-Soft PSA 8.0 RPMS
baseurl=http://3es.atomicrocketturtle.com/atomic/psa-8.0/$releaseverES/

Now things will be changing a bit with 8.1, but I have made any of those changes myself yet.

Once you've done this, do a "yum check-update" and you should either see something like this:

[root@server yum.repos.d]# yum check-update
Setting up repositories
psa-8.0 100% |=========================| 951 B 00:00
update 100% |=========================| 951 B 00:00
base 100% |=========================| 1.1 kB 00:00
atomic 100% |=========================| 951 B 00:00
addons 100% |=========================| 951 B 00:00
atomic-app-vault 100% |=========================| 951 B 00:00
extras 100% |=========================| 1.1 kB 00:00
Reading repository metadata in from local files
primary.xml.gz 100% |=========================| 51 kB 00:00
atomic : ################################################## 145/145
Added 0 new packages, deleted 2 old in 0.15 seconds
[root@server yum.repos.d]#

or a big list of packages on your server that it wants to update.

Now here is where you need to be careful. If you don't have any experience you don't want to simply put in "yum update" and let it update everything as there is a very good chance you'll cause some issues with your server. As you gain experience and understand how this works, you'll want to get yourself to the point where you are running "yum update" daily, or every couple of days making sure your system is up to date.

In any case - it would be a fantastic idea to pull out an old pc and install the same version of CentOS and Plesk you have on your production server and do some test setups first. Since I've done this I've learnt a tremendous amount.

Let me know if you get this far and we'll move on to installing dcc, razor and pyzor.

Luke

 

[Sekka]

You should have a directory as follows:

/etc/yum.repos.d/

Nope

I'm pretty sure yum is on there. I did,

yum --help

And it brough up the help for the commmand.

 

[lpittman]

Hey,

Okay I'm gonna shoot this out there now - if you expect to get any help from a community of this kind, you'll have to start by helping yourself. Meaning this; don't just read the lines I type, but read between the lines I type and try to understand what I'm trying to get you to do and why I'm trying to get you to do it.

That being said - I understand you are new to this and might not know how to go further than I'm asking, but at least try.

Linux stores configuration files (among other files) for the majority of its packages in /etc/ ... so you could expect to find a 'yum.conf' file in there if Yum is installed (which you've pointed out it is - good).

Do a 'vi' on that 'yum.conf' file and read it ... it should give you some info about where/how to add Repos. If it doesn't, use "man yum" which should pull up the manual pages for yum.

Try this and see where it gets you.

Luke

 

[jwdick]

You may NOT want to add:

[psa-8.0]
name=Atomic Rocket Turtle - $releasever - SW-Soft PSA 8.0 RPMS
baseurl=http://3es.atomicrocketturtle.com/atomic/psa-8.0/$releaseverES/


Art's releases for PSA are for RHEL and are older than Plesk's PSA releases already on your CentOS server. At least they were on mine. If you use this, make sure it is what you want to "update" before you do this.

 

[Sekka]

I'm taking lpittman's advice and getting use to yum for now.

There is no folder called /etc/yum.repos.d/ and in the conf there is no apparant indication on how to add this, or in the manual, so I'll go snooping about on the internet for some info.

My conf file just contains 4 things from what I can understand,

CentOS-3 - Addons
CentOS-3 - Base
CentOS-3 - Extras
CentOS-3 - Updates

I don't know if this helps at all.