Spamassassin - how to ignore old emails

[rchurch]

whenever I train spam assassin, the old emails appear next time around, all needing to reset again.

How can I stop that happening?

Does the forget check box stop some from coming up next time, or does it get Spamassassin to treat them as though they have never been tagged?

 

[iascadmin]

I believe that spamAssassin will display all emails in an inbox for a user.

And it defaults so that all email are "ham" unless set otherwise in that session.

To get email not to come up again, remove the email from the inbox.

In chekcing my own email accounts, I check in plesk first and train spamAssassin for an account, then check in horde and delete all spam - either manually or through a filter rule. And purge deletered email.

 

[creativevision]

I actually went through 350 emails and marked them all SPAM and had to wait 5 minutes for the page to refresh or update. But no success message. My browser said it was Done.

When I wen back to the page, it shows the exact same 350 emails all marked HAM.

What is the truth here? Did i spend 45 minutes marking 350 emails as SPAM for no reason? Does the SPAM ASSASSIN tool now identify all these emails as SPAM, because there is nothing, absolutely nothing on the server that lets you know it has been marked as SPAM or if it was a success. I went to one email account that had 10 SPAM emails and i changed the settings from HAM to SPAM, and clicked submit. It appears it worked, but they all say HAM still and are still there.

This is confusing. How does this work? Is it working?

 

[iascadmin]

Let me try to be a bit clearer.

SpamAssassin as implemented in Plesk works in concert with but separate from the other email packages.

You access and train SpamAssassin through the Plesk control panel.

You access email through horde or some other mail client, such as outlook.

Each time that you go into the plesk control panel and train SpamAssassin from a users email account setup in the control panel, SpamAssassin scans that user's inbox. The radio buttons in the training session are all set to ham by default, regardless as to what SpamAssassin has scored them - spam or ham.

To prevent previously marked spam from appearing in the SpamAssassin trainer in the plesk control panel, you must delete the spam from the inbox in a mail client. In horde, select each of your 350 spam email, and either blacklist which will take you to the hordes filters [not related to SpamAssassin], click on Apply All Filters, which will take some time to do, and then you will be returned to the inbox. Then click on Purge Deleter. In an other email client, you can either mark those spam email as Junk Mail, or just delete them.

Logout of your email client. Back out of SpamAssassin.

The next time you go into the control panel you should only see new email in the SpamAssassin trainer.

My oldest email address, over a decade of use now, gets over 500 pieces of spam a day. And I do this every day.

 

[iascadmin]

Ah, one other thing. If you do delete the spam in a mail client other than the horde webmail client, especially if you have the client setup to treat that email account as a POP3 account, assure that your email client is configured to delete email on the server when you delete them in the client. Otherwise, they will still be in the inbox, and still appear in the plesk control panel SpamAssassin training area.

 

[creativevision]

iascadmin,

Thanks for the great advice.

Is this an ongoing thing that needs to be done every day?

I thought the spam assassin tool was a tool that automatically does all this work for us? I am showing 1500 spam emails in hmy email box using HORDE.

After doing this for a week, will the spams start being alot less each day or month?

Also does the black list data show up in the PLESK server?

What kind of results can we expect?

 

[iascadmin]

As far as I know the black list in horde is separate from the blacklist and whitellist data in SpamAssassin.

SpamAssassin allows for both server wide and personal blacklists and whitelists.

SpamAssassin is a Bayesean statistics based tool. This means that it recursively updates the probabilities of the variables that it uses to score each email, marking it as spam or passing it as "ham". Each successive training becomes the "test" distribution" for each variable, modifying the "a priori" distribution of the previous training sessions, combining through mathematical conjugates to develop the "a posteriori" or current distribution for each variable. A very robust method.

Yes, this should be done every day, but as you go forward, and SpamAssassin learns, more email will be correctly marked as spam and as ham, and you can increase the aggressiveness of the scoring [lower the number of hits required]. The more you train it, the more accurate it will be.

I would recommend that you only have one filter rule in horde for filtering out JunkMail: that is that the filter in horde looks for the Spam tag that SpamAssassin uses to modify the subject of Spam. The default is ****SPAM****. Have the horde filter check for this tag in the subject line and move the spam to a Junk folder.

We don't recommend that you allow SpamAssassin to automatically delete spam. It could catch some email that you really want to recieve, especailly if you increase the sensitivity - decrease the number of hits required.

One other thing, as root, in an ssh session, use

ps -ef | grep spam

and assure that you are using psa-spamd and not spamd. This way you know that you are using the plesk interface to SapmAssassin. You can search the forum for other posts regarding this issue.

 

[creativevision]

Here are a few questions:

In the plesk control panel, here are settings:

1. Useage policy settings - do we choose server wide and personal settings?

2. Should the hits required for spam be set to 4 ???

In the spam filter configuration under the personal email account, what are these supposed to be?

1. Do we check the box for server wide settings?

2. When i click on training, and when i check emails as SPAM, it does not seem to do anything when i submit. It shows the same list but all as HAM.

 

[iascadmin]

Here are a few questions:

Here are some answers, maybe...

In the plesk control panel, here are settings:

1. Useage policy settings - do we choose server wide and personal settings?

2. Should the hits required for spam be set to 4 ???

If you have blacklist and/or white list information from which you feel all users on your server would benefit, select both server-wide and personal, and then enter blacklist and whitleist email addresses as *@domain.TLD or *.domain.TLD or sepcific email addresses. The first form applies the rule to all email addresses from that domain while the second applies to all email from all subdomains of that domain.

For server-wide settings, we stick with the default of 7 hits required; that is, a message must get a score of at least 7 [match seven of the criteria] to be tagged as SPAM. The lower the number of hits required, the more aggressive is the engine, and the more likely that acceptable email will be marked as spam. Also, remember, one person's spam is another person's ham. So keep the server-wide setting fairly loose.

In the spam filter configuration under the personal email account, what are these supposed to be?

1. Do we check the box for server wide settings?

2. When i click on training, and when i check emails as SPAM, it does not seem to do anything when i submit. It shows the same list but all as HAM.

Each user can decide whether or not they want to use the server-wide settings in addition to their own settings. The same with the white lists and black lists. If you have server-wide settings enabled, the administrator's list for each is added to the personal list. A user can add their own good and bad email addresses to the white and black lists, can delete an administrator's selection from their list, and can make the scoring [number of hits] more or less agressive for their personal account.

The second question here is the same that started this thread. Please read the above responses. BTW, when you click on OK after training emails as spam or ham, you should be taken back to the SpamAssassin main screen for that account in Plesk - the one with training at the top, black lists, whitelists, etc.. Don't go back to training until after you have deleted spam from the inbox ON THE SERVER for that account .

Hope this helps.

 

[xtreme2490]

Originally posted by iascadmin

One other thing, as root, in an ssh session, use

ps -ef | grep spam

and assure that you are using psa-spamd and not spamd. This way you know that you are using the plesk interface to SapmAssassin. You can search the forum for other posts regarding this issue. [/B]

[root@Rayden root]# ps -ef | grep spam
root 1960 1 0 Feb06 ? 00:00:01 /usr/bin/spamd -d -c -a -m5 -H
popuser 1972 1 0 Feb06 ? 00:00:32 /usr/bin/spamd --username=popuser --daemonize --helper-home-dir=/var/qmail --max-children 20 --create-prefs --nouser-config --virtual-config-dir=/var/qmail/mailnames/%d/%l --pidfile=/var/run/spamd_full.pid --socketpath=/tmp/spamd_full.sock
popuser 1974 1 0 Feb06 ? 00:00:01 /usr/bin/spamd --username=popuser --daemonize --helper-home-dir=/var/qmail --max-children 20 --create-prefs --nouser-config --virtual-config-dir=/var/qmail/mailnames/%d/%l --pidfile=/var/run/spamd_light.pid --socketpath=/tmp/spamd_light.sock --siteconfigpath=/dev/null
root 17725 17680 0 08:43 pts/0 00:00:00 grep spam

Does this means that there's something wrong with my configuration ?
If so , how to solve.

 

[iascadmin]

Your results looks correct for psa use of spamAssassin spamd.

Does anyone else see anything different?

 

[modom]

Is there an easier way to remove blacklisted emails in horde rather than one at a time?