Spam training from junk folders

[mdeneen]

Hi,

I've read this link about getting zmtrainsa to process all messages in the junk folder, not just those marked as junk from the web UI.

I was planning on using the following script to do this for me:

#!/bin/bash

for i in `zmprov gaa mydomain.com`
do
        /opt/zimbra/bin/zmtrainsa $i spam Junk >> /opt/zimbra/log/spamtrain.log
done

Then, I looked at the zimbra crontab. There is a line which calls zmtrainsa with a --cleanup option 45 minutes after the spam training starts. What does --cleanup do, and what happens if it takes more than 45 minutes for spam training?

If I am forcing spam training to happen on all messages, do I have to force a cleanup as well?

This is my first Zimbra install. I have to say, I'm really impressed!

Best Regards,
Mark

[phoenix]

You don't need to use zmtrainsa against the Junk folder, anything that's in there will also have gone to the spam training account. They are automatically checked and cleared by cron jobs.

[mdeneen]

From the other thread:

In the Zimbra Web interface, you need to press the "Junk" button to train the spam filters. Dragging a message to the "Junk" folder does not do the same thing.

There's already an enhancement request filed in bugzilla to make IMAP COPY to Junk train the spam filters. Please vote for it if you'd like to see it implemented!

Are you saying that if I move a message to the Junk folder in Thunderbird, that it will be processed by zmtrainsa? I tried this and no messages were processed. When I called "zmtrainsa userid@domain Spam Junk", the messages were processed.

Mark

[phoenix]

No, I was talking about the web client. If you want to do it from a client such a Thunderbird or Outlook then you need to forward them inline to the spam training account (and I mean the spam training account not the Junk folder) - it has been mentioned once or twice in the forums. There's also an RFE in bugzilla for a 'Junk folder' feature in Outlook (and possibly T'bird) - search and vote.

[mdeneen]

Originally Posted by phoenix

No, I was talking about the web client. If you want to do it from a client such a Thunderbird or Outlook then you need to forward them inline to the spam training account - it has been mentioned once or twice in the forums. There's also an RFE in bugzilla for a 'Junk folder' feature in Outlook (and possibly T'bird) - search and vote.

Ok, we're almost there!

I understand that you can forward inline or use the web client... but in the thread I linked to in my first post, it was mentioned that you could use zmtrainsa to train on a user spam folder. At the time, the issue was that zmtrainsa required a userid/password. This is no longer the case with the latest release of Zimbra.


I do not wish to give my users the ham/spam email addresses. What I am trying to do is manually train spamassassin for IMAP clients. When the RFE is complete, I will move over to the included solution.

My tests show that calling zmtrainsa on a particular folder appears to process the spam messages exactly like I expect it to. My only question is about what the --cleanup option does.

Thanks Again,
Mark

[dwmtractor]

Because if it is, we should be able to put a batch in crontab to do so, and it should process the contents of the folders that have been filtered by TBird or whatever other client people are using.

Bottom line, it looks like there are a bunch of us who want to use external clients instead of the web client and are looking for ways to make this easier. An auto-analysis of the contents of all junk folders, without regard to what client put the messages in there, would be a big step in that direction.

Dan

[gkra]

Originally Posted by dwmtractor

Bottom line, it looks like there are a bunch of us who want to use external clients instead of the web client and are looking for ways to make this easier. An auto-analysis of the contents of all junk folders, without regard to what client put the messages in there, would be a big step in that direction.

I have to agree with this. Most of our users prefer Thunderbird, Apple Mail, or Outlook for their daily email use, and it seems I've erroneously been telling them that "anything put in the Junk folder will be used to train the spam filters."

From my searching in the forums, it seems that that's not the case, which really isn't all that useful.

[gkra]

Originally Posted by phoenix

You don't need to use zmtrainsa against the Junk folder, anything that's in there will also have gone to the spam training account. They are automatically checked and cleared by cron jobs.

Given that most of our users are using IMAP clients, myself included, I guess the "checked and cleared" only applies to messages put in a user's Junk folder by using the web client's "Spam" button on a message.

Similarly, are messages which are auto-filed by Zimbra into the Junk folder considered for spam training and cleared?

Thanks,

Gregory

[phoenix]

Originally Posted by gkra

Similarly, are messages which are auto-filed by Zimbra into the Junk folder considered for spam training and cleared?

Yes, they are. They will have already gone through the spam checking system and are filed in the Junk folder (rather than removed as spam immediately) because of your spam settings. You should also check the Junk folder occasionally for any false positives.

There's an RFE for Junk folder functionality in IMAP clients, vote on that if you want the feature.

[gkra]

Originally Posted by phoenix

Yes, they are. They will have already gone through the spam checking system and are filed in the Junk folder (rather than removed as spam immediately) because of your spam settings.

Okay... There's a couple things I'm still not certain I understand, and I apologize if I'm being obtuse.

• Are the individual users' Junk folders supposed to be automatically cleared?
• If a message is marked as Junk by a user in the web client, should I see a copy of that message in the INBOX of the spam user account, or where is that collected?