LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Server (https://www.linuxquestions.org/questions/linux-server-73/)
-   -   Migration of spamassassin settings from an old to a new server (https://www.linuxquestions.org/questions/linux-server-73/migration-of-spamassassin-settings-from-an-old-to-a-new-server-789182/)

voltron81 02-15-2010 03:13 AM

Migration of spamassassin settings from an old to a new server
 
Hi to everybody.
I'm migrating a mailserver from an old to a new mailserver.
The new mailserver is based on ISPConfig 3.0.6.1.

I would like to migrate what spamassassin learnt in the past from the old mailserver to the new one.
Now the question:
- The 2 spamassassin versions are different. Is it a problem?
- Can I just migrate the database with this commands???
Code:

sa-learn --backup > sadb.bkp

sa-learn --restore ./sadb.bkp

-how to migrate SpamAssasins Auto White List and all others SpamAssassins databases?

Thanks
Michele

business_kid 02-15-2010 05:10 AM

IIRC, your settings and rules are in /etc/mailspamassassin, and your install elsewhere. Spamassassin is more or less designed to be updated and to keep your unique rules to avoid this problem. Have you read the spamassassin doc on upgrading?

voltron81 02-16-2010 03:36 AM

Hi, thanks for your reply.
I've read this http://spamassassin.apache.org/full/.../sa-learn.html and this: http://spamassassin.apache.org/full/.../sa-learn.html

So if I've understand correctly, with sa-learn-backup I'll bakup just the Bayes database.
To have a better learn of spamassassin, I need to do also something like:
Code:

sa-learn --spam /path/to/spam/folder
      sa-learn --ham /path/to/ham/folder

Is it correct?
Sorry but I don't have previous experience with spamassassin policy migration and I hope somebody can put me on the right way
Thanks
Michele

business_kid 02-16-2010 04:56 AM

I set up a system with this stuff years back. I used
Vipul's Razor - that caught over half of the spam, and was very accurate. I may have had one false positive, but not two.
DCC caught maybe 5%, but it's aimed at bulk mail, rather than spam. There are tricks the spammers are using to get around these two. They sometimes succeed, but the programmers get more clever.
There's something similar to the razor in python on sourceforge. The name escapes me. It caught squat - nothing. It may have improved.
Spamassassin (spamc/spamd)I had ready to bite at anything, and the dns & server based rules I found more effective than the page scanning ones. Mails made in M$ frontpage with .exe files attached should score in the double figures. You can add some custom rulesets

The key to this is where spamassassin has it's databases (.db files) If you don't destroy those, you are ok - I don't think you even have to learn the new system. You can check by saving off mails individually (mail1, mail2, etc). If they changed the database format between versions then an upgrade is messy.

The spamc man page will give you the options but I think it's
cat mail1 |spamc -R
for an onscreen report that isn't added. So check some of your mails that way. There is something for removing the report
spamassassin -r(?)


All times are GMT -5. The time now is 12:26 AM.