Page 1 of 1

SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 15 Mar 2018 09:25
by cowboy6
Hey guys.

I've successfully set EFA as our antispam server. Now I would like to use SA learn. When I tag one email for SA learn as Spam, I get this message:
Image

Interesting is that under Recent messages is Status Learned as Spam.
Image

So my question is... Is this SA learn working correctly?

One more thing... Under Spam report I don't see any Bayes.
Image

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 19 Mar 2018 03:38
by pdwalker
Under "Tools and Links" is another link to "SpamAssassin Bayes Database Info". When you click on that link, what do you get?

Mine looks like:
Screen Shot 2018-03-19 at 11.37.png
Screen Shot 2018-03-19 at 11.37.png (59.23 KiB) Viewed 6353 times

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 19 Mar 2018 06:15
by cowboy6
Hi, this is our Bayes Database Information:

Image

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 19 Mar 2018 07:34
by henk

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 19 Mar 2018 08:10
by pdwalker
Also, the bayes database needs to be trained on a minimum number of messages before it starts showing results. Feed it more spam and ham (100 minimum?) and then your system will benefit from the bayesian analysis.

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 20 Mar 2018 07:33
by cowboy6
SA Learn config: path "/root/.spamassassin" is inaccessible was resolved with EFA server restart (link from henk).

For Bayes learning I will have to wait to receive more then 100 mails as said pdwalker.

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 16 Apr 2018 08:20
by cowboy6
Bayes stared to work after 200 detected spams.

Re: SA Learn config: path "/root/.spamassassin" is inaccessible

Posted: 16 Apr 2018 11:02
by pdwalker
Keep an eye on your BAYES scores. If you are confident the BAYES is helping, then you can increase the spamscore on the top end.

For example, I use a score of 4.0 for BAYES_99 and a score of 2.0 for BAYES_999 (bayes_999 is additive to bayes_99 on my system thus the final score is +6.0). These numbers are higher than the default and it seems to work well for me)