MailScanner does rbl checks

[Scott Silva]

on 4/20/2011 7:49 AM George Elgin spake the following:
>>> Does this mean I'm not doing any RBL checking at all now? Or will SA
> 
> so with
>  
> Spam List = SPAMEATINGMONKEY,CBL,MANITU,spamhaus-ZEN
> and obviously set up rules here
> 
> spam.lists.conf:SPAMEATINGMONKEY                bl.spameatingmonkey.net
> 
> my objective is i don't want mails scanned that are clearly spam say FAIL more
> than one GOOD rbl ie.
> 
> Spam Lists To Reach High Score = 2,, in other words i thought the defaults are
> way too conservative//
> 
> my question though is "does this in fact cause MailScanner to NOT let SA spend
> time analysing these Mails ?".
> 
> OR
> 
> "would it be preferable to let postfix itself reject RBL'ed mails ?"
> 
> also for
> [gelgin at gfee MailScanner]$ grep skip_rbl_checks *
> spam.assassin.prefs.conf:# skip_rbl_checks      1
> 
> is 1 the default or do i uncomment if i don't want checks ? unlike amavis i am
> not getting a good feel that RBL checks are really happening...
> 
>     >> still do that?
>     >
>     > Yes, SA will still do those checks.
>     >
>     > That's the current recommended practice - using MailScanner to check
>     > RBLs is only advised in cases where you don't wish to use
>     > SpamAssassin but do wish to check multiple RBLs. 
> 
>     I've always done it the other way (MailScanner does the checks) but times
>     change.  To reverse it, is it as simple as changing these from
>     MailScanner.conf
>         Spam Lists To Be Spam = 1
>         Spam List = spamhaus-ZEN (or whatever one has in there)
>     and in the spam.assassin.prefs.conf setting skip_rbl_checks to 0 (zero)?
> 
>     Will the "Spam Lists To Be Spam" setting matter if spamassassin is doing
>     the checks instead of MailScanner?  If so, should that become zero as well?
> 
> 
If you trust the RBL, let the MTA or a milter reject. Once you receive and
close the connection, all you can safely do is discard, as any rejects after
this point can go to any innocent spoofed domains.