On Sat, Aug 26, 2006 at 03:04:59PM -0600, Gary V wrote:
> After I installed 0.1.13 beta over the top of 0.1.13 beta 1, I got:
> Aug 26 14:25:29 sfa postfix/policydweight[6299]: warning: cache: syntax
> error in file /etc/policyd-weight.conf: Global symbol "$BIND_ADDRESS"
> requires explicit package name at (eval 15) line 15.
Yes. "Old" versions cannot make use of new settings as the default value for
that variable must be initialized by policyd-weight. This is done by
calling "my $VAR" in plicyd-weight first, then evaluating the config file.
Old versions don't know that, so perl gets irritated. That's the flipsite of
using
use strict;
If you want to use "old" versions you need to remove "new" parameters from
the config file (otherwise it uses builtin defaults).
> This is after I stopped policyd-weight and the cache. Looks like I
> should have also deleted the cache file after doing so. A reboot must
> have done it for me. I assume /tmp/policyd-weight is the cache file I
> should have deleted (after stopping the processes).
Actually it should be /tmp/.poliyd-weight/polw.sock
I've noticed the bug in the sample config file and updated.
(NOTE: changing socket locations is always a bit not so flaw-less)
> ########################################
>
> I am confused about something. You say:
>
> $DEFER_ACTION = "450"; # possible values: DEFER_IF_PERMIT,
> # DEFER_IFREJECT,
> # 4xx response codes. See also access(5)
>
> $DEFER_LEVEL = 5; # DEFER mail only up to this level
> # scores greater than DEFER_LEVEL will be
> # rejected
>
> So I would think at a score of 1.5, policyd-weight would send a 450
> code, but I see a 550 code:
>
> Aug 26 14:31:54 sfa postfix/policydweight[2543]: weighted check:
> NOT_IN_SBL_XBL_SPAMHAUS=-1.5 NOT_IN_BL_NJABL=-1.5 NOT_IN_SPAMCOP=-1.5
> CL_IP_NE_HELO=1.5 RESOLVED_IP_IS_NOT_HELO=1.5 (check from:
> .example. - helo: .gary.) FROM_NOT_FAILED_HELO=3
> <client=192.168.1.41> <helo=gary> <[EMAIL PROTECTED]>
> <[EMAIL PROTECTED]>, rate: 1.5
>
> Aug 26 14:31:54 sfa postfix/policydweight[2543]: decided action=550
> Mail appeared to be SPAM or forged. Ask your Mail/DNS-Administrator
> to correct HELO and DNS MX settings or to get removed from DNSBLs;
> MTA helo: gary, MTA hostname: unknown[192.168.1.41] (helo/hostname
> mismatch)
>
> Maybe I don't understand. Personally, I would rather see the 5xx code
> as happened here.
You have not noticed:
$DEFER_STRING = "IN_SPAMCOP= BOGUS_MX=";
# a space separated case-sensitive list of
# strings on which if found in the $RET
# logging-string policyd-weight changes
# its action to $DEFER_ACTION in case
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
# of rejects.
# USE WITH CAUTION!
> and then manually whitelist chosen clients, I don't see why a
> mechanism like policyd-weight would ever want to say there has been a
> temporary failure.
This is to make exceptions for certain things which *could* mean a temporary
issue. Spamcop is such a case. BOGUS_MX is questionable, I've kept it because
of I was unsure whether such thing can be temporarily.
I.e. the above query would have let to a 4xx if the client would have been
in spamcop, and its score under 5.
This is because spamcop is *very* aggressive, but also very effective.
However, I am thinking about a way to chain DNSBL results with after-qeue
results from amavis (just to make optimal safe use of spamcop).
Problem with after-qeue (amavis) results is, I have seen mail from the
amavis-user and postfix-user list in the quarantine. So we need a way to make
safe use of after-queue results.
--
Robert Felber (PGP: 896CF30B)
Munich, Germany
____________________________________________________________
Policyd-weight Mailinglist - http://www.policyd-weight.org/
