From stuart at bmsi.com  Tue Sep  2 13:50:59 2008
From: stuart at bmsi.com (Stuart D. Gathman)
Date: Tue, 02 Sep 2008 13:50:59 -0400
Subject: [Pymilter] Re: Bug report for PySPF
In-Reply-To: <7FD163CA-016C-4F38-91C9-33620FBEBC4F@majid.org>
References: <7FD163CA-016C-4F38-91C9-33620FBEBC4F@majid.org>
Message-ID: <48BD7D03.1070909@bmsi.com>

Fazal Majid wrote:
> PySPF is failing on AOL SPF checks. This might be due to the fact AOL 
> has 2 SPF records.
>
> gondwana ~>dig txt aol.com
> ;; Warning: Message parser reports malformed message packet.
> ;; Truncated, retrying in TCP mode.
Large TXT result sets that require TCP didn't work in pyspf prior to 
2.0.5.  The exception from pydns suggests an old version of that as 
well.  Here is what I get with pyspf-2.0.5 and pydns-2.3.3

[stuart at cmsmail ~]$ spfquery --ip 123.23.24.25 --sender shkhadri at aol.com
neutral

spfquery: 123.23.24.25 is neither permitted nor denied by domain of aol.com
Received-SPF: Neutral (spfquery: 123.23.24.25 is neither permitted nor 
denied by domain of aol.com) client-ip=123.23.24.25; 
envelope-from="shkhadri at aol.com"; receiver=spfquery; mechanism=?all; 
identity=mailfrom

pyspf-2.0.4 and pydns-2.3.2 would get a 'none' result with no exception 
due to the packet truncation (TCP not supported).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: stuart.vcf
Type: text/x-vcard
Size: 315 bytes
Desc: not available
URL: <http://gathman.org/pipermail/pymilter/attachments/20080902/2bcc62f4/attachment.vcf>

From stuart at bmsi.com  Mon Sep  8 11:03:09 2008
From: stuart at bmsi.com (Stuart D. Gathman)
Date: Mon, 8 Sep 2008 11:03:09 -0400 (EDT)
Subject: [Pymilter] Re: Python-milter SIGHUP or restart
In-Reply-To: <20080908121202.GB2627@salstar.sk>
References: <20080908121202.GB2627@salstar.sk>
Message-ID: <Pine.LNX.4.64.0809081047000.28376@bmsred.bmsi.com>

On Mon, 8 Sep 2008, Jan ONDREJ (SAL) wrote:

>   I am using pymilter in my project SAGATOR. An user reported me a problem,
> that after this process receives an SIGHUP signal, process is ended and do
> now work more.
> 
>   This behaviour looks to be standard for milter and is hardcoded. I can't
> overwrite SIGHUP handler, because it's still overwritten in milter.

That is correct, libmilter takes over signal handling. 

>   But problem starts again, when I am trying to restart .runmilter().
> Next start in same process does not work. Is it a bug of python-milter or
> why this happens? An example scripts is attached to this email.

I've never seen a C milter that restarts runmilter() in the same process.

>   I am using SIGHUP to rotare log files. It's required for proper
> functionality of SAGATOR. Changing this signal may be hard or impossible,
> because there are other services, which use signals in sagator.
> 
>   I need to fix one of these problems to make my project functional again:
>     - my own SIGHUP handler
>     - ability to restart .runmilter()

-- 
	      Stuart D. Gathman <stuart at bmsi.com>
    Business Management Systems Inc.  Phone: 703 591-0911 Fax: 703 591-6154
"Confutatis maledictis, flammis acribus addictis" - background song for
a Microsoft sponsored "Where do you want to go from here?" commercial.


From stuart at bmsi.com  Mon Sep  8 11:16:01 2008
From: stuart at bmsi.com (Stuart D. Gathman)
Date: Mon, 8 Sep 2008 11:16:01 -0400 (EDT)
Subject: [Pymilter] Re: Python-milter SIGHUP or restart
In-Reply-To: <Pine.LNX.4.64.0809081047000.28376@bmsred.bmsi.com>
References: <20080908121202.GB2627@salstar.sk> <Pine.LNX.4.64.0809081047000.28376@bmsred.bmsi.com>
Message-ID: <Pine.LNX.4.64.0809081112270.28803@bmsred.bmsi.com>

On Mon, 8 Sep 2008, Stuart D. Gathman wrote:

> >   I am using SIGHUP to rotare log files. It's required for proper
> > functionality of SAGATOR. Changing this signal may be hard or impossible,
> > because there are other services, which use signals in sagator.
> > 
> >   I need to fix one of these problems to make my project functional again:
> >     - my own SIGHUP handler
> >     - ability to restart .runmilter()

I will research whether smfi_main (and hence runmilter) can be made
restartable (and if not, why not), but why not wrap your milter in a
subprocess?

-- 
	      Stuart D. Gathman <stuart at bmsi.com>
    Business Management Systems Inc.  Phone: 703 591-0911 Fax: 703 591-6154
"Confutatis maledictis, flammis acribus addictis" - background song for
a Microsoft sponsored "Where do you want to go from here?" commercial.