Re: Update to recommended TLS settings

2015-08-07 Thread Luigi Rosa

Viktor Dukhovni wrote on 06/08/2015 09:08:


Recent updates to the supported Postfix releases have updated the
default settings of the OpenSSL ciphers used for opportunistic TLS
from "export" to "medium.


Viktor,
thank you so much for this mini-howto.

As an added security I rotate DH parameters every night, this should enforce 
better forward secrecy, am I correct?




--


Ciao,
luigi

/
+--[Luigi Rosa]--
\

The wireless music box has no imaginable commercial value. Who would
pay for a message sent to no one in particular?
  --Associates of David Sarnoff responding to the latter's call
for investment in the radio, 1921


unknown macro name in expansion request

2015-08-07 Thread Rob Sterenborg (Lists)

Hi,

This Postfix 3.0.2.

I see this in maillog:

Aug 7 09:43:12 pfxp001 postfix/smtpd[18610]: warning: unknown macro name 
"myhostname_ext" in expansion request


Aug  7 09:43:34 pfxp001 postfix/postscreen[12699]: warning: unknown 
macro name "myhostname_ext" in expansion request



However:

# postconf mydomain_ext
mydomain_ext = .nl
# postconf myhostname_ext
myhostname_ext = mx2.$mydomain_ext


I'm also having mydomain_int and myhostname_int, but it's not 
complaining about that.


What is it I'm overlooking?


# postconf -n | grep myhostname_ext
myhostname_ext = mx2.$mydomain_ext
smtp_helo_name = $myhostname_ext
smtpd_banner = $myhostname_ext
smtpd_reject_footer = For assistance, call +31 . Please 
provide the following information: time ($localtime), client 
($client_address) and sending server ($myhostname_ext).


# postconf -n | grep myhostname_int
myhostname = $myhostname_int
myhostname_int = pfxp001.$mydomain_int
myorigin = $myhostname_int


==
Rob


Re: Update to recommended TLS settings

2015-08-07 Thread DTNX Postmaster
On 07 Aug 2015, at 06:14, Viktor Dukhovni  wrote:

> On Fri, Aug 07, 2015 at 02:55:42AM +0200, DTNX Postmaster wrote:
> 
>> For most systems, monitoring the status of their encryption just isn't 
>> done at all; they use the defaults their device or server came with at 
>> the time they purchased it, and rarely keep up with the times.
> 
> They don't need to.  There's nothing wrong with outdated crypto on
> systems that wouldn't even encrypt if encryption weren't on by
> default.  We'll get more decent security through a natural process
> of deployment of more capable systems and retirement of less capable
> systems.  Eventually, there'll be no demand for RC4 (for example),
> and we'll be able to turn it off with no noticeable degradation to
> cleartext.  Later still (another ~5 years?) we'll be able to turn
> off TLS 1.0...

Except when they DO need to; Microsoft, for example, has shipped 
Schannel with horrid defaults for a long time, which affects not just 
SMTP, but also the authenticated users on those systems. Many of those 
are not kept up to date, which leads to servers that actually support 
TLSv1.2, but offer only SSLv2, SSLv3 and TLSv1 in practice, with a 
suboptimal set of ciphers. Because it's turned off by default.

Arguing that these systems do not need to be maintained until they are 
replaced does us all a disservice. The era of Set and Forget is over; 
if you connect something to the Internet, maintain it as long as it has 
that connection.

Mvg,
Joni



Re: unknown macro name in expansion request

2015-08-07 Thread Wietse Venema
Rob Sterenborg (Lists):
> Hi,
> 
> This Postfix 3.0.2.
> 
> I see this in maillog:
> 
> Aug 7 09:43:12 pfxp001 postfix/smtpd[18610]: warning: unknown macro name 
> "myhostname_ext" in expansion request
> 
> Aug  7 09:43:34 pfxp001 postfix/postscreen[12699]: warning: unknown 
> macro name "myhostname_ext" in expansion request

As documented (!) smtpd_reject_footer does not expand main.cf
parameter settings.


   Notes:

   o  NOT SUPPORTED are other attributes such as sender, recipient, or
  main.cf parameters.



Wietse




Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread Vincent Pelletier
Hello,

I have a Mailman + postfix setup (old versions: mailman 2.1.11,
postfix 2.5.5), and the mail server of one of subscribers' domain
started rejecting mails when "From" header domain is its own domain
(foo.com MX rejecting incoming @foo.com mails). This of course happens
when a @foo.com member posts to the list: other @foo.com subscribers
never receive his post.

From my understanding of the "From" header, it is legal for it to have
a domain which is not hosted by sender (ie, the SMTP transaction
initiator).

So my plan is to change nothing on the Mailman level, but rather add an
exception in our postfix config to mangle "From" header of mails being
sent to that server.

In my understanding, I need 2 distinct postfix features:
- transport_maps for the "recipient-dependent" aspect
- header_check for the mangling part

Is my understanding correct so far ?

Here is how I implemented the above:

main.cf:
transport_maps = hash:/etc/postfix/transport_map

transport_map:
foo.com mangle_from:

master.cf:
mangle_from unix - - - - - smtpd
  -o header_checks=pcre:/etc/postfix/mangle_from.pcre

mangle_from.pcre:
/^From:[[:space:]]+.*@foo.com/ REPLACE From: 

But this fails between transport_map and master.cf:
mail.log:
Aug  7 12:56:46 ns208770 postfix/oqmgr[4535]: 2B67ABF017B: 
from=, size=5497, nrcpt=1 (queue active)
Aug  7 13:01:46 ns208770 postfix/error[22126]: 2B67ABF017B: to=<...@foo.com>, 
relay=none, delay=1833, delays=1533/300/0/0.16, dsn=4.3.0, status=deferred 
(mail transport unavailable)

mail.warn:
Aug  7 13:01:46 ns208770 postfix/oqmgr[4535]: warning: unexpected end-of-input 
from private/mangle_from socket while reading input attribute name
Aug  7 13:01:46 ns208770 postfix/oqmgr[4535]: warning: private/mangle_from 
socket: malformed response

What did I do wrong ?
Can I do anything without having to upgrade right now ?

Regards,
-- 
Vincent Pelletier
ERP5 - open source ERP/CRM for flexible enterprises


smime.p7s
Description: S/MIME cryptographic signature


Re: Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread wilfried.es...@essignetz.de
Hi,


maybe this, from postfix doc, will help you:

http://www.postfix.org/ADDRESS_REWRITING_README.html#generic


Willi



Am 07.08.2015 um 13:51 schrieb Vincent Pelletier:
> Hello,
> 
> I have a Mailman + postfix setup (old versions: mailman 2.1.11,
> postfix 2.5.5), and the mail server of one of subscribers' domain
> started rejecting mails when "From" header domain is its own domain
> (foo.com MX rejecting incoming @foo.com mails). This of course happens
> when a @foo.com member posts to the list: other @foo.com subscribers
> never receive his post.
> 
> From my understanding of the "From" header, it is legal for it to have
> a domain which is not hosted by sender (ie, the SMTP transaction
> initiator).
> 
> So my plan is to change nothing on the Mailman level, but rather add an
> exception in our postfix config to mangle "From" header of mails being
> sent to that server.
> 
> In my understanding, I need 2 distinct postfix features:
> - transport_maps for the "recipient-dependent" aspect
> - header_check for the mangling part
> 
> Is my understanding correct so far ?
> 
> Here is how I implemented the above:
> 
> main.cf:
> transport_maps = hash:/etc/postfix/transport_map
> 
> transport_map:
> foo.com mangle_from:
> 
> master.cf:
> mangle_from unix - - - - - smtpd
>   -o header_checks=pcre:/etc/postfix/mangle_from.pcre
> 
> mangle_from.pcre:
> /^From:[[:space:]]+.*@foo.com/ REPLACE From: 
> 
> But this fails between transport_map and master.cf:
> mail.log:
> Aug  7 12:56:46 ns208770 postfix/oqmgr[4535]: 2B67ABF017B: 
> from=, size=5497, nrcpt=1 (queue active)
> Aug  7 13:01:46 ns208770 postfix/error[22126]: 2B67ABF017B: to=<...@foo.com>, 
> relay=none, delay=1833, delays=1533/300/0/0.16, dsn=4.3.0, status=deferred 
> (mail transport unavailable)
> 
> mail.warn:
> Aug  7 13:01:46 ns208770 postfix/oqmgr[4535]: warning: unexpected 
> end-of-input from private/mangle_from socket while reading input attribute 
> name
> Aug  7 13:01:46 ns208770 postfix/oqmgr[4535]: warning: private/mangle_from 
> socket: malformed response
> 
> What did I do wrong ?
> Can I do anything without having to upgrade right now ?
> 
> Regards,
> 



Re: Update to recommended TLS settings

2015-08-07 Thread Viktor Dukhovni
On Fri, Aug 07, 2015 at 10:24:34AM +0200, Luigi Rosa wrote:

> >Recent updates to the supported Postfix releases have updated the
> >default settings of the OpenSSL ciphers used for opportunistic TLS
> >from "export" to "medium.
> 
> Viktor,
> thank you so much for this mini-howto.
> 
> As an added security I rotate DH parameters every night, this should enforce
> better forward secrecy, am I correct?

Yes, especially for 512-bit parameters (if EXPORT is still enabled).

Nightly is likely substantially more often than necessary for
1024-bit parameters, but they're cheap enough to generate.

I think many cryptographers would be surprised if 2048-bit prime
EDH groups were already today vulnerable to practical pre-computation
attacks, but there's no harm in rotating these also.

-- 
Viktor.


Re: postscreen dnsbl weighting with new(est) spamhaus return codes -- experiences/data?

2015-08-07 Thread PGNd
On Thu, Aug 6, 2015, at 10:22 AM, Chris Adams wrote:
> Those are dbl response codes, not zen.  You are mixing the two up

And you'd be correct!  pebkac.

thanks.



Re: Update to recommended TLS settings

2015-08-07 Thread Blake Hudson

Thank you Viktor.


Re: Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread Vincent Pelletier
On Fri, 07 Aug 2015 17:10:11 +0200,
"wilfried.es...@essignetz.de"  wrote :
> maybe this, from postfix doc, will help you:
> 
> http://www.postfix.org/ADDRESS_REWRITING_README.html#generic

The problem with this is that it will rewrite "To:" field too, so
sending to ...@foo.com pushes to their SMTP a mail for
nob...@nexedi.com - which they rightfully reject as a relay attempt.

Or I botched my configuration and/or missed an extra option.

Regards,
-- 
Vincent Pelletier
ERP5 - open source ERP/CRM for flexible enterprises


smime.p7s
Description: S/MIME cryptographic signature


Re: Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread wilfried.es...@essignetz.de
Am 07.08.2015 um 18:24 schrieb Vincent Pelletier:
> On Fri, 07 Aug 2015 17:10:11 +0200,
> "wilfried.es...@essignetz.de"  wrote :
>> maybe this, from postfix doc, will help you:
>>
>> http://www.postfix.org/ADDRESS_REWRITING_README.html#generic
> 
> The problem with this is that it will rewrite "To:" field too, so
> sending to ...@foo.com pushes to their SMTP a mail for
> nob...@nexedi.com - which they rightfully reject as a relay attempt.

That makes me wonder. Do they know the difference  between "envelope-to"
and "to"-field, or "envelope-from" and "from"-field? (This makes me
remember to the beginning of my fechmail era - they do not use fetchmail
or something similar?)

Usually smtp-Servers should only look to the the "envelope-to/-from",
which they exchange during the smtp dialogue. If a smtp-server is
checking only the from and to fields in the header of the received mail,
they will loose receipient information. That concerns not only mail from
lists, but also mail sent with adressess in BCC.

Possibly, you should try to talk to this special admin. Looks like he's
violating some really important rfc's.


Willi

> 
> Or I botched my configuration and/or missed an extra option.
> 
> Regards,
> 




Re: Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread Vincent Pelletier
On Fri, 07 Aug 2015 18:53:49 +0200,
"wilfried.es...@essignetz.de"  wrote :
> That makes me wonder. Do they know the difference  between "envelope-to"
> and "to"-field, or "envelope-from" and "from"-field? (This makes me
> remember to the beginning of my fechmail era - they do not use fetchmail
> or something similar?)

Good question, I do not know (and will have to wait for next Monday to
ask). I assume they do this to protect against social engineering (I
assume most MUA and users would mostly only pay attention to "From:"
header), so filtering based on the header would make some sense - but I
didn't check what envelope values are & their effect on rejection.

> Usually smtp-Servers should only look to the the "envelope-to/-from",
> which they exchange during the smtp dialogue. If a smtp-server is
> checking only the from and to fields in the header of the received mail,
> they will loose receipient information. That concerns not only mail from
> lists, but also mail sent with adressess in BCC.
> 
> Possibly, you should try to talk to this special admin. Looks like he's
> violating some really important rfc's.

This is my understanding of the issue so far, indeed.

Regards,
-- 
Vincent Pelletier
ERP5 - open source ERP/CRM for flexible enterprises


smime.p7s
Description: S/MIME cryptographic signature


Re: Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread Viktor Dukhovni
On Fri, Aug 07, 2015 at 01:51:34PM +0200, Vincent Pelletier wrote:

> foo.com mangle_from:
> 
> master.cf:
> mangle_from unix - - - - - smtpd
>   -o header_checks=pcre:/etc/postfix/mangle_from.pcre

Don't confuse "smtp" with "smtpd".

$ postconf -Mf smtp/unix
smtp   unix  -   -   n   -   -   smtp

replace the first "smtp" with "mangle_from", keep the rest the same
(unless you want "chroot"), and add a suitably indented "-o" line.

-- 
Viktor.


Re: Mailing list sending mail to server with strong anti-spoofing setting

2015-08-07 Thread Wietse Venema
Vincent Pelletier:
> Hello,
> 
> I have a Mailman + postfix setup (old versions: mailman 2.1.11,
> postfix 2.5.5), and the mail server of one of subscribers' domain
> started rejecting mails when "From" header domain is its own domain
> (foo.com MX rejecting incoming @foo.com mails). This of course happens
> when a @foo.com member posts to the list: other @foo.com subscribers
> never receive his post.

I suspect that they are enforcing a DMARC policy that requires
modifications to list managers so that they replace the From: header
address with the sender address of the mailing list.

Google for "dmarc mailman fix".

Wietse