Re: reject_unknown_sender_domain problem

10 msg	My Virtual users can't send messages to their o...
7 msg	AND parameters of Postfix question ?
2 msg	Px messing plain mail addresses, group members
4 msg	no logs
8 msg	myhostname
1 msg	how to manage destination status cache
2 msg	Temporary failure with Postfix and Cyrus
21 msg	/usr/lib/sasl2 failed to open directory
1 msg	Cyrus and Postfix

reject_unknown_sender_domain problem
\ Benny Pedersen (26 May 2007)
. \ Magnus Bäck (26 May 2007)
. \ Noel Jones (28 May 2007)

6 msg	Automatic message
14 msg	mailbox_command prepends a 'From ' ?
3 msg	virtual and lmtp for virtual hosting
10 msg	Still problems..
4 msg	Double transport
2 msg	MUA with SSL/TLS Client Certificate support (SO...
3 msg	Regarding S/MIME Messages
17 msg	address_verify_sender
10 msg	Question about Logs
9 msg	problem with bad syntax and Outlook 2007

Subject:	Re: reject_unknown_sender_domain problem
Group:	Postfix-users
From:	Noel Jones
Date:	28 May 2007

At 03:15 AM 5/26/2007, Benny Pedersen wrote:

>does it check mx ?
>does or just a ?
>
>both ?
>
>i see some envelope with just mx being rejected :/

If you check the docs you will see that postfix checks for existence
of either an A or MX record for the envelope sender domain. Seems to
work as advertised for me.

Some possibilities why you see unexpected behavior...
- temporary DNS errors. When postfix tried the lookup no result was
available. Postfix will always "defer" rather than "reject" if a
temporary error is detected. When you test later, the record is available.
- broken DNS server/forwarder. Some DNS implementations,
particularly those some in consumer-grade routers, seem to not work
very well. Make sure you use a "real" DNS server, preferably run a
caching DNS service on your mail server or local network.
- misinterpretation of evidence. Post unaltered evidence of your
suspected false positives for peer review.

You seem to be complaining about this same issue in SpamAssassin, yet
you don't post any evidence to the sa-users list either.

Claims of incorrect behavior without showing /any/ evidence is not
likely to be considered credible. Especially when the feature you
question is both widely used and widely accepted to work correctly.

We would be glad to help you understand what you see in your logs if
you care to share complete information.

--
Noel Jones