Prev: SMTP Greet Pause (re: Feature 20090228)
Next: Coping with relayed spam?
From: Mikael Bak on 8 Sep 2009 05:47
Steve Heaven wrote:
> On Mon, 2009-09-07 at 11:50 -0400, Sahil Tandon wrote:
>
>>
>> You should not accept mail for invalid recipients. Use existing
>> functionality to build a cache/database of valid recipients "on the fly".
>> See: http://www.postfix.org/ADDRESS_VERIFICATION_README.html#recipient
>
> We have no way of knowing if the recipient address is valid or not as we
> are only acting as a relay for the final destination.
> We cannot build a database of recipients on the fly as that information
> is held on the various servers of our clients, to which we do not have
> access.
>

Sahil Tandon gave you a link containing the solution to you problem. I
suggest you read it before you say it can't be done.

Tip: scoll up to "How address verification works".

Mikael
From: Clunk Werclick on 8 Sep 2009 05:57

From:
Clunk Werclick
<mailbackup19(a)googlemail.com>
Reply-to:
mailbackup19(a)googlemail.com
Cc:
postfix-users(a)postfix.org
Subject:
Re: relay_domains
vs
virtual_mailbox_domains
Date:
Tue, 08 Sep 2009
09:28:36 +0100
Mailer:
Evolution 2.24.3



On Tue, 2009-09-08 at 08:52 +0100, Steve Heaven wrote:
> On Mon, 2009-09-07 at 11:50 -0400, Sahil Tandon wrote:
>
> >
> > You should not accept mail for invalid recipients. Use existing
> > functionality to build a cache/database of valid recipients "on the
fly".
> > See:
http://www.postfix.org/ADDRESS_VERIFICATION_README.html#recipient
>
> We have no way of knowing if the recipient address is valid or not as
> we are only acting as a relay for the final destination.
> We cannot build a database of recipients on the fly as that
> information is held on the various servers of our clients, to which we
> do not have access.
>
Please forgive the bluntness - and drifting off a bit as I've not seen
all of this; If you are acting as a relay and not able to verify the
final recipients exist - you will quickly run into serious problems and
side effects.

Postfix provides a probing/discovery mechanism that spares you the need
to build maps - it's not ideal when compared to the sheer speed of SQL,
MAPS or LDAP, but it exists - so there is no excuse to accept mail for
invalid recipients with Postfix. The link given tells you how this
'probing' works.

Failing to verify final recipients means you will probably accept mail
that is sequentially refused, leaving you holding the baby and having to
bounce it. (Old Chinese Proverb say, man who gives 250 OK to SMTP, take
ownership and responsibility). With invalid recipients, the sender is
usually forged and as your relay has nothing left to do but bounce the
message, your IP(s) are going to become really unpopular *fast*, and
probably have it blacklisted in no time at all.

This is, of course, not only limited to invalid recipients. Accepting
any kind of mail for a destination that cannot be delivered gives the
same problem. Perhaps the recipient is valid, but the destination
refused the message because of the content/spam. You end up holding the
baby again.

If you really need the ability to catch all without bounce then the
final destination needs to absolutely white list everything your throw
at it - regardless of recipient or content. That is most certainly *not*
ideal without some serious UCE measures on the relay itself.

In commercial solutions I have seen, RELAYS have held the message and
not given a 250 until the final destination has taken it -or- (less
ideal) taken the message and put it into an 'outbound' Postfixen where
it is retried for 48-72 hours. This gives the Relay admin time to see it
and liase with the final destination host admin. This would be a real
headache if you wind up with thousands of messages in the queue for
invalid recipients, bringing us full circle to the topic once more.

Good luck with what it is you are doing.


--
-----------------------------------------------------------
C Werclick .Lot
Technical incompetent
Loyal Order Of The Teapot.

This e-mail and its attachments is intended only to be used as an e-mail
and an attachment. Any use of it for other purposes other than as an
e-mail and an attachment will not be covered by any warranty that may or
may not form part of this e-mail and attachment.



From: mouss on 8 Sep 2009 18:27
Steve Heaven a écrit :
> On Mon, 2009-09-07 at 11:50 -0400, Sahil Tandon wrote:
>
>>
>> You should not accept mail for invalid recipients. Use existing
>> functionality to build a cache/database of valid recipients "on the fly".
>> See: http://www.postfix.org/ADDRESS_VERIFICATION_README.html#recipient
>
> We have no way of knowing if the recipient address is valid or not as we
> are only acting as a relay for the final destination.
> We cannot build a database of recipients on the fly as that information
> is held on the various servers of our clients, to which we do not have
> access.


the old: "try to pass to next, until final server accepts or rejects"
is n more acceptable. recipients must be checked at the "edge".

postfix provides reject_unverified_recipient to help you for that
(assuming the next relay really validates the recipient).

It's been a time that most people acknowledge that backscatter is a
problem. those who take a selfish approach to mail should not be
surprised if they are blacklisted, and should not ask for help.
From: mouss on 8 Sep 2009 18:28
yar mailer got borked?

Clunk Werclick a �crit :
> From:
> Clunk Werclick
> <mailbackup19(a)googlemail.com>
> Reply-to:
> mailbackup19(a)googlemail.com
> Cc:
> postfix-users(a)postfix.org
> Subject:
> Re: relay_domains
> vs
> virtual_mailbox_domains
> Date:
> Tue, 08 Sep 2009
> 09:28:36 +0100
> Mailer:
> Evolution 2.24.3
>
>
>
> On Tue, 2009-09-08 at 08:52 +0100, Steve Heaven wrote:
>> On Mon, 2009-09-07 at 11:50 -0400, Sahil Tandon wrote:
>>
>>> You should not accept mail for invalid recipients. Use existing
>>> functionality to build a cache/database of valid recipients "on the
> fly".
>>> See:
> http://www.postfix.org/ADDRESS_VERIFICATION_README.html#recipient
>> We have no way of knowing if the recipient address is valid or not as
>> we are only acting as a relay for the final destination.
>> We cannot build a database of recipients on the fly as that
>> information is held on the various servers of our clients, to which we
>> do not have access.
>>
> Please forgive the bluntness - and drifting off a bit as I've not seen
> all of this; If you are acting as a relay and not able to verify the
> final recipients exist - you will quickly run into serious problems and
> side effects.
>
> Postfix provides a probing/discovery mechanism that spares you the need
> to build maps - it's not ideal when compared to the sheer speed of SQL,
> MAPS or LDAP, but it exists - so there is no excuse to accept mail for
> invalid recipients with Postfix. The link given tells you how this
> 'probing' works.
>
> Failing to verify final recipients means you will probably accept mail
> that is sequentially refused, leaving you holding the baby and having to
> bounce it. (Old Chinese Proverb say, man who gives 250 OK to SMTP, take
> ownership and responsibility). With invalid recipients, the sender is
> usually forged and as your relay has nothing left to do but bounce the
> message, your IP(s) are going to become really unpopular *fast*, and
> probably have it blacklisted in no time at all.
>
> This is, of course, not only limited to invalid recipients. Accepting
> any kind of mail for a destination that cannot be delivered gives the
> same problem. Perhaps the recipient is valid, but the destination
> refused the message because of the content/spam. You end up holding the
> baby again.
>
> If you really need the ability to catch all without bounce then the
> final destination needs to absolutely white list everything your throw
> at it - regardless of recipient or content. That is most certainly *not*
> ideal without some serious UCE measures on the relay itself.
>
> In commercial solutions I have seen, RELAYS have held the message and
> not given a 250 until the final destination has taken it -or- (less
> ideal) taken the message and put it into an 'outbound' Postfixen where
> it is retried for 48-72 hours. This gives the Relay admin time to see it
> and liase with the final destination host admin. This would be a real
> headache if you wind up with thousands of messages in the queue for
> invalid recipients, bringing us full circle to the topic once more.
>
> Good luck with what it is you are doing.
>
>

From: Steve Heaven on 9 Sep 2009 02:16
On Wed, 2009-09-09 at 00:27 +0200, mouss wrote:
> Steve Heaven a �crit :
> >
>
>
> the old: "try to pass to next, until final server accepts or rejects"
> is n more acceptable. recipients must be checked at the "edge".
>
> postfix provides reject_unverified_recipient to help you for that
> (assuming the next relay really validates the recipient).

That's the problem. Most of our clients that we relay mail for run
Microsoft SBS Exchange which doesnt verify probes. It accepts mail for
any user and sends an undeliverable report back to the sender.

--
thorNET
Internet Services, Consultancy & Training
www.thornet.co.uk

First  |  Prev  |  Next  |  Last
Pages: 1 2 3
Prev: SMTP Greet Pause (re: Feature 20090228)
Next: Coping with relayed spam?