dist/proto/POSTSCREEN_README.html

<!doctype html public "-//W3C//DTD HTML 4.01 Transitional//EN"
        "http://www.w3.org/TR/html4/loose.dtd">

<head>

<title>Postfix Postscreen Howto</title>

<meta http-equiv="Content-Type" content="text/html; charset=utf-8">

</head>

<body>

<h1><img src="postfix-logo.jpg" width="203" height="98" ALT="">Postfix Postscreen Howto</h1>

<hr>

<h2> <a name="intro">Introduction</a> </h2>

<p> This document describes features that are available in Postfix
3.6 and later. See <a href="POSTSCREEN_3_5_README.html">
POSTSCREEN_3_5_README.html</a> for Postfix versions 2.8 - 3.5. </p>

<p> The Postfix postscreen(8) daemon provides additional protection
against mail server overload. One postscreen(8) process handles
multiple inbound SMTP connections, and decides which clients may
talk to a Postfix SMTP server process.  By keeping spambots away,
postscreen(8) leaves more SMTP server processes available for
legitimate clients, and delays the onset of <a
href="STRESS_README.html">server overload</a> conditions. </p>

<p> postscreen(8) should not be used on SMTP ports that receive
mail from end-user clients (MUAs). In a typical deployment,
postscreen(8) handles the MX service on TCP port 25, while MUA
clients submit mail via the submission service on TCP port 587 which
requires client authentication. Alternatively, a site could set up
a dedicated, non-postscreen, "port 25" server that provides submission
service and client authentication, but no MX service.  </p>

<p> postscreen(8) maintains a temporary allowlist for clients that
pass its tests; by allowing allowlisted clients to skip tests,
postscreen(8) minimizes its impact on legitimate email traffic.
</p>

<p> postscreen(8) is part of a multi-layer defense. <p>

<ul>

<li> <p> As the first layer, postscreen(8) blocks connections from
zombies and other spambots that are responsible for about 90% of
all spam.  It is implemented as a single process to make this defense
as inexpensive as possible. </p>

<li> <p> The second layer implements more complex SMTP-level access
checks with <a href="SMTPD_ACCESS_README.html">Postfix SMTP servers</a>,
<a href="SMTPD_POLICY_README.html">policy daemons</a>, and
<a href="MILTER_README.html">Milter applications</a>. </p>

<li> <p> The third layer performs light-weight content inspection
with the Postfix built-in header_checks and body_checks. This can
block unacceptable attachments such as executable programs, and
worms or viruses with easy-to-recognize signatures. </p>

<li> <p> The fourth layer provides heavy-weight content inspection
with external content filters. Typical examples are <a
href="http://www.ijs.si/software/amavisd/">Amavisd-new</a>, <a
href="http://spamassassin.apache.org/">SpamAssassin</a>, and <a
href="MILTER_README.html">Milter applications</a>. </p>

</ul>

<p> Each layer reduces the spam volume. The general strategy is to
use the less expensive defenses first, and to use the more expensive
defenses only for the spam that remains. </p>

<p> Topics in this document: </p>

<ul>

<li> <a href="#intro">Introduction</a>

<li> <a href="#basic">The basic idea behind postscreen(8)</a>

<li> <a href="#general"> General operation </a>

<li> <a href="#quick">Quick tests before everything else</a>

<li> <a href="#before_220"> Tests before the 220 SMTP server greeting </a>

<li> <a href="#after_220">Tests after the 220 SMTP server greeting</a>

<li> <a href="#other_error">Other errors</a>

<li> <a href="#victory">When all tests succeed</a>

<li> <a href="#config"> Configuring the postscreen(8) service</a>

<li> <a href="#historical"> Historical notes and credits </a>

</ul>

<h2> <a name="basic">The basic idea behind postscreen(8)</a> </h2>

<p> Most email is spam, and most spam is sent out by zombies (malware
on compromised end-user computers).  Wietse expects that the zombie
problem will get worse before things improve, if ever. Without a
tool like postscreen(8) that keeps the zombies away, Postfix would be
spending most of its resources not receiving email. </p>

<p> The main challenge for postscreen(8) is to make an is-a-zombie
decision based on a single measurement. This is necessary because
many zombies try to fly under the radar and avoid spamming the same
site repeatedly.  Once postscreen(8) decides that a client is
not-a-zombie, it allowlists the client temporarily to avoid further
delays for legitimate mail. </p>

<p> Zombies have challenges too: they have only a limited amount
of time to deliver spam before their IP address becomes denylisted.
To speed up spam deliveries, zombies make compromises in their SMTP
protocol implementation.  For example, they speak before their turn,
or they ignore responses from SMTP servers and continue sending
mail even when the server tells them to go away. </p>

<p> postscreen(8) uses a variety of measurements to recognize
zombies.  First, postscreen(8) determines if the remote SMTP client
IP address is denylisted.  Second, postscreen(8) looks for protocol
compromises that are made to speed up delivery.  These are good
indicators for making is-a-zombie decisions based on single
measurements.  </p>

<p> postscreen(8) does not inspect message content. Message content
can vary from one delivery to the next, especially with clients
that (also) send legitimate email.  Content is not a good indicator
for making is-a-zombie decisions based on single measurements,
and that is the problem that postscreen(8) is focused on.  </p>

<h2> <a name="general"> General operation </a> </h2>

<p> For each connection from an SMTP client, postscreen(8) performs
a number of tests
in the order as described below.  Some tests introduce a delay of
a few seconds.  postscreen(8) maintains a temporary allowlist for
clients that pass its tests; by allowing allowlisted clients to
skip tests, postscreen(8) minimizes its impact on legitimate email
traffic.  </p>

<p> By default, postscreen(8) hands off all connections to a Postfix
SMTP server process after logging its findings. This mode is useful
for non-destructive testing. </p>

<p> In a typical production setting, postscreen(8) is configured
to reject mail from clients that fail one or more tests, after
logging the helo, sender and recipient information. </p>

<p> Note: postscreen(8) is not an SMTP proxy; this is intentional.
The purpose is to keep zombies away from Postfix, with minimal
overhead for legitimate clients. </p>

<h2> <a name="quick">Quick tests before everything else</a> </h2>

<p> Before engaging in SMTP-level tests. postscreen(8) queries a
number of local deny and allowlists. These tests speed up the
handling of known clients. </p>

<ul>

<li> <a href="#perm_allow_deny"> Permanent allow/denylist test </a>

<li> <a href="#temp_allow"> Temporary allowlist test </a>

<li> <a href="#allow_veto"> MX Policy test </a>

</ul>

<h3> <a name="perm_allow_deny"> Permanent allow/denylist test </a> </h3>

<p> The postscreen_access_list parameter (default: permit_mynetworks)
specifies a permanent access list for SMTP client IP addresses. Typically
one would specify something that allowlists local networks, followed
by a CIDR table for selective allow- and denylisting. </p>

<p> Example: </p>

<pre>
/etc/postfix/main.cf:
    postscreen_access_list = permit_mynetworks,
        cidr:/etc/postfix/postscreen_access.cidr

/etc/postfix/postscreen_access.cidr:
   # Rules are evaluated in the order as specified.
   # Denylist 192.168.* except 192.168.0.1.
   192.168.0.1          permit
   192.168.0.0/16       reject
</pre>

<p> See the postscreen_access_list manpage documentation for more
details.  </p>

<p> When the SMTP client address matches a "permit" action,
postscreen(8) logs this with the client address and port number as:
</p>

<blockquote>
<pre>
<b>ALLOWLISTED</b> <i>[address]:port</i>
</pre>
</blockquote>

<blockquote> <p> Use the respectful_logging configuration parameter to
select a deprecated form of this logging. </p> </blockquote>

<p> The allowlist action is not configurable: immediately hand off the
connection to a Postfix SMTP server process. </p>

<p> When the SMTP client address matches a "reject" action,
postscreen(8) logs this with the client address and port number as:
</p>

<blockquote>
<pre>
<b>DENYLISTED</b> <i>[address]:port</i>
</pre>
</blockquote>

<blockquote> <p> Use the respectful_logging configuration parameter to
select a deprecated form of this logging. </p> </blockquote>

<p> The postscreen_denylist_action parameter specifies the action
that is taken next.  See "<a href="#fail_before_220">When tests
fail before the 220 SMTP server greeting</a>" below. </p>

<h3> <a name="temp_allow"> Temporary allowlist test </a> </h3>

<p> The postscreen(8) daemon maintains a <i>temporary</i>
allowlist for SMTP client IP addresses that have passed all
the tests described below. The postscreen_cache_map parameter
specifies the location of the temporary allowlist.  The
temporary allowlist is not used for SMTP client addresses
that appear on the <i>permanent</i> access list. </p>

<p> By default the temporary allowlist is not shared with other
postscreen(8) daemons. See
<a href="#temp_allow_sharing"> Sharing
the temporary allowlist </a> below for alternatives. </p>

<p> When the SMTP client address appears on the temporary
allowlist, postscreen(8) logs this with the client address and port
number as: </p>

<pre>
    <b>PASS OLD</b> <i>[address]:port</i>
</pre>

<p> The action is not configurable: immediately hand off the
connection to a Postfix SMTP server process.  The client is
excluded from further tests until its temporary allowlist
entry expires, as controlled with the postscreen_*_ttl
parameters.  Expired entries are silently renewed if possible. </p>

<h3> <a name="allow_veto"> MX Policy test </a> </h3>

<p> When the remote SMTP client is not on the static access list
or temporary allowlist, postscreen(8) can implement a number of
allowlist tests, before it grants the client a temporary allowlist
status that allows it to talk to a Postfix SMTP server process. </p>

<p> When postscreen(8) is configured to monitor all primary and
backup MX addresses, it can refuse to allowlist clients that connect
to a backup MX address only (an old spammer trick to take advantage
of backup MX hosts with weaker anti-spam policies than primary MX
hosts). </p>

<blockquote> <p> NOTE: The following solution is for small sites.
Larger sites would have to share the postscreen(8) cache between
primary and backup MTAs, which would introduce a common point of
failure.  </p> </blockquote>

<ul>

<li> <p> First, configure the host to listen on both primary and
backup MX addresses. Use the appropriate <tt>ifconfig</tt> or <tt>ip</tt>
command for the local operating system, or update the appropriate
configuration files and "refresh" the network protocol stack. </p>

<p> <p> Second, configure Postfix to listen on the new IP address
(this step is needed when you have specified inet_interfaces in
main.cf). </p>

<li> <p> Then, configure postscreen(8) to deny the temporary allowlist
status on the backup MX address(es).  An example for Wietse's
server is: </p>

<pre>
/etc/postfix/main.cf:
    postscreen_allowlist_interfaces = !168.100.189.8 static:all
</pre>

<p> Translation: allow clients to obtain the temporary allowlist
status on all server IP addresses except 168.100.189.8, which is a
backup MX address.  </p>

</ul>

<p> When a non-allowlisted client connects the backup MX address,
postscreen(8) logs this with the client address and port number as:
</p>

<blockquote> <pre>
<b>CONNECT from</b> <i>[address]:port</i> <b>to [168.100.189.8]:25</b>
<b>ALLOWLIST VETO</b> <i>[address]:port</i>
</pre> </blockquote>

<blockquote> <p> Use the respectful_logging configuration parameter to
select a deprecated form of this logging. </p> </blockquote>

<p> Translation: the client at <i>[address]:port</i> connected to
the backup MX address 168.100.189.8 while it was not allowlisted.
The client will not be granted the temporary allowlist status, even
if passes all the allowlist tests described below. </p>

<h2> <a name="before_220"> Tests before the 220 SMTP server greeting </a> </h2>

<p> The postscreen_greet_wait parameter specifies a short time
interval before the "220 <i>text</i>..." server greeting, where
postscreen(8) can run a number of tests in parallel. </p>

<p> When a good client passes these tests, and no "<a
href="#after_220">deep protocol tests</a>"
are configured, postscreen(8)
adds the client to the temporary allowlist and hands off the "live"
connection to a Postfix SMTP server process.  The client can then
continue as if postscreen(8) never even existed (except of course
for the short postscreen_greet_wait delay).  </p>

<ul>

<li> <a href="#pregreet"> Pregreet test </a>

<li> <a href="#dnsbl"> DNS Allow/denylist test </a>

<li> <a href="#fail_before_220">When tests fail before the 220 SMTP server greeting</a>

</ul>

<h3> <a name="pregreet"> Pregreet test </a> </h3>

<p> The SMTP protocol is a classic example of a protocol where the
server speaks before the client. postscreen(8) detects zombies
that are in a hurry and that speak before their turn. This test is
enabled by default. </p>

<p> The postscreen_greet_banner parameter specifies the <i>text</i>
portion of a "220-<i>text</i>..." teaser banner (default: $smtpd_banner).
Note that this becomes the first part of a multi-line server greeting.
The postscreen(8) daemon sends this before the postscreen_greet_wait
timer is started.  The purpose of the teaser banner is to confuse
zombies so that they speak before their turn. It has no effect on
SMTP clients that correctly implement the protocol.  </p>

<p> To avoid problems with poorly-implemented SMTP engines in network
appliances or network testing tools, either exclude them from all
tests with the postscreen_access_list feature or else specify
an empty teaser banner: </p>

<pre>
/etc/postfix/main.cf:
    # Exclude broken clients by allowlisting. Clients in mynetworks
    # should always be allowlisted.
    postscreen_access_list = permit_mynetworks,
        cidr:/etc/postfix/postscreen_access.cidr

/etc/postfix/postscreen_access.cidr:
    192.168.254.0/24 permit
</pre>

<pre>
/etc/postfix/main.cf:
    # Disable the teaser banner (try allowlisting first if you can).
    postscreen_greet_banner =
</pre>

<p> When an SMTP client sends a command before the
postscreen_greet_wait time has elapsed, postscreen(8) logs this as:
</p>

<pre>
    <b>PREGREET</b> <i>count</i> <b>after</b> <i>time</i> <b>from</b> <i>[address]:port text...</i>
</pre>

<p> Translation: the client at <i>[address]:port</i> sent <i>count</i>
bytes before its turn to speak. This happened <i>time</i> seconds
after the postscreen_greet_wait timer was started.  The <i>text</i>
is what the client sent (truncated to 100 bytes, and with non-printable
characters replaced with C-style escapes such as \r for carriage-return
and \n for newline). </p>

<p> The postscreen_greet_action parameter specifies the action that
is taken next.  See "<a href="#fail_before_220">When tests fail
before the 220 SMTP server greeting</a>" below. </p>

<h3> <a name="dnsbl"> DNS Allow/denylist test </a> </h3>

<p> The postscreen_dnsbl_sites parameter (default: empty) specifies
a list of DNS blocklist servers with optional filters and weight
factors (positive weights for denylisting, negative for allowlisting).
These servers will be queried in parallel with the reverse client
IP address.  This test is disabled by default. </p>

<blockquote>
<p>
CAUTION: when postscreen rejects mail, its SMTP reply contains the
DNSBL domain name. Use the postscreen_dnsbl_reply_map feature to
hide "password" information in DNSBL domain names.
</p>
</blockquote>

<p> When the postscreen_greet_wait time has elapsed, and the combined
DNSBL score is equal to or greater than the postscreen_dnsbl_threshold
parameter value, postscreen(8) logs this as: </p>

<pre>
    <b>DNSBL rank</b> <i>count</i> <b>for</b> <i>[address]:port</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> has a combined
DNSBL score of <i>count</i>. </p>

<p> The postscreen_dnsbl_action parameter specifies the action that
is taken when the combined DNSBL score is equal to or greater than
the threshold.  See "<a href="#fail_before_220">When tests fail
before the 220 SMTP server greeting</a>" below. </p>

<h3> <a name="fail_before_220">When tests fail before the 220 SMTP server greeting</a> </h3>

<p> When the client address matches the permanent denylist, or
when the client fails the pregreet or DNSBL tests, the action is
specified with postscreen_denylist_action, postscreen_greet_action,
or postscreen_dnsbl_action, respectively. </p>

<dl>

<dt> <b>ignore</b> (default) </dt>

<dd> Ignore the failure of this test. Allow other tests to complete.
Repeat this test the next time the client connects.  This option
is useful for testing and collecting statistics without blocking
mail. </dd>

<dt> <b>enforce</b> </dt>

<dd> Allow other tests to complete.  Reject attempts to deliver mail
with a 550 SMTP reply, and log the helo/sender/recipient information.
Repeat this test the next time the client connects. </dd>

<dt> <b>drop</b> </dt>

<dd> Drop the connection immediately with a 521 SMTP reply.  Repeat
this test the next time the client connects. </dd>

</dl>

<h2> <a name="after_220">Tests after the 220 SMTP server greeting</a> </h2>

<p> In this phase of the protocol, postscreen(8) implements a
number of "deep protocol" tests. These tests use an SMTP protocol
engine that is built into the postscreen(8) server. </p>

<p> Important note: these protocol tests are disabled by default.
They are more intrusive than the pregreet and DNSBL tests, and they
have limitations as discussed next. </p>

<ul>

<li> <p> The main limitation of "after 220 greeting" tests is that
a new client must disconnect after passing these tests (reason:
postscreen is not a proxy).  Then the client must reconnect from
the same IP address before it can deliver mail.  The following
measures may help to avoid email delays: </p>

<ul>

<li> <p> Allow "good" clients to skip tests with the
postscreen_dnsbl_allowlist_threshold feature. This is especially effective
for large providers that usually don't retry from the same IP
address. </p>

<li> <p> Small sites: Configure postscreen(8) to listen on multiple
IP addresses, published in DNS as different IP addresses for the
same MX hostname or for different MX hostnames. This avoids mail
delivery delays with clients that reconnect immediately from the
same IP address.  </p>

<li> <p> Large sites: Share the postscreen(8) cache between different
Postfix MTAs with a large-enough memcache_table(5). Again, this
avoids mail delivery delays with clients that reconnect immediately
from the same IP address. </p>

</ul>

<li> <p> postscreen(8)'s built-in SMTP engine does not implement the
AUTH, XCLIENT, and XFORWARD features. If you need to make these
services available on port 25, then do not enable the tests after
the 220 server greeting. </p>

<li> <p> End-user clients should connect directly to the submission
service, so that they never have to deal with postscreen(8)'s tests.
</p>

</ul>

<p> The following "after 220 greeting" tests are available: </p>

<ul>

<li> <a href="#pipelining">Command pipelining test</a>

<li> <a href="#non_smtp">Non-SMTP command test</a>

<li> <a href="#barelf">Bare newline test</a>

<li> <a href="#fail_after_220">When tests fail after the 220 SMTP server greeting</a>

</ul>

<h3> <a name="pipelining">Command pipelining test</a> </h3>

<p> By default, SMTP is a half-duplex protocol: the sender and
receiver send one command and one response at a time.  Unlike the
Postfix SMTP server, postscreen(8) does not announce support
for ESMTP command pipelining.  Therefore, clients are not allowed
to send multiple commands. postscreen(8)'s
<a href="#after_220">deep
protocol test</a> for this is disabled by default. </p>

<p> With "postscreen_pipelining_enable = yes", postscreen(8) detects
zombies that send multiple commands, instead of sending one command
and waiting for the server to reply.  </p>

<p> This test is opportunistically enabled when postscreen(8) has
to use the built-in SMTP engine anyway. This is to make postscreen(8)
logging more informative. </p>

<p> When a client sends multiple commands, postscreen(8) logs this
as: </p>

<pre>
    <b>COMMAND PIPELINING from</b> <i>[address]:port</i> <b>after</b> <i>command</i>: <i>text</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> sent
multiple SMTP commands, instead of sending one command and then
waiting for the server to reply. This happened after the client
sent <i>command</i>. The <i>text</i> shows part of the input that
was sent too early; it is not logged with Postfix 2.8. </p>

<p> The postscreen_pipelining_action parameter specifies the action
that is taken next.  See "<a href="#fail_after_220">When tests fail
after the 220 SMTP server greeting</a>" below. </p>

<h3> <a name="non_smtp">Non-SMTP command test</a> </h3>

<p> Some spambots send their mail through open proxies. A symptom
of this is the usage of commands such as CONNECT and other non-SMTP
commands. Just like the Postfix SMTP server's smtpd_forbidden_commands
feature, postscreen(8) has an equivalent postscreen_forbidden_commands
feature to block these clients. postscreen(8)'s
<a href="#after_220">deep
protocol test</a> for this is disabled by default.  </p>

<p> With "postscreen_non_smtp_command_enable = yes", postscreen(8)
detects zombies that send commands specified with the
postscreen_forbidden_commands parameter. This also detects commands
with the syntax of a message header label. The latter is a symptom
that the client is sending message content after ignoring all the
responses from postscreen(8) that reject mail. </p>

<p> This test is opportunistically enabled when postscreen(8) has
to use the built-in SMTP engine anyway. This is to make postscreen(8)
logging more informative.  </p>

<p> When a client sends non-SMTP commands, postscreen(8) logs this
as: </p>

<pre>
    <b>NON-SMTP COMMAND from</b> <i>[address]:port</i> <b>after</b> <i>command: text</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> sent a
command that matches the postscreen_forbidden_commands
parameter, or that has the syntax of a message header label (text
followed by optional space and ":").
The "<tt><b>after</b> <i>command</i></tt>" portion is logged with
Postfix 2.10 and later. </p>

<p> The postscreen_non_smtp_command_action parameter specifies
the action that is taken next.  See "<a href="#fail_after_220">When
tests fail after the 220 SMTP server greeting</a>" below. </p>

<h3> <a name="barelf">Bare newline test</a> </h3>

<p> SMTP is a line-oriented protocol: lines have a limited length,
and are terminated with &lt;CR&gt;&lt;LF&gt;. Lines ending in a
"bare" &lt;LF&gt;, that is newline not preceded by carriage return,
are not allowed in SMTP.  postscreen(8)'s
<a href="#after_220">deep
protocol test</a> for this is disabled by default.  </p>

<p> With "postscreen_bare_newline_enable = yes", postscreen(8)
detects clients that send lines ending in bare newline characters.
</p>

<p> This test is opportunistically enabled when postscreen(8) has
to use the built-in SMTP engine anyway. This is to make postscreen(8)
logging more informative.  </p>

<p> When a client sends bare newline characters, postscreen(8) logs
this as:
</p>

<pre>
    <b>BARE NEWLINE from</b> <i>[address]:port</i> <b>after</b> <i>command</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> sent a bare
newline character, that is newline not preceded by carriage
return.
The "<tt><b>after</b> <i>command</i></tt>" portion is logged with
Postfix 2.10 and later. </p>

<p> The postscreen_bare_newline_action parameter specifies the
action that is taken next.  See "<a href="#fail_after_220">When
tests fail after the 220 SMTP server greeting</a>" below. </p>

<h3> <a name="fail_after_220">When tests fail after the 220 SMTP server greeting</a> </h3>

<p> When the client fails the pipelining, non-SMTP command or bare
newline tests, the action is specified with postscreen_pipelining_action,
postscreen_non_smtp_command_action or postscreen_bare_newline_action,
respectively. </p>

<dl>

<dt> <b>ignore</b> (default for bare newline) </dt>

<dd> Ignore the failure of this test. Allow other tests to complete.
Do NOT repeat this test before the result from some other test
expires.

This option is useful for testing and collecting statistics without
blocking mail permanently. </dd>

<dt> <b>enforce</b> (default for pipelining) </dt>

<dd> Allow other tests to complete.  Reject attempts to deliver
mail with a 550 SMTP reply, and log the helo/sender/recipient
information.  Repeat this test the next time the client connects.
</dd>

<dt> <b>drop</b> (default for non-SMTP commands) </dt>

<dd> Drop the connection immediately with a 521 SMTP reply.  Repeat
this test the next time the client connects.  This action is
compatible with the Postfix SMTP server's smtpd_forbidden_commands
feature. </dd>

</dl>

<h2> <a name="other_error">Other errors</a> </h2>

<p> When an SMTP client hangs up unexpectedly, postscreen(8) logs
this as: </p>

<pre>
    <b>HANGUP after</b> <i>time</i> <b>from</b> <i>[address]:port</i> <b>in</b> <i>test name</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> disconnected
unexpectedly, <i>time</i> seconds after the start of the
test named <i>test name</i>. </p>

<p> There is no punishment for hanging up. A client that hangs up
without sending the QUIT command can still pass all postscreen(8)
tests. </p>

<!--

<p> While an unexpired penalty is in effect, an SMTP client is not
allowed to pass any tests, and  postscreen(8) logs each connection
with the remaining amount of penalty time as: </p>

<pre>
    <b>PENALTY</b> <i>time</i> <b>for</b> <i>[address]:port</i>
</pre>

<p> During this time, all attempts by the client to deliver mail
will be deferred with a 450 SMTP status.  </p>

-->

<p> The following errors are reported by the built-in SMTP engine.
This engine never accepts mail, therefore it has per-session limits
on the number of commands and on the session length. </p>

<pre>
    <b>COMMAND TIME LIMIT</b> <b>from</b> <i>[address]:port</i> <b>after</b> <i>command</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> reached the
per-command time limit as specified with the postscreen_command_time_limit
parameter.  The session is terminated immediately.
The "<tt><b>after</b> <i>command</i></tt>" portion is logged with
Postfix 2.10 and later. </p>

<pre>
    <b>COMMAND COUNT LIMIT from</b> <i>[address]:port</i> <b>after</b> <i>command</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> reached the
per-session command count limit as specified with the
postscreen_command_count_limit parameter.  The session is terminated
immediately.
The "<tt><b>after</b> <i>command</i></tt>" portion is logged with
Postfix 2.10 and later. </p>

<pre>
    <b>COMMAND LENGTH LIMIT from</b> <i>[address]:port</i> <b>after</b> <i>command</i>
</pre>

<p> Translation: the SMTP client at <i>[address]:port</i> reached the
per-command length limit, as specified with the line_length_limit
parameter.  The session is terminated immediately.
The "<tt><b>after</b> <i>command</i></tt>" portion is logged with
Postfix 2.10 and later. </p>

<p> When an SMTP client makes too many connections at the same time,
postscreen(8) rejects the connection with a 421 status code and logs: </p>

<pre>
    <b>NOQUEUE: reject: CONNECT from</b> <i>[address]:port</i><b>: too many connections</b>
</pre>

<p> The postscreen_client_connection_count_limit parameter controls this limit. </p>

<p> When an SMTP client connects after postscreen(8) has reached a
connection count limit, postscreen(8) rejects the connection with
a 421 status code and logs: </p>

<pre>
    <b>NOQUEUE: reject: CONNECT from</b> <i>[address]:port</i><b>: all screening ports busy</b>
    <b>NOQUEUE: reject: CONNECT from</b> <i>[address]:port</i><b>: all server ports busy</b>
</pre>

<p> The postscreen_pre_queue_limit and postscreen_post_queue_limit
parameters control these limits.  </p>

<h2> <a name="victory">When all tests succeed</a> </h2>

<p> When a new SMTP client passes all tests (i.e. it is not allowlisted
via some mechanism), postscreen(8) logs this as: </p>

<pre>
    <b>PASS NEW</b> <i>[address]:port</i>
</pre>

<p> Where <i>[address]:port</i> are the client IP address and port.
Then, postscreen(8)
creates a temporary allowlist entry that excludes the client IP
address from further tests until the temporary allowlist entry
expires, as controlled with the postscreen_*_ttl parameters. </p>

<p> When no "<a href="#after_220">deep protocol tests</a>" are
configured, postscreen(8) hands off the "live" connection to a Postfix
SMTP server process.  The client can then continue as if postscreen(8)
never even existed (except for the short postscreen_greet_wait delay).
</p>

<p> When any "<a href="#after_220">deep protocol tests</a>" are
configured, postscreen(8) cannot hand off the "live" connection to
a Postfix SMTP server process in the middle of the session.  Instead,
postscreen(8) defers mail delivery attempts with a 4XX status, logs
the helo/sender/recipient information, and waits for the client to
disconnect.  The next time the client connects it will be allowed
to talk to a Postfix SMTP server process to deliver its mail.
postscreen(8) mitigates the impact of this limitation by giving
<a href="#after_220">deep protocol tests</a> a long expiration
time. </p>

<h2> <a name="config"> Configuring the postscreen(8) service</a>
</h2>

<p> postscreen(8) has been tested on FreeBSD [4-8], Linux 2.[4-6]
and Solaris 9 systems. </p>

<ul>

<li> <a href="#enable"> Turning on postscreen(8) without blocking
mail</a>

<li> <a href="#starttls"> postscreen(8) TLS configuration </a>

<li> <a href="#blocking"> Blocking mail with postscreen(8) </a>

<li> <a href="#turnoff"> Turning off postscreen(8) </a>

<li> <a href="#temp_allow_sharing"> Sharing the temporary allowlist
</a>

</ul>

<h3> <a name="enable"> Turning on postscreen(8) without blocking mail</a> </h3>

<p> To enable the postscreen(8) service and log client information
without blocking mail: </p>

<ol>

<li> <p> Make sure that local clients and systems with non-standard
SMTP implementations are excluded from any postscreen(8) tests. The
default is to exclude all clients in mynetworks. To exclude additional
clients, for example, third-party performance monitoring tools (these
tend to have broken SMTP implementations): </p>

<pre>
/etc/postfix/main.cf:
    # Exclude broken clients by allowlisting. Clients in mynetworks
    # should always be allowlisted.
    postscreen_access_list = permit_mynetworks,
        cidr:/etc/postfix/postscreen_access.cidr

/etc/postfix/postscreen_access.cidr:
    192.168.254.0/24 permit
</pre>

<li> <p> Comment out the "<tt>smtp  inet ... smtpd</tt>" service
in master.cf, including any "<tt>-o parameter=value</tt>" entries
that follow.  </p>

<pre>
/etc/postfix/master.cf:
    #smtp      inet  n       -       n       -       -       smtpd
    #    -o parameter=value ...
</pre>

<li> <p> Uncomment the new "<tt>smtpd pass ... smtpd</tt>" service
in master.cf, and duplicate any "<tt>-o parameter=value</tt>" entries
from the smtpd service that was commented out in the previous step.
</p>

<pre>
/etc/postfix/master.cf:
    smtpd     pass  -       -       n       -       -       smtpd
        -o parameter=value ...
</pre>

<li> <p> Uncomment the new "<tt>smtp inet ... postscreen</tt>"
service in master.cf. </p>

<pre>
/etc/postfix/master.cf:
    smtp      inet  n       -       n       -       1       postscreen
</pre>

<li> <p> Uncomment the new "<tt>tlsproxy unix ... tlsproxy</tt>"
service in master.cf.  This service implements STARTTLS support for
postscreen(8). </p>

<pre>
/etc/postfix/master.cf:
    tlsproxy  unix  -       -       n       -       0       tlsproxy
</pre>

<li> <p> Uncomment the new "<tt>dnsblog  unix ... dnsblog</tt>"
service in master.cf.  This service does DNSBL lookups for postscreen(8)
and logs results. </p>

<pre>
/etc/postfix/master.cf:
    dnsblog   unix  -       -       n       -       0       dnsblog
</pre>

<li> <p> To enable DNSBL lookups, list some DNS blocklist sites in
main.cf, separated by whitespace. Different sites can have different
weights. For example:

<pre>
/etc/postfix/main.cf:
    postscreen_dnsbl_threshold = 2
    postscreen_dnsbl_sites = zen.spamhaus.org*2
        bl.spamcop.net*1 b.barracudacentral.org*1
</pre>

<p> Note: if your DNSBL queries have a "secret" in the domain name,
you must censor this information from the postscreen(8) SMTP replies.
For example: </p>

<pre>
/etc/postfix/main.cf:
    postscreen_dnsbl_reply_map = texthash:/etc/postfix/dnsbl_reply
</pre>

<pre>
/etc/postfix/dnsbl_reply:
    # Secret DNSBL name           Name in postscreen(8) replies
    secret.zen.dq.spamhaus.net    zen.spamhaus.org
</pre>

<p> The texthash: format is similar to hash: except that there is
no need to run postmap(1) before the file can be used, and that it
does not detect changes after the file is read. It is new with
Postfix version 2.8. </p>

<li> <p> Read the new configuration with "<tt>postfix reload</tt>".
</p>

</ol>

<p> Notes: </p>

<ul>

<li> <p> Some postscreen(8) configuration parameters implement
stress-dependent behavior. This is supported only when the default
value is stress-dependent (that is, "postconf -d <i>parametername</i>"
output shows
"<i>parametername</i>&nbsp;=&nbsp;${stress?<i>something</i>}${stress:<i>something</i>}" or
"<i>parametername</i>&nbsp;=&nbsp;${stress?{<i>something</i>}:{<i>something</i>}}").
Other parameters always evaluate as if the stress value is the empty
string. </p>

<li> <p> See "<a href="#before_220">Tests before the 220 SMTP server
greeting</a>" for details about the logging from these
postscreen(8) tests. </p>

<li> <p> If you run Postfix 2.6 or earlier you must stop and start
the master daemon ("<tt>postfix stop; postfix start</tt>").  This
is needed because the Postfix "pass" master service type did not
work reliably on all systems. </p>

</ul>

<h3> <a name="starttls"> postscreen(8) TLS configuration </a> </h3>

<p> postscreen(8) TLS support is available for remote SMTP clients
that aren't allowlisted, including clients that need to renew their
temporary allowlist status.  When a remote SMTP client requests TLS
service, postscreen(8) invisibly hands off the connection to a
tlsproxy(8) process. Then, tlsproxy(8) encrypts and decrypts the
traffic between postscreen(8) and the remote SMTP client. One
tlsproxy(8) process can handle multiple SMTP sessions. The number
of tlsproxy(8) processes slowly increases with server load, but it
should always be much smaller than the number of postscreen(8) TLS
sessions.  </p>

<p> TLS support for postscreen(8) and tlsproxy(8) uses the same
parameters as with smtpd(8). We recommend that you keep the relevant
configuration parameters in main.cf.  If you must specify "-o
smtpd_mumble=value" parameter overrides in master.cf for a
postscreen-protected smtpd(8) service, then you should specify those
same parameter overrides for the postscreen(8) and tlsproxy(8)
services. </p>

<h3> <a name="blocking"> Blocking mail with postscreen(8) </a> </h3>

<p> For compatibility with smtpd(8), postscreen(8) implements the
soft_bounce safety feature. This causes Postfix to reject mail with
a "try again" reply code. </p>

<ul>

<li> <p> To turn this on for all of Postfix, specify "<tt>soft_bounce
= yes</tt>" in main.cf. </p>

<li> <p> To turn this on for postscreen(8) only, append "<tt>-o
soft_bounce=yes</tt>" (note: NO SPACES around '=') to the postscreen
entry in master.cf. <p>

</ul>

<p> Execute "<tt>postfix reload</tt>" to make the change effective. </p>

<p> After testing, do not forget to remove the soft_bounce feature,
otherwise senders won't receive their non-delivery notification
until many days later.  </p>

<p> To use the postscreen(8) service to block mail, edit main.cf and
specify one or more of: </p>

<ul>

<li> <p> "<tt>postscreen_dnsbl_action = enforce</tt>", to reject
clients that are on DNS blocklists, and to log the helo/sender/recipient
information. With good DNSBLs this reduces the amount of load on
Postfix SMTP servers dramatically.  </p>

<li> <p> "<tt>postscreen_greet_action = enforce</tt>", to reject
clients that talk before their turn, and to log the helo/sender/recipient
information. This stops over half of all known-to-be illegitimate
connections to Wietse's mail server. It is backup protection for
zombies that haven't yet been denylisted. </p>

<li> <p> You can also enable "<a href="#after_220">deep protocol
tests</a>", but these are more intrusive than the pregreet or DNSBL
tests. </p>

<p> When a good client passes the "<a href="#after_220">deep
protocol tests</a>",
postscreen(8) adds the client to the temporary
allowlist but it cannot hand off the "live" connection to a Postfix
SMTP server process in the middle of the session. Instead, postscreen(8)
defers mail delivery attempts with a 4XX status, logs the
helo/sender/recipient information, and waits for the client to
disconnect. </p>

<p> When the good client comes back in a later session, it is allowed
to talk directly to a Postfix SMTP server.  See "<a href="#after_220">Tests
after the 220 SMTP server greeting</a>" above for limitations with
AUTH and other features that clients may need.  </p>

<p> An unexpected benefit from "<a href="#after_220">deep protocol
tests</a>" is that some "good" clients don't return after the 4XX
reply; these clients were not so good after all. </p>

<p> Unfortunately, some senders will retry requests from different
IP addresses, and may never get allowlisted.  For this reason,
Wietse stopped using "<a href="#after_220">deep protocol tests</a>"
on his own internet-facing mail server.  </p>

<li> <p> There is also support for permanent denylisting and
allowlisting; see the description of the postscreen_access_list
parameter for details. </p>

</ul>

<h3> <a name="turnoff"> Turning off postscreen(8) </a> </h3>

<p> To turn off postscreen(8) and handle mail directly with Postfix
SMTP server processes: </p>

<ol>

<li> <p> Comment out the "<tt>smtp inet ... postscreen</tt>" service
in master.cf, including any "<tt>-o parameter=value</tt>" entries
that follow. </p>

<pre>
/etc/postfix/master.cf:
    #smtp      inet  n       -       n       -       1       postscreen
    #    -o parameter=value ...
</pre>

<li> <p> Comment out the "<tt>dnsblog  unix ... dnsblog</tt>" service
in master.cf.  </p>

<pre>
/etc/postfix/master.cf:
    #dnsblog   unix  -       -       n       -       0       dnsblog
</pre>

<li> <p> Comment out the "<tt>smtpd pass ... smtpd</tt>" service
in master.cf, including any "<tt>-o parameter=value</tt>" entries
that follow. </p>

<pre>
/etc/postfix/master.cf:
    #smtpd     pass  -       -       n       -       -       smtpd
    #    -o parameter=value ...
</pre>

<li> <p> Comment out the "<tt>tlsproxy unix ... tlsproxy</tt>"
service in master.cf, including any "<tt>-o parameter=value</tt>"
entries that follow. </p>

<pre>
/etc/postfix/master.cf:
    #tlsproxy  unix  -       -       n       -       0       tlsproxy
    #    -o parameter=value ...
</pre>

<li> <p> Uncomment the "<tt>smtp  inet ... smtpd</tt>" service in
master.cf, including any "<tt>-o parameter=value</tt>" entries that
may follow.  </p>

<pre>
/etc/postfix/master.cf:
    smtp       inet  n       -       n       -       -       smtpd
        -o parameter=value ...
</pre>

<li> <p> Read the new configuration with "<tt>postfix reload</tt>".
</p>

</ol>

<h3> <a name="temp_allow_sharing"> Sharing the temporary allowlist </a> </h3>

<p> By default, the temporary allowlist is not shared between
multiple postscreen(8) daemons.  To enable sharing, choose one
of the following options: </p>

<ul>

<li> <p> A non-persistent memcache: temporary allowlist can be shared
    between postscreen(8) daemons on the same host or different
    hosts.  Disable cache cleanup (postscreen_cache_cleanup_interval
    = 0) in all postscreen(8) daemons because memcache: has no
    first-next API (but see example 4 below for memcache: with
    persistent backup). This requires Postfix 2.9 or later. </p>

    <pre>
    # Example 1: non-persistent memcache: allowlist.
    /etc/postfix/main.cf:
	postscreen_cache_map = memcache:/etc/postfix/postscreen_cache
	postscreen_cache_cleanup_interval = 0

    /etc/postfix/postscreen_cache:
	memcache = inet:127.0.0.1:11211
	key_format = postscreen:%s
    </pre>

<li> <p>
    A persistent lmdb: temporary allowlist can be shared between
    postscreen(8) daemons that run under the same master(8) daemon,
    or under different master(8) daemons on the same host.  Disable
    cache cleanup (postscreen_cache_cleanup_interval = 0) in all
    postscreen(8) daemons except one that is responsible for cache
    cleanup. This requires Postfix 2.11 or later. </p>

    <pre>
    # Example 2: persistent lmdb: allowlist.
    /etc/postfix/main.cf:
	postscreen_cache_map = lmdb:$data_directory/postscreen_cache
	# See note 1 below.
	# postscreen_cache_cleanup_interval = 0
    </pre>

<li> <p> Other kinds of persistent temporary allowlist can be shared
    only between postscreen(8) daemons that run under the same
    master(8) daemon. In this case, temporary allowlist access must
    be shared through the proxymap(8) daemon. This requires Postfix
    2.9 or later. </p>

    <pre>
    # Example 3: proxied btree: allowlist.
    /etc/postfix/main.cf:
	postscreen_cache_map =
	    proxy:btree:/var/lib/postfix/postscreen_cache
	# See note 1 below.
	# postscreen_cache_cleanup_interval = 0

    # Example 4: proxied btree: allowlist with memcache: accelerator.
    /etc/postfix/main.cf:
	postscreen_cache_map = memcache:/etc/postfix/postscreen_cache
	proxy_write_maps =
	    proxy:btree:/var/lib/postfix/postscreen_cache
	    ... other proxied tables ...
	# See note 1 below.
	# postscreen_cache_cleanup_interval = 0

    /etc/postfix/postscreen_cache:
	# Note: the $data_directory macro is not defined in this context.
	memcache = inet:127.0.0.1:11211
	backup = proxy:btree:/var/lib/postfix/postscreen_cache
	key_format = postscreen:%s
    </pre>

    <p> Note 1: disable cache cleanup (postscreen_cache_cleanup_interval
    = 0) in all postscreen(8) daemons except one that is responsible
    for cache cleanup. </p>

    <p> Note 2: postscreen(8) cache sharing via proxymap(8) requires Postfix
    2.9 or later; earlier proxymap(8) implementations don't support
    cache cleanup.  </p>

</ul>

<h2> <a name="historical"> Historical notes and credits </a> </h2>

<p> Many ideas in postscreen(8) were explored in earlier work by
Michael Tokarev, in OpenBSD spamd, and in MailChannels Traffic
Control. </p>

<p> Wietse threw together a crude prototype with pregreet and dnsbl
support in June 2009, because he needed something new for a Mailserver
conference presentation in July. Ralf Hildebrandt ran this code on
several servers to collect real-world statistics. This version used
the dnsblog(8) ad-hoc DNS client program. </p>

<p> Wietse needed new material for a LISA conference presentation
in November 2010, so he added support for DNSBL weights and filters
in August, followed by a major code rewrite, deep protocol tests,
helo/sender/recipient logging, and stress-adaptive behavior in
September. Ralf Hildebrandt ran this code on several servers to
collect real-world statistics. This version still used the embarrassing
dnsblog(8) ad-hoc DNS client program.  </p>

<p> Wietse added STARTTLS support in December 2010. This makes
postscreen(8) usable for sites that require TLS support.  The
implementation introduces the tlsproxy(8) event-driven TLS proxy
that decrypts/encrypts the sessions for multiple SMTP clients. </p>

<p> The tlsproxy(8) implementation led to the discovery of a "new"
class of vulnerability (<a
href="http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2011-0411"
>CVE-2011-0411</a>) that affected multiple implementations of SMTP,
POP, IMAP, NNTP, and FTP over TLS. </p>

<p> postscreen(8) was officially released as part of the Postfix
2.8 stable release in January 2011.</p>

<p> Noel Jones helped with the Postfix 3.6 transition towards respectful
documentation. </p>

</body>
</html>