Citing Lars:

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">

<html><head>

<title>Apache non-IP Virtual Hosts</title>

</head>

<!-- Background white, links blue (unvisited), navy (visited), red (active) -->

<BODY

 BGCOLOR="#FFFFFF"

 TEXT="#000000"

 LINK="#0000FF"

 VLINK="#000080"

 ALINK="#FF0000"

>

<!--#include virtual="header.html" -->

<h1 ALIGN="CENTER">Apache non-IP Virtual Hosts</h1>

<strong>See Also:</strong>

<a href="virtual-host.html">Virtual Host Support</a>

<hr>

<h2>What is a Virtual Host</h2>

<p>The "Virtual Host" refers to the practice of maintaining more than

one server on one machine, as differentiated by their apparent

hostname. For example, it is often desirable for companies sharing a

web server to have their own domains, with web servers accessible as

<code>www.company1.com</code> and <code>www.company2.com</code>,

without requiring the user to know any extra path information.</p>

<p>Apache was one of the first servers to support virtual hosts right

out of the box, but since the base <code>HTTP</code> (HyperText

Transport Protocol) standard does not allow any method for the server

to determine the hostname it is being addressed as, Apache's virtual

host support has required a separate IP address for each

server. Documentation on using this approach (which still works very

well) <a href="virtual-host.html">is available</a>.

<p>While the approach described above works, with the available IP

address space growing smaller, and the number of domains increasing,

it is not the most elegant solution, and is hard to implement on some

machines. The <code>HTTP/1.1</code> protocol contains a method for the

server to identify what name it is being addressed as. Apache 1.1 and

later support this approach as well as the traditional

IP-address-per-hostname method.</p>

<p>The benefits of using the new virtual host support is a practically

unlimited number of servers, ease of configuration and use, and

requires no additional hardware or software. The main disadvantage is

that the user's browser must support this part of the protocol. The

latest versions of many browsers (including Netscape Navigator 2.0 and

later) do, but many browsers, especially older ones, do not. This can

cause problems, although a possible solution is addressed below.</p>

<h2>Using non-IP Virtual Hosts</h2>

<p>Using the new virtual hosts is quite easy, and superficially looks

like the old method. You simply add to one of the Apache configuration

files (most likely <code>httpd.conf</code> or <code>srm.conf</code>)

code similar to the following:</p>

<pre>

    <VirtualHost www.apache.org>

    ServerName www.apache.org

    DocumentRoot /usr/web/apache

    </VirtualHost>

</pre>

<p>Of course, any additional directives can (and should) be placed

into the <code>&lt;VirtualHost&gt;</code> section. To make this work,

all that is needed is to make sure that the <code>www.apache.org</code>

DNS entry points to the same IP address as the main

server. Optionally, you could simply use that IP address in the

&lt;VirtualHost&gt; entry.</p>

<p>Additionally, many servers may wish to be accessible by more than

one name. For example, the Apache server might want to be accessible

as <code>apache.org</code>, or <code>ftp.apache.org</code>, assuming

the IP addresses pointed to the same server. In fact, one might want it

so that all addresses at <code>apache.org</code> were picked up by the

server. This is possible with the <code>ServerAlias</code>

directive, placed inside the <VirtualHost> section. For

example:</p>

<pre>

    ServerAlias apache.org *.apache.org

</pre>

<p>Note that you can use <code>*</code> and <code>?</code> as wild-card

characters.</p>

<p>You also might need ServerAlias if you are serving local users who

do not always include the domain name.  For example, if local users are

familiar with typing "www" or "www.physics" then you will need to add

<code>ServerAlias www www.physics</code>.  It isn't possible for the

server to know what domain the client uses for their name resolution

because the client doesn't provide that information in the request.</p>

<h2>Security Considerations</h2>

Apache allows all virtual hosts to be made accessible via the

<code>Host:</code> header through all IP interfaces, even those which

are configured to use different IP interfaces.  For example, if the

configuration for <code>www.foo.com</code> contained a virtual host

section for <code>www.bar.com</code>, and <code>www.bar.com</code> was

a separate IP interface, such that

non-<code>Host:</code>-header-supporting browsers can use it, as

before with Apache 1.0.  If a request is made to

<code>www.foo.com</code> and the request includes the header

<code>Host: www.bar.com</code>, a page from <code>www.bar.com</code>

will be sent.

<P>

This is a security concern if you are controlling access to a

particular server based on IP-layer controls, such as from within a

firewall or router.  Let's say <code>www.bar.com</code> in the above

example was instead an intra-net server called

<code>private.foo.com</code>, and the router used by foo.com only let

internal users access <code>private.foo.com</code>.  Obviously,

<code>Host:</code> header functionality now allows someone who has

access to <code>www.foo.com</code> to get

<code>private.foo.com</code>, if they send a <code>Host:

private.foo.com</code> header.  It is important to note that this

condition exists only if you only implement this policy at the IP

layer - all security controls used by Apache (i.e., <A

HREF="mod/mod_access.html">allow, deny from,</A> etc.) are consistently

respected.

<h2>Compatibility with Older Browsers</h2>

<p>As mentioned earlier, a majority of browsers do not send the

required data for the new virtual hosts to work properly. These

browsers will always be sent to the main server's pages. There is a

workaround, albeit a slightly cumbersome one:</p>

<p>To continue the <code>www.apache.org</code> example (Note: Apache's

web server does not actually function in this manner), we might use the

new <code>ServerPath</code> directive in the <code>www.apache.org</code> virtual host,

for example:

<pre>

    ServerPath /apache

</pre>

<p>What does this mean? It means that a request for any file beginning

with "<code>/apache</code>" will be looked for in the Apache

docs. This means that the pages can be accessed as

<code>http://www.apache.org/apache/</code> for all browsers, although

new browsers can also access it as

<code>http://www.apache.org/</code>.</p>

<p>In order to make this work, put a link on your main server's page

to <code>http://www.apache.org/apache/</code> (Note: Do not use

<code>http://www.apache.org/</code> - this would create an endless

loop). Then, in the virtual host's pages, be sure to use either purely

relative links (e.g. "<code>file.html</code>" or

"<code>../icons/image.gif</code>" or links containing the prefacing

<code>/apache/</code>

(e.g. "<code>http://www.apache.org/apache/file.html</code>" or

"<code>/apache/docs/1.1/index.html</code>").</p>

<p>This requires a bit of

discipline, but adherence to these guidelines will, for the most part,

ensure that your pages will work with all browsers, new and old. When

a new browser contacts <code>http://www.apache.org/</code>, they will

be directly taken to the Apache pages. Older browsers will be able to

click on the link from the main server, go to

<code>http://www.apache.org/apache/</code>, and then access the

pages.</p>

<!--#include virtual="footer.html" -->

</BODY>

</HTML>

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">

<html><head>

<title>An In-Depth Discussion of VirtualHost Matching</title>

</head>

<!-- Background white, links blue (unvisited), navy (visited), red (active) -->

<BODY

 BGCOLOR="#FFFFFF"

 TEXT="#000000"

 LINK="#0000FF"

 VLINK="#000080"

 ALINK="#FF0000"

>

<!--#include virtual="header.html" -->

<h1 ALIGN="CENTER">An In-Depth Discussion of VirtualHost Matching</h1>

<p>This is a very rough document that was probably out of date the moment

it was written.  It attempts to explain exactly what the code does when

deciding what virtual host to serve a hit from.  It's provided on the

assumption that something is better than nothing.  The server version

under discussion is Apache 1.2.

<p>If you just want to &quot;make it work&quot; without understanding

how, there's a <a href="#whatworks">What Works</a> section at the bottom.

<h3>Config File Parsing</h3>

<p>There is a main_server which consists of all the definitions appearing

outside of <CODE>VirtualHost</CODE> sections.  There are virtual servers,

called <EM>vhosts</EM>, which are defined by

<A

 HREF="mod/core.html#virtualhost"

><SAMP>VirtualHost</SAMP></A>

sections.

<p>The directives

<A

 HREF="mod/core.html#port"

><SAMP>Port</SAMP></A>,

<A

 HREF="mod/core.html#servername"

><SAMP>ServerName</SAMP></A>,

<A

 HREF="mod/core.html#serverpath"

><SAMP>ServerPath</SAMP></A>,

and

<A

 HREF="mod/core.html#serveralias"

><SAMP>ServerAlias</SAMP></A>

can appear anywhere within the definition of

a server.  However, each appearance overrides the previous appearance

(within that server).

<p>The default value of the <code>Port</code> field for main_server

is 80.  The main_server has no default <code>ServerName</code>,

<code>ServerPath</code>, or <code>ServerAlias</code>.

<p>In the absence of any

<A

 HREF="mod/core.html#listen"

><SAMP>Listen</SAMP></A>

directives, the (final if there

are multiple) <code>Port</code> directive in the main_server indicates

which port httpd will listen on.

<p> The <code>Port</code> and <code>ServerName</code> directives for

any server main or virtual are used when generating URLs such as during

redirects.

<p> Each address appearing in the <code>VirtualHost</code> directive

can have an optional port.  If the port is unspecified it defaults to

the value of the main_server's most recent <code>Port</code> statement.

The special port <SAMP>*</SAMP> indicates a wildcard that matches any port.

Collectively the entire set of addresses (including multiple

<SAMP>A</SAMP> record

results from DNS lookups) are called the vhost's <EM>address set</EM>.

<p> The magic <code>_default_</code> address has significance during

the matching algorithm.  It essentially matches any unspecified address.

<p> After parsing the <code>VirtualHost</code> directive, the vhost server

is given a default <code>Port</code> equal to the port assigned to the

first name in its <code>VirtualHost</code> directive.  The complete

list of names in the <code>VirtualHost</code> directive are treated

just like a <code>ServerAlias</code> (but are not overridden by any

<code>ServerAlias</code> statement).  Note that subsequent <code>Port</code>

statements for this vhost will not affect the ports assigned in the

address set.

<p>

All vhosts are stored in a list which is in the reverse order that

they appeared in the config file.  For example, if the config file is:

<blockquote><pre>

    <VirtualHost A>

    ...

    </VirtualHost>

    <VirtualHost B>

    ...

    </VirtualHost>

    <VirtualHost C>

    ...

    </VirtualHost>

</pre></blockquote>

Then the list will be ordered: main_server, C, B, A.  Keep this in mind.

<p>

After parsing has completed, the list of servers is scanned, and various

merges and default values are set.  In particular:

<ol>

<li>If a vhost has no

    <A

     HREF="mod/core.html#serveradmin"

    ><code>ServerAdmin</code></A>,

    <A

     HREF="mod/core.html#resourceconfig"

    ><code>ResourceConfig</code></A>,

    <A

     HREF="mod/core.html#accessconfig"

    ><code>AccessConfig</code></A>,

    <A

     HREF="mod/core.html#timeout"

    ><code>Timeout</code></A>,

    <A

     HREF="mod/core.html#keepalivetimeout"

    ><code>KeepAliveTimeout</code></A>,

    <A

     HREF="mod/core.html#keepalive"

    ><code>KeepAlive</code></A>,

    <A

     HREF="mod/core.html#maxkeepaliverequests"

    ><code>MaxKeepAliveRequests</code></A>,

    or

    <A

     HREF="mod/core.html#sendbuffersize"

    ><code>SendBufferSize</code></A>

    directive then the respective value is

    inherited from the main_server.  (That is, inherited from whatever

    the final setting of that value is in the main_server.)

<li>The &quot;lookup defaults&quot; that define the default directory

    permissions

    for a vhost are merged with those of the main server.  This includes

    any per-directory configuration information for any module.

<li>The per-server configs for each module from the main_server are

    merged into the vhost server.

</ol>

Essentially, the main_server is treated as "defaults" or a

"base" on

which to build each vhost.  But the positioning of these main_server

definitions in the config file is largely irrelevant -- the entire

config of the main_server has been parsed when this final merging occurs.

So even if a main_server definition appears after a vhost definition

it might affect the vhost definition.

<p> If the main_server has no <code>ServerName</code> at this point,

then the hostname of the machine that httpd is running on is used

instead.  We will call the <EM>main_server address set</EM> those IP

addresses returned by a DNS lookup on the <code>ServerName</code> of

the main_server.

<p> Now a pass is made through the vhosts to fill in any missing

<code>ServerName</code> fields and to classify the vhost as either

an <EM>IP-based</EM> vhost or a <EM>name-based</EM> vhost.  A vhost is

considered a name-based vhost if any of its address set overlaps the

main_server (the port associated with each address must match the

main_server's <code>Port</code>).  Otherwise it is considered an IP-based

vhost.

<p> For any undefined <code>ServerName</code> fields, a name-based vhost

defaults to the address given first in the <code>VirtualHost</code>

statement defining the vhost.  Any vhost that includes the magic

<SAMP>_default_</SAMP> wildcard is given the same <code>ServerName</code> as

the main_server.  Otherwise the vhost (which is necessarily an IP-based

vhost) is given a <code>ServerName</code> based on the result of a reverse

DNS lookup on the first address given in the <code>VirtualHost</code>

statement.

<p>

<h3>Vhost Matching</h3>

<p><strong>Apache 1.3 differs from what is documented

here, and documentation still has to be written.</strong>

<p>

The server determines which vhost to use for a request as follows:

<p> <code>find_virtual_server</code>: When the connection is first made

by the client, the local IP address (the IP address to which the client

connected) is looked up in the server list.  A vhost is matched if it

is an IP-based vhost, the IP address matches and the port matches

(taking into account wildcards).

<p> If no vhosts are matched then the last occurrence, if it appears,

of a <SAMP>_default_</SAMP> address (which if you recall the ordering of the

server list mentioned above means that this would be the first occurrence

of <SAMP>_default_</SAMP> in the config file) is matched.

<p> In any event, if nothing above has matched, then the main_server is

matched.

<p> The vhost resulting from the above search is stored with data

about the connection.  We'll call this the <EM>connection vhost</EM>.

The connection vhost is constant over all requests in a particular TCP/IP

session -- that is, over all requests in a KeepAlive/persistent session.

<p> For each request made on the connection the following sequence of

events further determines the actual vhost that will be used to serve

the request.

<p> <code>check_fulluri</code>: If the requestURI is an absoluteURI, that

is it includes <code>http://hostname/</code>, then an attempt is made to

determine if the hostname's address (and optional port) match that of

the connection vhost.  If it does then the hostname portion of the URI

is saved as the <EM>request_hostname</EM>.  If it does not match, then the

URI remains untouched.  <STRONG>Note</STRONG>: to achieve this address

comparison,

the hostname supplied goes through a DNS lookup unless it matches the

<code>ServerName</code> or the local IP address of the client's socket.

<p> <code>parse_uri</code>: If the URI begins with a protocol

(<EM>i.e.</EM>, <code>http:</code>, <code>ftp:</code>) then the request is

considered a proxy request.  Note that even though we may have stripped

an <code>http://hostname/</code> in the previous step, this could still

be a proxy request.

<p> <code>read_request</code>: If the request does not have a hostname

from the earlier step, then any <code>Host:</code> header sent by the

client is used as the request hostname.

<p> <code>check_hostalias</code>: If the request now has a hostname,

then an attempt is made to match for this hostname.  The first step

of this match is to compare any port, if one was given in the request,

against the <code>Port</code> field of the connection vhost.  If there's

a mismatch then the vhost used for the request is the connection vhost.

(This is a bug, see observations.)

<p>

If the port matches, then httpd scans the list of vhosts starting with

the next server <STRONG>after</STRONG> the connection vhost.  This scan does not

stop if there are any matches, it goes through all possible vhosts,

and in the end uses the last match it found.  The comparisons performed

are as follows:

<ul>

<li>Compare the request hostname:port with the vhost

    <code>ServerName</code> and <code>Port</code>.

<li>Compare the request hostname against any and all addresses given in

    the <code>VirtualHost</code> directive for this vhost.

<li>Compare the request hostname against the <code>ServerAlias</code>

    given for the vhost.

</ul>

<p>

<code>check_serverpath</code>: If the request has no hostname

(back up a few paragraphs) then a scan similar to the one

in <code>check_hostalias</code> is performed to match any

<code>ServerPath</code> directives given in the vhosts.  Note that the

<STRONG>last match</STRONG> is used regardless (again consider the ordering of

the virtual hosts).

<h3>Observations</h3>

<ul>

<li>It is difficult to define an IP-based vhost for the machine's

    "main IP address".  You essentially have to create a bogus

    <code>ServerName</code> for the main_server that does not match the

    machine's IPs.

    <P>

<li>During the scans in both <code>check_hostalias</code> and

    <code>check_serverpath</code> no check is made that the vhost being

    scanned is actually a name-based vhost.  This means, for example, that

    it's possible to match an IP-based vhost through another address.  But

    because the scan starts in the vhost list at the first vhost that

    matched the local IP address of the connection, not all IP-based vhosts

    can be matched.

    <p>

    Consider the config file above with three vhosts A, B, C.  Suppose

    that B is a named-based vhost, and A and C are IP-based vhosts.  If

    a request comes in on B or C's address containing a header

    &quot;<SAMP>Host: A</SAMP>&quot; then

    it will be served from A's config.  If a request comes in on A's

    address then it will always be served from A's config regardless of

    any Host: header.

    </p>

<li>Unless you have a <SAMP>_default_</SAMP> vhost,

    it doesn't matter if you mix name-based vhosts in amongst IP-based

    vhosts.  During the <code>find_virtual_server</code> phase above no

    named-based vhost will be matched, so the main_server will remain the

    connection vhost.  Then scans will cover all vhosts in the vhost list.

    <p>

    If you do have a <SAMP>_default_</SAMP> vhost, then you cannot place

    named-based vhosts after it in the config.  This is because on any

    connection to the main server IPs the connection vhost will always be

    the <SAMP>_default_</SAMP> vhost since none of the name-based are

    considered during <code>find_virtual_server</code>.

    </p>

<li>You should never specify DNS names in <code>VirtualHost</code>

    directives because it will force your server to rely on DNS to boot.

    Furthermore it poses a security threat if you do not control the

    DNS for all the domains listed.

    <a href="dns-caveats.html">There's more information

    available on this and the next two topics</a>.

    <p>

<li><code>ServerName</code> should always be set for each vhost.  Otherwise

    A DNS lookup is required for each vhost.

    <p>

<li>A DNS lookup is always required for the main_server's

    <code>ServerName</code> (or to generate that if it isn't specified

    in the config).

    <p>

<li>If a <code>ServerPath</code> directive exists which is a prefix of

    another <code>ServerPath</code> directive that appears later in

    the configuration file, then the former will always be matched

    and the latter will never be matched.  (That is assuming that no

    Host header was available to disambiguate the two.)

    <p>

<li>If a vhost that would otherwise be a name-vhost includes a

    <code>Port</code> statement that doesn't match the main_server

    <code>Port</code> then it will be considered an IP-based vhost.

    Then <code>find_virtual_server</code> will match it (because

    the ports associated with each address in the address set default

    to the port of the main_server) as the connection vhost.  Then

    <code>check_hostalias</code> will refuse to check any other name-based

    vhost because of the port mismatch.  The result is that the vhost

    will steal all hits going to the main_server address.

    <p>

<li>If two IP-based vhosts have an address in common, the vhost appearing

    later in the file is always matched.  Such a thing might happen

    inadvertently.  If the config has name-based vhosts and for some reason

    the main_server <code>ServerName</code> resolves to the wrong address

    then all the name-based vhosts will be parsed as ip-based vhosts.

    Then the last of them will steal all the hits.

    <P>

<li>The last name-based vhost in the config is always matched for any hit

    which doesn't match one of the other name-based vhosts.

</ul>

<h3><a name="whatworks">What Works</a></h3>

<p>In addition to the tips on the <a href="dns-caveats.html#tips">DNS

Issues</a> page, here are some further tips:

<ul>

<li>Place all main_server definitions before any VirtualHost definitions.

(This is to aid the readability of the configuration -- the post-config

merging process makes it non-obvious that definitions mixed in around

virtualhosts might affect all virtualhosts.)

<p>

<li>Arrange your VirtualHosts such

that all name-based virtual hosts come first, followed by IP-based

virtual hosts, followed by any <SAMP>_default_</SAMP> virtual host

<p>

<li>Avoid <code>ServerPaths</code> which are prefixes of other

<code>ServerPaths</code>.  If you cannot avoid this then you have to

ensure that the longer (more specific) prefix vhost appears earlier in

the configuration file than the shorter (less specific) prefix

(<EM>i.e.</EM>, &quot;ServerPath /abc&quot; should appear after

"ServerPath /abcdef").

<p>

<li>Do not use <EM>port-based</EM> vhosts in the same server as

name-based vhosts.  A loose definition for port-based is a vhost which

is determined by the port on the server (<em>i.e.</em>, one server with

ports 8000, 8080, and 80 - all of which have different configurations).

<p>

</ul>

<!--#include virtual="footer.html" -->

</BODY>

</HTML>