Web Server Management: Running Apache 2 on Red

Hat Linux

Bob Dowling
University of Cambridge Computing Service
[email protected]
Web Server Management: Running Apache 2 on Red Hat Linux
by Bob Dowling

Installation:This course will first illustrate how to load the Apache 2 package on a Red Hat
Linux version 9 system. This will be specific to this operating system.
ConfigurationThe course will then demonstrate how to configure the web server from the
ground up. The course does not teach tweaking of the default configuration but rather the
writing of a configuration from scratch. The configuration will be suitable for a system running
multiple virtual hosts.
Other utilities:The course will also discuss the associated utilities for log file rotation and
reporting.
Table of Contents
1. Installing the software .....................................................................................................1
The web server group of packages ............................................................................1
Installing the packages ................................................................................................2
Changes made to the system ......................................................................................4
Quick and Dirty Web Server.......................................................................................7
2. The site’s design..............................................................................................................11
The server we want....................................................................................................11
What is a virtual host? ...............................................................................................11
Structures of HTTP queries and responses.............................................................12
3. Getting started.................................................................................................................15
4. Supporting MIME types................................................................................................23
MIME types on a Red Hat Linux system................................................................23
Loading and using the MIME module....................................................................24
5. Symbolic links.................................................................................................................27
6. Handling directories ......................................................................................................31
Using default documents ..........................................................................................32
Automatic indexing of directories ...........................................................................33
Manipulating columns.....................................................................................36
Manipulating rows ...........................................................................................39
Adding icons to the listing ..............................................................................42
Adding text to the listing.................................................................................44
Using an HTML table.......................................................................................46
Summary of the auto-indexing module ........................................................47
Using both modules...................................................................................................49
7. Logging .............................................................................................................................51
The error log................................................................................................................51
Access logs ..................................................................................................................53
Log rotation.................................................................................................................57
Log file analysis ..........................................................................................................59
8. Users’ own web pages....................................................................................................63
9. Delegated control............................................................................................................65
10. Access control ................................................................................................................71
Access control by client IP address..........................................................................71
Access control by client identity ..............................................................................73
Variations on a theme of user identification...........................................................77
Mix and match: Location and Authentication .......................................................78
Blocking access based on a file’s name ...................................................................79
11. Conclusion .....................................................................................................................81
A. Apache modules.............................................................................................................85
B. Reference information for logging .............................................................................89

iii
iv
Chapter 1. Installing the software

Packages:
We will describe the packages that Red Hat install if you request the Web Server group
at installation. We will select the minimal set we need.

Running rpm:
We will demonstrate the use of the rpm utility to install these packages.

Changes to system:
We will briefly review what changes have been made to the system.

The web server group of packages

At installation the administrator is offered the choice of software to install. If the
installer opted for the Server installation option or selected Custom and selected the
Web Server group then all the necessary software is already installed.

Packages included in the Web Server group

httpd
The core web server package. Without this, we don’t have a web server.

httpd-manual
This is the on-line manual for the web server.

mod_python
This allows the use of Python programs alongside the web server in a manner
similar to, but more versatile than, classic CGI programs.

mod_perl
This is exactly the same as mod_python except that it supports the Perl program-
ming language.

mod_ssl
This provides support for HTTPS, the encrypted version of HTTP. This is cov-
ered in a different Computing Service course (Web Server Management: Securing
Access to Web Servers).

hwcrypto
This provides the facility to use hardware cryptographic systems to boost the
speed of HTTPS communications.

php
PHP provides another scripting language for the web server to replace CGI.

php-imap
This package provides the hooks for PHP to use the IMAP protocol to talk to
email systems.

1
Chapter 1. Installing the software

php-ldap
This package provides the hooks for PHP to perform lookups in LDAP directo-
ries.

php-pgsql
This package provides the hooks for PHP to work with PostgreSQL relational
databases.

mod_auth_pgsql
This package provides the facility to use a PostgreSQL database to do password
lookups for access controls rather than the plain text files that are often used.

squid
Squid is a proxy caching server.

tux
Tux is an alternative web server. We will not be discussing tux in this course.

webalizer
Webalizer is a Web server log analysis program. We will examine its use in the
Section called Log file analysis in Chapter 7.
There are other packages released with Red Hat Linux version 9 which depend on
the httpd package and which enhance its functionality. In particular, if you want to
connect to a database running MySQL rather than PostgreSQL then the corresponding
packages are mod_auth_mysql and php-mysql.
For the purposes of this course we will need only some of the packages. We will also
make use of another system package for log rotation. This is a base system package
and should already be installed on the system.

Packages used directly in this course

• httpd
• webalizer
• logrotate

Installing the packages

Location:
We detail where to find the packages. This element is Cambridge-specific.

rpm:
Then we describe exactly how to install the packages.

Components
Finally we review what’s actually been added to the system (before we delete large
chunks of it).

To install the packages we will have to be root. We either log in as root or su to root.
If you su please be sure to use the - option to get the right environment.

2
 Chapter 1. Installing the software

$ \/bin/su -
Password: password
#
Figure 1-1. Changing to be root

Unix Support keeps the Red Hat distributions on an NFS server

nfs-uxsup.csx.cam.ac.uk in the directory /linux/redhat. To access it we will create
a directory to mount the server’s directory tree on, mount it, install the packages
and then unmount the distribution again.

# mkdir /mnt/redhat
# mount -o ro nfs-uxsup.csx.cam.ac.uk:/linux/redhat /mnt/redhat
# cd /mnt/redhat
# ls
5.2 7.1 8.0 code enterprise rawhide
6.2 7.2 9 contrib local_extras updates
7.0 7.3 beta current preview
Figure 1-2. Mounting the NFS server’s Red Hat Linux distributions

Next, we will move to the directory that has the packages for the specific version
of Red Hat Linux that we run. Red Hat have distributions for various different lan-
guages and hardware types. Each of these is available at various different versions.
We start by identifying these to navigate to a directory which is the contents of the
corresponding Red Hat CD.

Locating the CD image

• Version: We are using version 9.
• Language: We want the English language version, identified by code en. Actually,
Unix Support only mirrors the English versions.
• Product: Red Hat distribute a variety of products including the operating system.
They also ship documentation, extra utilities etc. We want the operating system
product, identified by os.
• Architecture: We want the Intel architecture, identified by code i386.

# pwd
/mnt/redhat
# cd 9/en/os/i386
# ls
autorun README.zh_CN
dosutils README.zh_TW
EULA RedHat
GPL RELEASE-NOTES
images RELEASE-NOTES-de.html
isolinux RELEASE-NOTES-es.html
jpk RELEASE-NOTES-fr.html
README RELEASE-NOTES.html
README-Accessibility RELEASE-NOTES-it.html
README.de RELEASE-NOTES-ja.html
README.es RELEASE-NOTES-ko.html
README.fr RELEASE-NOTES-zh_CN.html
README.it RELEASE-NOTES-zh_TW.html
README.ja RPM-GPG-KEY
README.ko SRPMS
Figure 1-3. Navigating to the CD image

This has brought us to the copy of the relevant Red Hat Linux CD. Now we have to
locate the packages. These can be found in the subdirectory RedHat/RPMS.

3
Chapter 1. Installing the software

# cd RedHat/RPMS
# ls
4Suite-0.11.1-10.i386.rpm
a2ps-4.13b-24.i386.rpm
abiword-1.0.2-6.i386.rpm
...
zlib-1.1.4-4.i386.rpm
zlib-devel-1.1.4-4.i386.rpm
zsh-4.0.4-8.i386.rpm
Figure 1-4. Navigating to the package repository

Each of these .rpm files contains a Red Hat package. We are now in a position to in-
stall the ones we want. Each has a file name based on the package name. We will in-
stall the software with the rpm command’s --install option. These filenames tend
to be quite long. Tab-completion can save you a lot of typing. Type the beginning of
a filename and press Tab. The shell will complete the filename as far as it can.

# pwd
/mnt/redhat/9/en/os/i386/RedHat/RPMS
# rpm --install httpd-2.0.40-8.i386.rpm
# rpm --install webalizer-2.01_10-9.i386.rpm
Figure 1-5. Installing the packages

The two filenames could have been given as a list on a single rpm command line.
We have been slightly unfair here. We have installed the version of the web server
that comes with Red Hat Linux version 9. However, there may have been releases
since then that we should upgrade to. You may recall an updates directory in
/mnt/redhat. This contains updates to software that are released after the initial
release.
We will use the rpm command’s --freshen option to update our version of the core
httpd package.

# cd /mnt/redhat/updates/9/en/os/i386
# ls httpd-* logrotate-* webalizer-*
ls: logrotate-*: No such file or directory
ls: webalizer-*: No such file or directory
httpd-2.0.40-11.i386.rpm httpd-manual-2.0.40-11.i386.rpm
httpd-devel-2.0.40-11.i386.rpm
# rpm --freshen httpd-2.0.40-11.i386.rpm
Figure 1-6. Looking for updated packages

Finally we will unmount the NFS file server. To do this we have to leave any of the
directories we have mounted from it as we can’t whip the carpet out from under our
own feet.

# pwd
/mnt/redhat/9/en/os/i386/RedHat/RPMS
# cd
# pwd
/root
# umount /mnt/redhat
Figure 1-7. Disconnecting from the file server

4
 Chapter 1. Installing the software

Changes made to the system

We should quickly examine what changes have been made to the system by the in-
stallation of these packages. The httpd package has obviously installed a web server
(in /usr/sbin/httpd) but it has also installed files in a few other locations.

httpd.conf
conf
magic
conf.d README

/etc/httpd logs /var/logs/httpd empty

modules /usr/lib/httpd/modules mod_*.so

run /var/run *.pid

Figure 1-8. Configuration hierarchy

The /etc/httpd directory contains all the configuration for the web server and is
where the server looks for everything except the web pages themselves. The conf
subdirectory is for the main configuration file, httpd.conf. The conf.d directory is
for extra elements of configuration which can be automatically included in the server
configuration.
Because this directory is where the server looks for most things there are some sym-
bolic links leaving this directory and pointing to where the system expects such things
to go.
The logs symbolic link points into the /var/logs directory where Red Hat Linux
stores its logs. An httpd subdirectory is used because there will typically be more
than one log file in use at any time. This subdirectory holds them together. We will
discuss log files in detail in Chapter 7.
The modules symbolic link points into /usr/lib where Red Hat Linux keeps its li-
braries. Again, an httpd subdirectory is used to keep all the Apache libraries to-
gether. The libraries are called “modules” under Apache and much of the course will
be devoted to what they can do and how to get them to do it.
The run symbolic link points to /var/run. This directory on a Red Hat Linux system
is used by long-running processes to declare that they are running and to indicate
how to contact them. In practice this means that it is full of PID files. These are files,
named after the process they refer to, containing the process ID (PID) of the process.
For Apache, the PID file is called httpd.pid. In practice, we will not need to know
about this file except to understand how the Apache shutdown procedure works in
Chapter 3.

5
Chapter 1. Installing the software

cgi−bin
*.html.var
error
bottom.html
include spacer.html
/var/www top.html

msfree.png
html usage
webalizer.png

icons *.png
Figure 1-9. Initial document hierarchy

The /var/www directory is the default location for files served by the web server. One
of its subdirectories, cgi-bin is empty as Apache does not ship with CGI scripts any
more. This course does not cover CGI programming. There is a different University
Computing Service course, CGI Scripting for Programmers: Introduction, for this.
The html subdirectory is similarly almost empty. Its only content is a directory usage
which has nothing to do with how to run the web server but is, rather, where the
webalizer log analysis application puts its reports on how much the server has been
used. We will cover this in depth in the Section called Log file analysis in Chapter 7. The
html directory is the basic website. Anything put here will appear on the website. See
the Section called Quick and Dirty Web Server for how to get a web site up and running
as quickly as possible.
The icons subdirectory contains the icons used in the automatically generated list-
ings. We will use these extensively when we discuss automatic indexing of directories
in the Section called Automatic indexing of directories in Chapter 6.
The most complex installed directory structure is in error. This contains a large num-
ber of files whose names are HTTP error names with .html.var as a suffix. Each of
these contains the web page that will be displayed if the corresponding error condi-
tion happens. We will not be discussing error documents in this course. Each error
page has the error in four different languages and the web server is geared up to
serve the appropriate language from the choice. (This is what the .var suffix is all
about.)
Note that the /var/www directory tree is owned by root. Any changes to the website
as the system currently stands need to be done by root
Two other changes have been made to the system. A user and group have been cre-
ated for the web server to run as.

$ grep apache /etc/passwd /etc/group

/etc/passwd:apache:x:48:48:Apache:/var/www:/sbin/nologin
/etc/group:apache:x:48:
Figure 1-10. The apache user and group

User programs included with the web server

6
 Chapter 1. Installing the software

• /usr/bin/ab: This a stress-tester for the web server. Please do not stress-test
people’s servers without their explicit permission. Otherwise you may find them
stress-testing your skull with a pickaxe handle.
• /usr/bin/htdbm, /usr/bin/htdigest, /usr/bin/htpasswd: These manipulate
user and password information for web access controls. We will see htpasswd
later in the Section called Access control by client identity in Chapter 10 but we will
not be considering the other commands in this course.
• /usr/bin/logresolve: If a log file contains IP addresses rather than DNS names
for clients then this program will run through the log file and write out a copy with
hostnames replacing IP addresses. Because it caches resolved addresses it does this
rather efficiently. We will be covering log files (and why they might have IP ad-
dresses rather than hostnames) in Chapter 7.

System programs included with the web server

• /usr/sbin/apachectl: This is the script that is provided by the Apache team to
simplify turning on and off the service. However, to keep the startup and shut-
down of the server consistent with the rest of the system the standard startup
scripts don’t use this for the main work.
• /usr/sbin/httpd: This is the web server itself.
• /usr/sbin/rotatelogs: This is an Apache utility providing for rotating log files.
However, it is not used by a Red Hat Linux system because there is a system-
wide log rotation facility which is used instead for consistency with the rest of the
system. This will be considered in detail in the Section called Log rotation in Chapter
7.
• /usr/sbin/suexec: This is a helper program for Apache that lets the server run
external programs (e.g. CGI programs) as a different user than the user running
the web service itself. As we will not be covering CGI programming in this course
we will not be making any use of this program. As it is a setuid root program, you
may want to remove it if you don’t need it.

Quick and Dirty Web Server

The majority of this course concerns the (re)configuration of the web server. How-
ever, we should briefly describe what a system administrator should do if he or she
is happy with the default (which is not a bad set of defaults, by the way).
If the system administrator is happy to do all the changes to the web site as root then
nothing more needs to be done other than turning on the web server.
To enable the web server (so that it gets started at system boot) the system adminis-
trator needs to use the chkconfig command.

# chkconfig --list httpd

httpd 0:off 1:off 2:off 3:off 4:off 5:off 6:off
# chkconfig httpd on
# chkconfig --list httpd
httpd 0:off 1:off 2:on 3:on 4:on 5:on 6:off
Figure 1-11. Enabling the web server

The next time the system is rebooted, the web server will be started. If you don’t
want to wait until a reboot, or don’t want to reboot, then it can be manually started
by running the script that would be run at boot time.

7
Chapter 1. Installing the software

# cd
# /etc/init.d/httpd start
Starting httpd: [ OK ]
Figure 1-12. Manually starting the web server

If you take this easy approach then you need to know the following few facts.

What you get for the quick & dirty approach

• If the server’s DNS name is server then any file placed in
/var/www/html/some/path/here/file.html will be presented as URL
http://server/some/path/here/file.html.
• Two log files will be maintained in the directory /var/log/httpd called
access_log and error_log. These will be rotated weekly and four weeks’ worth
of logs will be kept.
You can make life much simpler for yourself (as the system administrator) if you
create a group of users who are allowed to edit the document tree /var/www/html.
We will create a group, webadmin, who will have access to the site.

Setting up and using the webadmin group

1. Creating the group

# groupadd -r webadmin
The -r option on groupadd sets up a system group. These are no different from
user groups in reality, but Red Hat Linux assigns them from a different range of
numeric IDs to keep them apart.
2. Setting up /var/www/html
Next we have to change /var/www/html so that this newly created group has
sway over it. We need to do a number of things.

• We must change the group of the directory to be webadmin. It starts out con-
trolled by the root group.
• We must change the permissions so that this group can add things.
• We must set the permissions so that anything created in the directory also is
controlled by the webadmin group.
The change of group is done with the chgrp command and the two changes of
permissions can be done with a single use of the chmod command.
# chgrp webadmin /var/www/html
# chmod g+ws /var/www/html
3. Adding users to the group
We will add the users alice and bob to the webadmin group.
You can directly edit the file /etc/group to add users to the group line. They
should be comma-separated with no spaces and no trailing comma.
apache:x:48:
webalizer:x:67:
webadmin:x:101:alice,bob
Alternatively, we can use the usermod command to change the groups that the
users are in. The -G option sets a user’s groups.

8
 Chapter 1. Installing the software

Warning
usermod’s -G option sets the user’s groups. It does not add to
them. You must quote all the user’s groups. Any groups the user
was previously in that are not quoted will be lost by the user.

Suppose alice is in group alpha already. Then to add her to webadmin we must
state that she is in webadmin and alpha.
# usermod -G alpha,webadmin alice
If bob is in no other group, then the command used is easier.
# usermod -G webadmin bob

Note: The users will have to log in again to pick up the groups they have been added
to.

9
Chapter 1. Installing the software

10
Chapter 2. The site’s design

This chapter will describe the design of a web site that we will set as our goal for this
course and discuss a small amount of theory.

Site description:
We will describe the web site that we want to create. This will be a website with a
number of modern features. In particular we will demand the facility to run multiple
virtual hosts (that is, different websites running off the same server).

Virtual hosting:
There will be a brief diversion while we describe exactly how virtual hosting is pos-
sible. There are a variety of different ways to achieve this goal and we will describe
the two most common.

The server we want

We are going to describe the server in the terms that an academic would use to de-
scribe it to a computer officer in the University: annoyingly vague. This is an excuse
for this course to introduce new features one at a time, of course.

Sites:
We are told that the server must serve two web sites, chalk.dept.cam.ac.uk and
cheese.dept.cam.ac.uk whose web pages will be under the control of two different
groups of users.

Facilities:
The “usual facilities” should be provided. This is too vague a specification in reality,
but it is typically all the average academic or manager will ever ask for. In this course
we will assume that this means index pages, automatic directory listing, user home
pages and access controls. To illustrate how to create tied-down servers we will also
design the server to be this and no more.

Logging:
Logs should be kept for as long as possible and usage information should be avail-
able on the web. We will have to consider the DPA implications of this part of the
specification, of course.

What is a virtual host?

There are (at least) three different ways that a single web server can host more than
one web site.

Multiple ports:
A web server can listen on more than the default port (number 80) and offer different
web sites on each port. To identify a non-standard port, its number must follow the
server name in the URL: http://www.dept.cam.ac.uk:port/some/path/here/. The
receiving system uses the port number of the incoming query to distinguish between
web sites.

11
Chapter 2. The site’s design

Multiple addresses:
A single system need not have a single IP address. It can have many and each can
have a different web site attached to it. This leads to two different server names ap-
pearing in standard URLs (i.e. there’s no :port element in the URL) but the two
server names correspond to the two different IP addresses of the system and corre-
spond to different web sites. The receiving web server uses the destination IP address
on the incoming query to distinguish web sites.

Multiple aliases
Also known as name-based virtual hosting, this is the most common form of virtual
hosting. The server only has a single IP address, but two different names in the DNS
correspond to that address. So chalk.dept.cam.ac.uk and cheese.dept.cam.ac.uk both
map on to the same IP address and therefore the same server. This raises the question
of how the web server can distinguish requests to the two different web sites.

Structures of HTTP queries and responses

To answer this question, we need to consider, briefly, the nature of a web request.
Exactly what gets sent to a server when a URL is requested? (And for that matter,
what gets sent back?)

GET /index.html HTTP/1.1

Host: noether.csi.cam.ac.uk
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003
Accept: text/html,text/plain;q=0.8,image/png,image/jpeg,image/gif;q=0.2,*/*;q=0.1
Accept-Language: en, es;q=0.50
Accept-Encoding: gzip, deflate, compress;q=0.9
Accept-Charset: ISO-8859-1, utf-8;q=0.66, *;q=0.33
Connection: keep-alive
Keep-Alive: 300

Figure 2-1. Incoming HTTP query for http://noether.csi.cam.ac.uk/index.html

To understand name-based virtual hosting consider just the first two lines. The GET
request only refers to the local element of the URL. The second line specifies the
hostname that is being asked for it.

HTTP Query Structure

GET
The first line declares that this is a request from a client that wishes to read in-
formation from the server. GET is the most common HTTP method.

/index.html
The second term in the first line is the local element of the URL requested. Note
that the leading part of the URL containing the server name has been stripped
out.

HTTP/1.1
The final element declares that the query is couched in the language of version
1.1 of the HTTP standard.

12
 Chapter 2. The site’s design

Host: noether.csi.cam.ac.uk
The second line indicates which server the query was addressed to. It is this
element of the query that allows a web server to distinguish between web sites
based purely on their names, regardless of the port number(s) or IP address(es)
used.

User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003

This optional line identifies the browser. Some servers, designed by stupid peo-
ple, vary the output according to this header. It is a hint and can be trivially
changed on many browsers.
In this case Mozilla identifies the browser as one of the Netscape/Mozilla fam-
ily and 5.0 ties it down to a version of Mozilla. X11 indicates that it is a browser
with access to an X windowing system, U identifies it as a Unix and Linux i686
specifies what sort of Unix. en-US specifies the locale the browser is working in.
rc:1.0.1 specifies the version of the browser. Gecko specifies the rendering en-
gine the browser uses and 20021003 is the release date (as version number) of
the version.

Accept: text/html,text/plain;q=0.8,image/png,image/jpeg,image/gif;q=0.2,*/*;q=0.1
This specifies the formats the browser can accept and how keen it is on them.
Servers can be configured to negotiate various different formats of response de-
pending on these parameters.
text/html means that the browser is happy to accept MIME content type
text/html and text/plain means that it can accept plain text too. The qualifier
q=0.8 means that, given a choice, the browser would prefer to receive text/html
(default q=1.0) than text/plain (q=0.8). Similarly for images, the browser would
prefer image/png or image/jpeg to the image/gif format. Finally it will accept
any format (*/*) but is not keen on them (q=0.1).
We will meet MIME content types again in Chapter 4.

Accept-Language: en, es;q=0.50

Just as it is possible to negotiate formats it is possible to negotiate languages. A
page might appear in more than one language and the browser specifies what
languages it can cope with and how desirable they are. The author is learning
Spanish and has Spanish as a second choice in the language selections.

Accept-Encoding: gzip, deflate, compress;q=0.9

Just as there was negotiation over MIME content type there can also be nego-
tation over MIME transfer encoding. This is a mechanism for the server and
browser to agree on a way to (typically) compress the data stream prior to trans-
fer.

Accept-Charset: ISO-8859-1, utf-8;q=0.66, *;q=0.33

The final topic for negotiation is the character set of any text that will be sent. In
this case, ISO Latin 1 is preferred, with UTF-8 a second choice and everything
else coming third.

Connection: keep-alive
This tells the server that it need not close the network connection after sending
back the response to the query as other requests may be sent down the same con-
nection. As setting up and tearing down connections are expensive operations
this is a major fficiency boost.

13
Chapter 2. The site’s design

Keep-Alive: 300
This instructs the server to keep the connection alive for 300 seconds in case
there are any more requests. After 300 seconds of idleness the server will drop
the connection.
For the record, I’ll also post the response. To make the example work, I’ve installed a
trivial index.html web page. We will use this later

HTTP/1.1 200 OK
Date: Wed, 19 Mar 2003 09:41:23 GMT
Server: Apache/2.0.40 (Red Hat Linux)
Last-Modified: Wed, 19 Mar 2003 09:40:21 GMT
ETag: "e011-13e-12910b40"
Accept-Ranges: bytes
Content-Length: 318
Connection: Keep-Alive
Content-Type: text/html; charset=ISO-8859-1

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

"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
<head>
<title>The DEFAULT web site</title>
</head><body>
<h1>Welcome to DEFAULT</h1>
<p>This is the DEFAULT web site.</p>
</body>
</html>
Figure 2-2. Outgoing HTTP response from the server

14
Chapter 3. Getting started

Clean slate:
We will start by removing the existing configuration script. This may seem dramatic
but this course seeks to explain every single line of the configuration file that will
finally be written. After we delete the file we will note that the web server won’t
start.

Simplest configuration:
The aim of this section is to give us enough configuration so that the server will at
least start, even if it won’t do anything useful.

Two more lines

We will add two more lines to the most basic configuration file we can have. The
first will just help the configuration file make more sense and the second will turn off
many defaults so we can see them explicitly when we need them.

Deleting the configuration file is easy. Go on, you know you’ve always wanted to do
it! What’s the worst that could happen?

# rm /etc/httpd/conf/httpd.conf
rm: remove regular file ‘/etc/httpd/conf/httpd.conf’? y
Figure 3-1. Deleting the configuration file

The web server will not start now. First it will complain about not having a configu-
ration file.

# /etc/init.d/httpd start
Starting httpd: grep: /etc/httpd/conf/httpd.conf: No such file or directory
httpd: could not open document config file /etc/httpd/conf/httpd.conf
[FAILED]
Figure 3-2. Failing to start the server with no configuration

Close observers will notice that it complains twice. The first is an error from the grep
command. This is just a Red Hat Linux feature that we can ignore. (The Red Hat
Linux start script checks to make sure that you aren’t using an Apache 1 configuration
file.) The second error message comes from httpd and is the one we are interested in.
Next, we will create an empty configuration file and see that that just changes the
error message.

# touch /etc/httpd/conf/httpd.conf
# /etc/init.d/httpd start
Starting httpd: no listening sockets available, shutting down
[FAILED]
Figure 3-3. Failing to start the server with an empty configuration

It must be admitted that as error messages go, “no listening sockets available, shut-
ting down” is a fairly obscure way of saying “you’ve not told me what to do”. Actu-
ally it means, “you’ve not told me to listen for any incoming requests so I might as
well quit now”.

15
Chapter 3. Getting started

We will start by detailing an absolutely minimal configuration file that gets the server
launched but nothing else.

Listen 80
Figure 3-4. httpd.conf: The Listen command

The command to tell the server to listen for connections is Listen. This takes one
argument, specifying which interface and port to listen on. The default port assigned
to web services by the Internet authorities is port 80. Quoting just a port number
means to listen on that port number on every IP-enabled interface. Simply to launch
the web server this is all we need!

# /etc/init.d/httpd start
Starting httpd: [ OK ]
Figure 3-5. Launching the web server with a minimal configuration

Unfortunately, the launched web server then immediately shuts down. By default,
the web server will log error messages in /var/log/httpd/error_log. We can look
in there for clues as to what we need next.

[Mon Mar 17 17:51:13 2003] [notice] Apache/2.0.40 (Red Hat Linux)

configured -- resuming normal operations
[Mon Mar 17 17:51:13 2003] [alert] (2)No such file or directory:
getpwuid: couldn’t determine user name from uid 4294967295, you
probably need to modify the User directive
...
[Mon Mar 17 17:51:14 2003] [alert] Child 9315 returned a Fatal error...
Apache is exiting!
Figure 3-6. error_log: Launch and fail

What does this error message mean? It means that the web server needs to know
who to run as. You will recall that the software installation created a user and group
for the server to run as. We need to tell it to use them. This is done with the User and
Group commands in the configuration file.

User apache
Group apache
Figure 3-7. httpd.conf: User and group

We have cheated. There is one configuration line we have omitted because in the Red
Hat Linux build of Apache there is an appropriate, compiled-in default. A number
of files will be referred to in this configuration file (such as the modules we will start
to meet in Chapter 4 and the log files we will meet in Chapter 7) and we want to
give relative path names for these files. We need to specify one absolute path for
these to be relative to. We add the ServerRoot command to identify this directory for
completeness.
The second “unnecessary” line we will add has the effect of turning off various set-
tings which default to being on. We do this for two reasons. The first is didactic; we
want to meet these options explicitly when they become relevant rather than relying
on defaults. The second is our decision to provide what was specified and no more.
This line will turn off everything and we must explicitly turn on what we want.

16
 Chapter 3. Getting started

Listen 80
User apache
Group apache
ServerRoot /etc/httpd
Options None
Figure 3-8. httpd.conf: Minimal version

Syntax summary: Getting launched

Listen
This specifies the network interface the server should listen on. If only a port
is specified then it will listen on every active network address. (Typically the
system’s own network address and the loopback address.)

User
This specifies the system user ID that the server process should run as. This user
was created by the Red Hat Linux Apache package.

Group
This specifies the system group ID that the server process should run as. This
group was created by the Red Hat Linux Apache package.

ServerRoot
All unqualified filenames in the configuration file will be resolved with respect
to this directory except for the actual web pages which will be handled by a
different command.

Options
This command sets various parameters in the configuration. We will meet these
through the course as we turn them on.
And if we start the web server now, with this five line configuration file, it launches
just fine and stays running.

# /etc/init.d/httpd start
Starting httpd: [ OK ]
# tail -1 /var/log/httpd/error_log
[Mon Mar 17 18:02:56 2003] [notice] Apache/2.0.40 (Red Hat Linux) configured -- resuming norma
# ps -ef | grep httpd
root 9344 1 0 18:02 ? 00:00:00 /usr/sbin/httpd
apache 9345 9344 0 18:02 ? 00:00:00 /usr/sbin/httpd
apache 9346 9344 0 18:02 ? 00:00:00 /usr/sbin/httpd
apache 9349 9344 0 18:02 ? 00:00:00 /usr/sbin/httpd
apache 9350 9344 0 18:02 ? 00:00:00 /usr/sbin/httpd
apache 9351 9344 0 18:02 ? 00:00:00 /usr/sbin/httpd
root 9357 8826 0 18:05 pts/3 00:00:00 grep httpd
Figure 3-9. Successful launch

But, as the figure shows, it’s not a single daemon that gets launched. There are six of
them. The first column of the ps output gives the owner of the process and the second
gives the process ID or PID. One of the server processes is owned by user root and
the others by user apache. That root-owned process is the parent process of all the
other processes. What happens is that the startup script that we manually invoked
launched the parent, root-owned process (PID 9344). It in turn launched five child
processes owned by apache (PIDs 9345–9351).
Why? Well, the idea is that the parent process does not service any request at all. Its
sole purpose is to keep an eye on the child processes. If one of them dies for any
17
Chapter 3. Getting started

reason the parent decides whether or not to replace it. (If they have all been idle for
the past 48 hours it may decide that four processes are plenty.) If they are all kept too
busy the parent may choose to start up some more processes to share the load. The
set of child processes is called the server pool and is the traditional mechanism that
Apache has always used to provide rapid responses. This way of working is called
the pre-forked model.
There are other models other than pre-forked, but Red Hat Linux has this enabled by
default and it is what we will use throughout the course.

Figure 3-10. No documents!

At the moment, the server has nothing to serve. Every attempt to request a
page from it results in a 404, not found error. If we look in the error log file,
/var/log/httpd/error_log we will see the error message:

[Tue Mar 18 13:42:10 2003] [error] File does not exist: /etc/httpd/htdocs

which indicates that the server is looking in the wrong place!.

We need to tell it where to look. However, as we are planning on hosting two web
sites we ought to be thinking about two locations, one for each value of the Host:
header. We should also think about what to do with requests that have neither
chalk.dept.cam.ac.uk nor cheese.dept.cam.ac.uk as the Host: header’s value.

Use valid DNS names

The names used for the virtual hosts cannot just be made up. They
must be registered in the DNS for the IP address of the server being
used. Typically there will be a “real” name for the server (given by the
DNS A record) and a number of aliases for the web sites (given by DNS
CNAME records).

We shall create two subdirectories, CHALK and CHEESE, of /var/www for the two web-
sites. We will also create two groups, chalk and cheese, which will contain the people
18
 Chapter 3. Getting started

entitled to update the sites.

# cd /var/www
# mkdir CHALK CHEESE
# groupadd -r chalk
# groupadd -r cheese
# chgrp chalk CHALK
# chgrp cheese CHEESE
# chmod g+ws CHALK
# chmod g+ws CHEESE
Figure 3-11. Setting up the two web sites

To let us know we have reached the right directory we will put a file, index.html in
each directory identifying it.

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

"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
<head>
<title>The CHALK web site</title>
</head><body>
<h1>Welcome to CHALK</h1>
<p>This is the CHALK web site.</p>
</body>
</html>
Figure 3-12. The chalk index.html file

Now we must tell the web server to use these two directories appropriately.

NameVirtualHost *

<VirtualHost *>
ServerName chalk.dept.cam.ac.uk
DocumentRoot /var/www/CHALK
</VirtualHost>

<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
</VirtualHost>
Figure 3-13. httpd.conf: Setting up the virtual hosts

To set up a named-based virtual host we add a section like the one shown in the
figure above to the configuration file. Two such sections should be added, one for
chalk and one for cheese. So what does it mean?

Syntax summary: Virtual hosts

NameVirtualHost interface
This instructs the web server to run name-based virtual hosts on interface. If
the specified interface is * then all available interfaces are used.

<VirtualHost>
The VirtualHost section describes a single virtual host. Everything from the
<VirtualHost interface> to </VirtualHost> sets parameters for a single vir-
tual host. The interface specified must match one previously set up for named-
based virtual hosting by a NameVirtualHost command.

19
Chapter 3. Getting started

ServerName
This sets the name of the server for the virtual host. If a query’s Host: header
does not match this then the virtual host block will not be applied.

DocumentRoot
This command specifies where the server should look for its documents for the
particular virtual host. This is where we get to split up our various hosts into
different directories.
We do not need to restart our web server after each change to the configuration file.
A rather faster mechanism is to cause it to reread its file to note changes. This is done
by using the reload option on the startup script.

# /etc/init.d/httpd reload
Reloading httpd: [ OK ]
Figure 3-14. Getting the server to reread its configuration file

We are now running one web server supporting two web sites. However, if we re-
quest the index.html page from chalk.dept.cam.ac.uk then we get the source of the
homepage and not the HTML rendering of it. We still have work to do.

Figure 3-15. Contacting the chalk.dept.cam.ac.uk web site; index.html

For completeness we should cover the assorted options that can be passed to the
startup script beyond the start, restart and reload options we have met already..

Options to the startup script /etc/init.d/httpd

start
Starts the web server.

20
 Chapter 3. Getting started

stop
Stops the web server.

restart
Stops and starts the web server.

condrestart
Stops and starts the web server if the PID file exists that suggests that the web
server was started via this script, rather than just manually.

status
Indicates whether or not the web server is running.

fullstatus
This option does not run on Red Hat Linux.

reload
Causes a running web server to reread its configuration file(s) and to reopen its
log files.

graceful
Equivalent to restart but politer.

help
Not much help!

configtest
Does not launch a web sever but forces it to parse the configuration file for syn-
tactic validity.

21
Chapter 3. Getting started

22
Chapter 4. Supporting MIME types

MIME types:
We will start with a very brief discussion about what MIME types are and in partic-
ular what MIME content types are. We will also see how they are associated with file
suffixes in a particular system configuration file.

Modules:
We will then introduce the concept of the module and, in particular, the module that
allows the web server to interpret MIME types.

Let’s take another look at the headers that get sent back by a fully configured web
server.

HTTP/1.1 200 OK
Date: Wed, 19 Mar 2003 09:41:23 GMT
Server: Apache/2.0.40 (Red Hat Linux)
Last-Modified: Wed, 19 Mar 2003 09:40:21 GMT
ETag: "e011-13e-12910b40"
Accept-Ranges: bytes
Content-Length: 318
Connection: close
Content-Type: text/html; charset=ISO-8859-1
Figure 4-1. Outgoing HTTP response headers from a working server

In particular note the Content-Type: header. This identifies the document served as
being of MIME content type text/html. This informs the browser that the document
should be parsed as HTML rather than as plain text. It also tells the browser that the
underlying character set used for the web page is ISO-8859-1.
Now let’s look at the headers coming from our server as it currently stands.

HTTP/1.1 200 OK
Date: Wed, 19 Mar 2003 10:02:11 GMT
Server: Apache/2.0.40 (Red Hat Linux)
Last-Modified: Tue, 18 Mar 2003 14:22:06 GMT
ETag: "2813-138-e4577f80"
Accept-Ranges: bytes
Content-Length: 312
Connection: close
Content-Type: text/plain
Figure 4-2. Outgoing HTTP response headers from our server

The principal difference is that the Content-Type: header now reads text/plain.

MIME types on a Red Hat Linux system

So, how does the system associate MIME content types with files? There are two
ways.

23
Chapter 4. Supporting MIME types

Content analysis
The first is to look in the file’s content and deduce the MIME content type from the
content.
You can see this mechanism in action with the file command. This command can give
a “human readable” description of a file’s content type or, with the -i option, it can
give a MIME content type.

$ file course.pdf
course.pdf: PDF document, version 1.2
$ file course.ps
course.ps: PostScript document text conforming at level 2.0
$ file -i course.ps
file: Using regular magic file ‘/usr/share/magic.mime’
course.ps: application/postscript
$ file -i course.ps 2> /dev/null
course.ps: application/postscript
$ file -i course.pdf 2> /dev/null
course.pdf: application/pdf
Figure 4-3. Using the file command

As you will deduce from the warning message printed on standard error, the file
/usr/share/magic.mime is used to store the information about how to map from
content to MIME type. The default file /usr/share/magic is used for the more ver-
bose descriptions.

File name analysis

The other approach is to use the file name. In particular it is traditional that files
should have particular suffices according to their MIME content types. This is the
most commonly used approach.
This approach is taken by other utilities than just the web server and there is a system
wide file giving the correspondence between file names and MIME content types.
This file is /etc/mime.types which is part of the base system (as part of the mailcap
package).

application/msword doc
application/pdf pdf
application/postscript ai eps ps
application/rtf rtf
application/x-bzip2 bz2
application/x-dvi dvi
application/xml
audio/mpeg mpga mp2 mp3
image/png png
model/vrml wrl vrml
text/html html htm
text/plain asc txt
video/mpeg mpeg mpg mpe
video/quicktime qt mov
Figure 4-4. Extracts from /etc/mime.types

Apache is capable of both modes of operation. We will use the latter as it is more
common. This is for historical reasons and is not a reflection on the relative values of
the two mechanisms.

24
 Chapter 4. Supporting MIME types

Loading and using the MIME module

Modules:
All the various elements of web server functionality are split out into modules. These
are shared libraries that the web server loads when instructed to by its configuration
file. In turn, the presence of a module causes new commands to be available in the
configuration file corresponding to the newly available functionality.
We start by adding the line to the configuration file that loads the module.

LoadModule mime_module modules/mod_mime.so

TypesConfig /etc/mime.types
Figure 4-5. httpd.conf: Loading the MIME module

The LoadModule command takes two arguments. The second is the filename of the
shared library that it needs. Note that the pathname of the module is given relative to
the /etc/httpd directory. The first argument is the name of the module within that
file.
Normally you would consult the documentation to determine what a module’s name
is but there is a filthy hack you can often use to determine it automatically.

$ nm /etc/httpd/modules/mod_mime.so | grep --word D

000031e0 D mime_module
Figure 4-6. Getting module names from shared libraries

Don’t worry about what the line means in detail. If you really want to know then
read the fine manual page for nm. The third element in the answer is the module
name.
A list of all the common modules, together with their library file names, module
names and brief descriptions is given in Appendix A at the end of these notes.
The TypesConfig command indicates the file that has the correspondences between
file name suffixes and MIME content types.
So how does our web server work now? The pages are now presented as HTML.

25
Chapter 4. Supporting MIME types

Figure 4-7. chalk.dept.cam.ac.uk index page as HTML

Syntax summary: Loading modules

LoadModule library.so name
Load the module named name found in the shared library file library.so.

Syntax summary: mime_module

TypesConfig file
This command specifies the name of the file that does the lookup from
filename suffix to MIME content type. On a Red Hat Linux system this file is
/etc/mime.types and is installed by the mailcap package.

26
Chapter 5. Symbolic links

Symbolic links
It is demonstrated that the web server does not follow symbolic links unless explicitly
directed to do so.

Options FollowSymLinks
The instruction to make the server follow symbolic links is introduced.

Options
There is general discussion about the Options command.

Options SymLinksIfOwnerMatch
The more restrictive command is also introduced.

We can now see the index.html file as expected but if we create a symbolic link
called main.html to index.html and ask for that we get a failure.

$ cd /var/www/CHALK/
$ ln -s index.html main.html
$ ls -l
total 4
-rw-rw-r-- 1 rjd4 chalk 312 may 7 10:01 index.html
lrwxrwxrwx 1 rjd4 chalk 10 may 7 10:19 main.html -> index.html
Figure 5-1. Creating the symbiolic link main.html

We note that when we try to access the symbolic link we get a 403 “Forbidden” error.
The web server has found the symbolic link but has decided not to follow it.

27
Chapter 5. Symbolic links

Figure 5-2. We are forbidden to access main.html

To instruct the web server to follow symbolic links we need to set an option. You will
recall we unset all options with Options None in the configuration file. Now we need
to turn on one of them.
We can do this with the command Options FollowSymLinks but this has a cer-
tain subtly we need to understand. The command Options FollowSymLinks sets
the FollowSymLinks option and unsets all of the others. The Options command fol-
lowed by a list of options is absolute; precisely the options specified will be set and
no others. For this reason we will introduce the syntax for setting (and unsetting)
individual options while leaving the others unchanged.

Options +FollowSymLinks
Figure 5-3. Setting a single option

There is an analogous syntax with a minus sign for turning off options while leaving
others untouched.

28
 Chapter 5. Symbolic links

Figure 5-4. We are permitted to access main.html

Because symbolic links might be used to circumvent access controls in the web
server there is a modified version of this option with the rather unwieldy name
SymLinksIfOwnerMatch. This instructs the web server to follow the symbolic link if
and only if the symbolic link’s owner (typically the user who created it) and the
target’s owner match.

Syntax Summary: The Options command

Options option1 option2 option3
Set options option1, option2 and option3 and unset all others.

Options +option1 +option2 +option3

Set options option1, option2 and option3 leaving all other options unchanged.

Options -option1 -option2 -option3

Unset options option1, option2 and option3 leaving all other unchanged.

Options +option1 +option2 -option3

Set options option1, option2 and unset option3 leaving all other unchanged.

Syntax Summary: Options for Symbolic Links

Options FollowSymLinks
The server will follow any symbolic link.

Options SymLinksIfOwnerMatch
The server will follow any symbolic link owned by the same user as its target.

29
Chapter 5. Symbolic links

30
Chapter 6. Handling directories

Directory URLs
Some URLs (typically those that end in /) correspond to directories rather than plain
files. We need to determine how to deal with these.

dir_module
The dir_module module provides the facility of using a default file (typically called
index.html) in a directory on behalf of the directory itself.

autoindex_module
The autoindex_module module provides the facility of generating a listing of the files
and subdirectories within a directory.

Combining approaches
We will also consider how to combine the two approaches so that an index.html file
is used if present and a directory listing is used if not.

A document has a MIME content type. We’ve already seen that the URLs that cor-
respond to files can now be served with the correct MIME type, so long as the file’s
name has the appropriate suffix. But what about directories? In particular a “top
level” URL such as http://cheese.dept.cam.ac.uk/ will trigger a request for /.
There’s no such file so the server can’t serve it yet, let alone determine a MIME con-
tent type for it.

Figure 6-1. / causing problems

31
Chapter 6. Handling directories

There are two solutions to this issue. The first, and simplest, is to nominate a filename
(such as index.html) and instruct that if a directory is requested which contains a
file with this name then that file should be treated as the target of the request. This is
the approach we will follow in the Section called Using default documents.
The other approach is to give a listing of the directory’s contents, typically in HTML
format. We will cover this in the Section called Automatic indexing of directories.

Using default documents

Loading the module
We will add lines to the configuration file to load the module and to set up a default
file. Because we want this functionality for both websites we place the instruction
outside and before the virtual host sections.

LoadModule dir_module modules/mod_dir.so

DirectoryIndex index.html index.htm
Figure 6-2. httpd.conf: Using dir_module

Note that we have a new command DirectoryIndex which is the only additional
command provided by dir_module. It is passed a list of the defaults to use if the
directory is looked for. If a directory is requested then the web server will look for
index.html in the directory because that filename is the first argument to the com-
mand. If index.html is missing then the server will look for index.htm, the second
quoted name. If neither is present then the web server will give a not found error.
Remember the server must reload its configuration to pick up these new instructions.

Figure 6-3. / equivalent to /index.html

32
 Chapter 6. Handling directories

Syntax summary: dir_module

DirectoryIndex
Specify the documents to be searched for given a directory query.
Document names are resolved relative to the directory queried unless they start
with a / when they are resolved relative to the document root.

Automatic indexing of directories

The second approach we will consider is that of automatically generating a list of the
entries in the directory.
The relevant module is rather old and clunky, hailing back to the days when browsers
didn’t support tables in HTML, but it is in very widespread use so we need to con-
sider it. We will start by loading the module and removing dir_module (for simplicity
at this stage).
This is also the largest module (in terms of number of commands) we will cover in
this course. If you can cope with this one, you can cope with any of them.

LoadModule autoindex_module modules/mod_autoindex.so

Figure 6-4. httpd.conf: Loading autoindex_module

If we just load the module then we see that, instead of getting a 404 “Not found” error
we get a 403 “Forbidden” error instead. The web server now knows how to handle
directories but has decided not to.

Figure 6-5. Refusal to handle a directory

As with symbolic links above (see Chapter 5) we need to set an option to instruct the
module to do its job. Note that this use of Options follows the loading of the module.
33
Chapter 6. Handling directories

Several options we will meet rely on a specific module and their use must follow the
LoadModule lin in the configuration file.

LoadModule autoindex_module modules/mod_autoindex.so

Options +Indexes
Figure 6-6. httpd.conf: Loading and enabling autoindex_module

And now, if we ask for the / URL we get the list of the single file (index.html) that
appears in the top-level directory.

Figure 6-7. The top-level listing

By default, the index produced is a very simple one (an itemized list in HTML). The
module provides a command, IndexOptions, which allows us to present a bit more
information about the files.
To see it in operation we will add a simple IndexOptions command to our configu-
ration file to turn on “fancy indexing”.

IndexOptions FancyIndexing
Figure 6-8. httpd.conf: Turning on fancy indexing

Now when we ask for the / URL we get a different format of output.

34
 Chapter 6. Handling directories

Figure 6-9. The fancy index of /

Columns in default fancy index

• Name
• Size
• Last modified
• Description
The layout of this page is very primitive. Arguably the presentation of this
information should be an HTML table, but the auto-indexing module predates the
widespread support for tables in web browsers so a plain text layout, limited to
80 columns wide, is used by default. We will show how to override this default
below. Because the text is so limited in width, facilities are provided to select which
columns are shown and there is limited support for changing the widths of columns.
There are also controls on which rows get shown.
So we can investigate this module more closely, we will populate the
cheese.dept.cam.ac.uk web site.

35
Chapter 6. Handling directories

Figure 6-10. A populated cheese.dept.cam.ac.uk website

Manipulating columns
We will start by making some use of the Description column. There are mechanisms
for manually adding descriptions to entries in the list, but these additions are usually
delegated to the directory in question rather than done in the main configuration file.
Therefore we will leave this topic until we cover delegation in Chapter 9. Instead,
we will use an Apache facility to parse the HTML in web pages, extract the <TITLE>
entries and uses these as the descriptions. This is done by adding another option to
the IndexOptions command.

IndexOptions FancyIndexing ScanHTMLTitles

Figure 6-11. httpd.conf: Adding title-based descriptions

Now the right-hand column has content.

36
 Chapter 6. Handling directories

Figure 6-12. The cheese.dept.cam.ac.uk website with descriptions

Next, we will remove unwanted columns. For the sake or argument, suppose we
don’t want the Size or Last modified columns. These can be suppressed with two
more indexing options.

IndexOptions FancyIndexing ScanHTMLTitles SuppressSize SuppressLastModified

Figure 6-13. httpd.conf: Suppressing the Size and Last updated columns

37
Chapter 6. Handling directories

Figure 6-14. The displayed listing with columns suppressed

Finally, we will look at modifying the widths of the various columns we have. We
will modify beta.html to have a very long title, pushing the Description column too
far.

Figure 6-15. Too long a description

Because the module dates back to the very beginning, it is designed around the idea
of the text-based browser with an 80-character width. The module truncates the de-
38
 Chapter 6. Handling directories

scription of the file if it would push the row beyond this point and indicates the trun-
cation with a “>”. There are two indexing options which can be used to alleviate this
behaviour, NameWidth and DescriptionWidth. Each of these can be used to specify
an absolute character width of the column or can be used to instruct the server to
determine the longest item in the list and adjust accordingly. We will use this second
approach.

IndexOptions FancyIndexing ... NameWidth=* DescriptionWidth=* ...

Figure 6-16. httpd.conf: Auto-adjusting the column widths

Figure 6-17. A very wide description column

Manipulating rows
The first thing we will do is to suppress certain rows from the listing. Why would we
want to do this? Well, suppose the web developers edit their files in place (i.e. in the
directory managed by the web server) with an editor (emacs, say) that while editing
a file (delta.html, say) creates work files (#delta.html#) while it is running and
leaves behind backup files (delta.html~) when it is finished. We don’t want these
files appearing in the listings.

39
Chapter 6. Handling directories

Figure 6-18. Unwanted files in the listing

IndexIgnore "#*#" "*~"

Figure 6-19. httpd.conf: Ignoring certain files

Note that the two expressions to be ignored are placed in quotes. This is not typically
necessary but under certain circumstances it is required, so the author tries to keep in
the habit of doing it always. In this case the “#” character is the comment character in
httpd.conf files. If it was not enclosed in quotes then everything on the IndexIgnore
line beyond the first “#” would be ignored.

40
 Chapter 6. Handling directories

Figure 6-20. Unwanted files removed from the listing

Warning
Just because a file name is not in the listing does not mean that it
cannot be downloaded. If I see alpha.html and guess that there might
be an alpha.html~ I can still request it and the server will serve it to
me. We will deal with blocking these downloads in the Section called
Blocking access based on a file’s name in Chapter 10.

We also have the issue of ordering to consider. At the moment the directories
and plain files are intermingled. We can split these off with the IndexOptions
FoldersFirst option. There is no equivalent FoldersLast option.

IndexOptions FancyIndexing ... FoldersFirst

Figure 6-21. httpd.confPutting subdirectories first in the listing

41
Chapter 6. Handling directories

Figure 6-22. Putting subdirectories first in the listing

Note: By now you may be starting to get confused about when a facility stands alone as
a command like IndexIgnore and when it is an IndexOptions option like FoldersFirst.
Me too.

Adding icons to the listing

The next prettying up of the listing will be to add icons to the listing. Typically, icons
are used to represent the MIME content type of the file. We will use the icons in the
/var/www/icons directory which are provided for this purpose.
We are immediately presented with a problem. The icons directory is not in either
web site’s DocumentRoot. To make it available to all our web sites we are going to
introduce another facility: aliasing. This comes courtesy of the alias_module module
and its Alias command.

LoadModule alias_module modules/mod_alias.so

Alias /icons/ /var/www/icons/
Figure 6-23. Setting up an alias

The Alias command overrides the DocumentRoot for specific URLs. In this case any
URL whose local part starts with /icons/ (n.b. the trailing slash) will be looked up
in /var/www/icons/. Is we place this directive before the definitions of the virtual
hosts then it will apply to both.
Once the module has been loaded, the Alias command may be run multiple times,
both inside and outside of the virtual host sections. If it appears within a virtual host’s
paragraph then it applies to just that virtual host.
The file icon.sheet.png in the icons directory gives a quick lookup of all the icons
provided. Now we have access to the icons we need to know how to make use of

42
 Chapter 6. Handling directories

them in directory listings. The auto-indexing module provides a slew of commands

for this purpose. The trick to producing self-consistent indexes is to use as few as
possible. We will set up distinct icons for the following entries.

Categories with distinct icons

• HTML web pages
• Plain text pages
• Any other “text” format
• Any image format
• Any audio format
• Any movie format
• PostScript
• Portable Document Format (PDF)
• Any other file content type
• Subdirectories
• The parent directory
The command that associates an icon with a MIME content type is AddIconByType.
However, we will also specify the ALT text for text-based browsers with the analo-
gous AddAltByType command.

AddIconByType /icons/layout.gif text/html

AddIconByType /icons/text.gif text/plain
AddIconByType /icons/generic.gif text/*

AddIconByType /icons/image2.gif image/*

AddIconByType /icons/sound1.gif audio/*
AddIconByType /icons/movie.gif video/*

AddIconByType /icons/ps.gif application/postscript

AddIconByType /icons/pdf.gif application/pdf

DefaultIcon /icons/ball.gray.gif

AddAltByType "HTML file" text/html

AddAltByType "Plain text" text/plain
AddAltByType "Text" text/*

AddAltByType "Static image" image/*

AddAltByType "Audio" audio/*
AddAltByType "Video" video/*

AddAltByType "PostScript" application/postscript

AddAltByType "PDF" application/pdf
Figure 6-24. Specifying icons for MIME content types

Note: Normally I would recommend using the PNG icons rather than the GIF icons to
avoid possible future patent problems with Unisys. However, whoever did the conversion
got the background transparency wrong and you should use the GIF icons for the time
being until the PNGs are fixed.

We still have a problem with directories. There is no MIME content type for a di-
rectory so we must use other facilities. The following is a filthy hack introduced by
Apache version 1 and preserved into version 2.

43
Chapter 6. Handling directories

AddIcon /icons/dir.gif "^^DIRECTORY^^"

AddIcon /icons/back.gif ".."

AddAlt "Directory" "^^DIRECTORY^^"

AddAlt "Up" ".."
Figure 6-25. httpd.conf: Support for directories

There is one last change we should make after all this playing with icons. An icon
is an embedded image. It is generally regarded as a “good thing” to annotate the
<IMG/> tag with the height and width of the image in pixels. We can specify these
with the IndexOptions options IconWidth and IconHeight. These cause the web
server to include width="20" height="22" in the <IMG/> tag’s attributes. These
numbers are wired into Apache and correspond to the icons it ships with, all of which
are 20 22 pixels in size.

IndexOptions FancyIndexing ... IconWidth IconHeight

Figure 6-26. httpd.conf: Specifying icon sizes

Conclusion
And now our listings look a bit more colourful. But this is a lot of effort for limited
presentational value.

Figure 6-27. Listing with icons

Adding text to the listing

In addition to having a listing of files, it is possible to place text above and below
the listing. This can either be in the form of plain text or full-blown HTML. We will
concentrate on the latter.

44
 Chapter 6. Handling directories

To add HTML above the listing the configuration must identify a header file. This file
must have a name that identifies it as having MIME content type text/html. In the
simple case, however, the file’s content, should not be a full HTML document but just
the HTML body component (without the leading BODY tag) for the text to appear
above the listing. Everything else will be automatically generated. We identify this
file (should it exist) with the HeaderName command.

HeaderName HEADER.html
Figure 6-28. httpd.conf: Specifying a header file

<p>Here is an HTML fragment to appear above

the file listing.</p>
Figure 6-29. The HEADER.html file

Figure 6-30. The file listing with header file included

Note that the HEADER.html file appears in the listing too. Typically this is not wanted
as it is already “doing its job” by having its contents appear at the top of the page.
The file HEADER.html would be a good candidate for the IndexIgnore command.
Sometimes, you do want to modify the HTML headers provided by the web
server. For this purpose there is an option to the IndexOptions command:
SuppressHTMLPreamble. With this option in place, the HEADER.html file should be a
complete “top half” of an HTML document.

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

"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
<head>
<title>An index page</title>
45
Chapter 6. Handling directories

</head>
<body bgcolor="#ffffff" text="#000000"
link="#003399" alink="#cc0000" vlink="#cc3333">
<p>Here is an HTML half page to appear above
the file listing.</p>
Figure 6-31. HEADER.html with headers

IndexOptions FancyIndexing ... SuppressHTMLPreamble

HeaderName HEADER.html
Figure 6-32. httpd.conf: Providing our own HTML header

Figure 6-33. Links with different colours, courtesy of the <BODY> tag

In addition to placing text above the listing it is possible to place it underneath too.
A sensible name for this would be a “footer” to correspond with “header”. It’s called
a “readme”. Ho hum. The corresponding command is ReadmeName and this is re-
quired to be an HTML fragment which does not contain the </BODY> or </HTML>
tags.

Using an HTML table

We commented above that the data presented in the file listings is inherently tabular
and would be better presented as an HTML table. This is now available, as an “ex-
perimental feature” in versions of Apache beyond 2.0.23. (Red Hat Linux version 9
ships with 2.0.40). The author can find no mechanisms for setting the attributes of the
table from within Apache except to use stylesheets in the header file for the directory.
If you use HTML tables (and the author thinks it is a good idea) then you still need
to use the NameWidth=* and DescriptionWidth=* options to IndexOptions.

46
 Chapter 6. Handling directories

IndexOptions ... HTMLTable ...

Figure 6-34. httpd.conf: Using HTML tables for the index

Figure 6-35. The listing in HTML table format

Summary of the auto-indexing module

In the summary of the commands provided by autoindex_module given below only
the commands and options discussed in this course are covered. There are many
more. If you can’t get the result you want with the commands given to date then
consult the full Apache documentation. You might get lucky.

Note: Commands and options that only make sense if fancy indexing is turned on are
marked with an “(f)”.

Syntax summary: autoindex_module

IndexOptions
Sets various parameters for how the index should look. The list below gives the
options.

IndexIgnore "name" "name" ...

Takes a list of filenames or shell-style wildcarded filenames for file names. Files
whose names match one or more of the patterns are not listed in the index.

47
Chapter 6. Handling directories

AddIconByType icon mime_type (f)

Specifies the icon that should be used for a particlar MIME content type. The
MIME content type can either be fully specified (e.g. text/html) or partially spec-
ified (e.g. text/*).

AddAltByType "text" mime_type (f)

Specifies the ALT attribute in the <IMG/> tag. If you are expecting text-only
browsers you might want to keep this short and of constant width (three
characters is traditional). Alternatively, ditch the icons altogether.

DefaultIcon icon (f)

This specifies the icon to be used if nothing else matches. There does not appear
to be an equivalent DefaultAlt command.

AddIcon icon name (f)

This specifies an icon for a particualt file name. Typically this should be avoided
but it is the best way to match the parent directory .. and other directories with
the pseudo-filename ^^DIRECTORY^^.

AddAlt "text" name (f)

This specifies ALT text alongside AddIcon’s images.

HeaderName name name ...

This identifies the file whose contents should be placed above the file listing.
The first file in the list that exists is used. These file names typically appear in the
IndexIgnore instruction.
The files can be either plain text, an HTML body fragment or an entire “top half”
of an HTML page. To stop the server adding its own HTML top half see the
IndexOptions option SupporessHTMLHeader.

ReadmeName name name ...

Exactly as HeaderName but it corresponds to the text below the listing. This can
only be plain text or an HTML body fragment.

Syntax summary: Options to IndexOptions

FancyIndexing
Turns on the four-column (by default) indexing mode rather than plain, bullet-
list indexing mode.

ScanHTMLTitles (f)
The Description column is filled in with the HTML <TITLE>.

SuppressSize (f)
The Size column is not displayed.

SuppressLastModified (f)
The Last modified column is not displayed.

SuppressDescription (f)
The Description column is not displayed.

48
 Chapter 6. Handling directories

FoldersFirst
List the subdirectories before the plain files.

SuppressHTMLPreamble
Instructs the web server not to create the HTML preamble but to rely on the
header file (identified by the HeaderName command) to provide it instead.

NameWidth=width (f)
Specifies the width of the Name column. If the width is a number then it is that
many characters wide. If it is an asterisk then it is as wide as the longest name
(with a lower bound).

DescriptionWidth=width (f)
Specifies the width of the Description column. If the width is a number then it
is that many characters wide. If it is an asterisk then it is as wide as the longest
description (with a lower bound).

IconWidth (f)
If icons are being used then this specifies that the web server should include the
WIDTH attribute in the <IMG/> tag.
If it is used on its own then it inserts a width of 20 pixels. This is the width
of every icon shipped in the icons directory. Alternatively, it can be used as
IconWidth=N whereupon it will use N as the width. All icons must be of the
same width; Apache has no mechanism to analyse icons on the fly.

IconHeight (f)
If icons are being used then this specifies that the web server should include the
HEIGHT attribute in the <IMG/> tag.
If it is used on its own then it inserts a height of 22 pixels. This is the height
of every icon shipped in the icons directory. Alternatively, it can be used as
IconHeight=N whereupon it will use N as the height. All icons must be of the
same height; Apache has no mechanism to analyse icons on the fly.

HTMLTable (f)
This instructs Apache to use an HTML table rather than a <PRE> block to present
the listing.

Using both modules

What happens if we run both modules? The answer is “it depends on what order we
load them in”. If we load autoindex_module and then dir_module the latter is never
used. The reason for this is that Apache runs though its modules in order looking
for one that can handle the request. If a directory query is received then, even if the
directory contains an index.html file, autoindex_module can handle it and does so.
If we load dir_module first then the situation is far more useful. If there is
an index.html file then it is used. If there isn’t then the directory is indexed
automatically. But there’s a little more. If both are in play then the ScanHTMLTitles
option gives us a little more. If a subdirectory contains an index.html file then its
<TITLE> is used as a description for the directory in the parent directory’s automatic
index.

49
Chapter 6. Handling directories

LoadModule dir_module modules/mod_dir.so

DirectoryIndex index.html

LoadModule autoindex_module modules/mod_autoindex.so

IndexOptions FancyIndexing ScanHTMLTitles SuppressSize SuppressLastModified ...
Figure 6-36. httpd.conf: Running with both directory modules

Figure 6-37. A listing with a directory’s description

50
Chapter 7. Logging

Error log
We will examine the error log to see what is logged and to change the amount of
logging done.

log_config_module
We will load and use a module that allows us to configure exactly what we log for
each request.

Log file rotation

We will illustrate the Red Hat Linux system-wide mechanism for log rotation and
briefly mention, and then discard, an Apache-specific way to do the same thing.

Log file analysis

We will illustrate the use of the webalizer package for processing the log files to pro-
vide a set of web pages illustrating the use of the server.

Legalities
There will be a brief description of the legal implications of keeping log files. The
author is not a lawyer and everything will change as either Parliament implements
yet another piece of contradictory legislation or a senile fool of a high court judge
farts while setting precedent.

The error log

The log file /var/log/httpd/error_log contains the error reports and warnings
from the web server. By default, the directory /var/log/httpd is readable only by
root. You may want to change this on your system. Within the directory, the files are
created world-readable. Only the directory’s permissions need be changed.
Let’s consider a number of the typical entries in the error log as it currently stands.

[Mon Mar 24 10:23:11 2003] [notice] SIGHUP received. Attempting to restart

[Mon Mar 24 10:23:11 2003] [notice] Apache/2.0.40 (Red Hat Linux) configured
-- resuming normal operations
Figure 7-1. error_log: Reloads

Our first example will be seen in the log files from this course more than any other
lines (we hope!). The line that starts “SIGHUP received” is the logged entry that
means we requested a reload of the configuration file. A SIGHUP is an operating sys-
tem signal sent to a running process (the web server) instructing it to do something.
In our case it is to reread the configuration file.
The line that (hopefully) follows it is the line from Apache that says it has been
(re)configured and that it is “resuming normal operations”, i.e. serving web pages
again.

51
Chapter 7. Logging

[Mon Mar 24 09:50:58 2003] [notice] caught SIGTERM, shutting down

Figure 7-2. error_log: Shutting down

Analogous to SIGHUP is SIGTERM which is another of these operating system sig-

nals, but one with a universal function. It is the instruction to a process to shut down.

[Wed Mar 26 09:25:50 2003] [error] [client 131.111.10.87] File does

not exist: /var/www/CHEESE/nonesuch.html
Figure 7-3. error_log: File not found

Another commonly occuring error_log line is the one corresponding to a request

for a file that does not exist.
So far, the error_log lines we have looked at have had a standard format:

• [date]

• [severity]

• message

Table 7-1. Logging levels

Level Meaning Example

emerg System is unusable. The “Child cannot open lock file.
entire web service has Exiting”
failed.
alert Action must be taken “getpwuid: couldn’t determine
immediately. The entire user name from uid”
web service has either
failed already or is
imminent danger of failing
completely.
crit Critical condition. A child “socket: Failed to get a socket,
server has failed and exiting child”
others are likely to follow.
error A request on a single “File does not exist:
request. /var/www/CHEESE/nonesuch.html”
warn Warnings. “child process 10108 still did
not exit, sending a SIGTERM”
notice Normal behaviour but “Apache/2.0.40 (Red Hat Linux)
worthy of notice. configured -- resuming normal
operations”
info Purely informational “Server seems busy, (you may need
to increase StartServers, or
Min/MaxSpareServers)...”
debug Debugging. “Opening config file ...”

We can change the level of the logging (of formatted messages) with the LogLevel
command. Either globally, or within specific virtual hosts’ sections we can issue the
command LogLevel debug, say, to get more debugging.
Messages issued from a running web server are well formatted. However, if you
make a syntax error in the httpd.conf file then the server won’t launch and the
error message is rather more stark.
52
 Chapter 7. Logging

Syntax error on line 10 of /etc/httpd/conf/httpd.conf:

Invalid directory indexing option
Figure 7-4. error_log: Unformatted error messages

It is also possible to move the error log file, or to do without the file altogether (but
still log errors).

ErrorLog /var/httpd/logs/error.log
Figure 7-5. httpd.conf: Logging to a different file

The ErrorLog directive gives the name of the error log file (relative to the
server root, /etc/httpd if it not given as an absolute path). By default the log
file is specified as logs/error_log. You will recall that /etc/httpd/logs
is a symbolic link to /var/log/httpd so error logs are stored, by default, in
/var/log/httpd/error_log. Any filename given to this command that does not
start with a / will be resolved relative to the server root.
If the file name given is “syslog” then logging is not done to /etc/httpd/syslog
but rather all error logs are passed to the operating system’s system logger. This can
be useful if you arrange for your system logs to be transmitted off-machine to a cen-
tral logging engine which you want to receive Apache error logs too.
Finally, if the file name starts with a pipe character, |, then what follows is interpreted
as a command which should be forked and executed and which has the error log lines
passed to it on its standard input.

Syntax summary: Error logging commands

ErrorLog logfile
If logfile begins with “/” then this file will be used for the error log.
If logfile is “syslog” then the error logs will be passed to the system logger
with facility local7.
If logfile is “syslog:facility ” then the error logs will be passed to the sys-
tem logger with facility facility. This must be one of the facilities known to
the system logger; you can’t just make up your own.
If logfile begins with “|” then what follows will be run as a command to re-
ceive the log lines on standard input.
Anything else is interpreted as a filename relative to the server root (/etc/httpd
by default).

LogLevel level
Any errors generated at logging levels equal to or more serious than level will
be logged.

Access logs
To date the only log file we have met is the error log. There is no logging in our
current server configuration when things aren’t going wrong. Clearly we want to log
the requests that are made of our server. These are the access logs.
We need to decide what we want to log and where to log it to. We may want more
than one log file for different sets of data.
As (almost) ever, the means to get this functionality is to load a module:
log_config_module from mod_log_config.so.
53
Chapter 7. Logging

This provides us with one particularly useful command: CustomLog. This allows us
to specify what information to record and where to record it for each query/response
at the server. This power comes at the price of almost complete syntactic obscurity at
first glance. But in all honesty it’s not that bad.
Suppose we wanted to record just the following information about each query pro-
cessed by the server:

• Time & date

• URL requested
• Name of system making the request
• Whether the request was successful

LoadModule log_config_module modules/mod_log_config.so

CustomLog logs/request.log "%t %U %h %s"

Figure 7-6. httpd.conf: Basic logging of requests

Each of the elements beginning with a percentage character is called an escape code
and is converted into some piece of logged information. A complete list of the codes
is given in Appendix B.

Four escape sequences

%t
The time of the request

%U
The URL requested

%h
The client hostname

%s
Status code of the request
To illustrate what they indicate and what they don’t, we will request three URLs and
note a number of problems in the logged output.

The requested URLs

• http://cheese.dept.cam.ac.uk/

• http://chalk.dept.cam.ac.uk/

• http://cheese.dept.cam.ac.uk/gamma.html

[08/May/2003:10:26:02 +0100] / 131.111.11.148 200

[08/May/2003:10:26:10 +0100] /index.html 131.111.11.148 200
[08/May/2003:10:26:29 +0100] /gamma.html 131.111.11.148 200
Figure 7-7. request.log: The corresponding output

Problems with the output as it stands

54
 Chapter 7. Logging

• There is no record of whether chalk.dept.cam.ac.uk or cheese.dept.cam.ac.uk was

queried.
• The hostnames are addresses, not names
The simplest way to address the issue of which website was queried is to move the
CustomLog lines into the virtual host sections and to have two different log files.
This gives them the flexibility to log different things too.
If we really wanted a single log file with the virtual host information we could use
the %v escape code to record it.

<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
CustomLog logs/cheese.log "%t %U %h %s"
</VirtualHost>
Figure 7-8. Setting a log file for cheese.dept.cam.ac.uk

To enable the use of hostnames rather than addresses, we must instruct the web
server to do DNS lookups for the IP addresses on each incoming query. We will do
this with the HostnameLookups command. This command is a core Apache com-
mand and not part of the logging module. It is also required if you plan to do any
access controls based on host names as we will be in the Section called Access control
by client IP address in Chapter 10. We will set this on globally. If either website wanted
to record IP addresses rather than hostnames then it can do so by using %s rather
than %h.

HostnameLookups On
Figure 7-9. Turning on hostname lookups

[26/Mar/2003:14:09:54 +0000] / noether.csi.cam.ac.uk 200

[26/Mar/2003:14:09:54 +0000] /icons/dir.gif noether.csi.cam.ac.uk 304
[26/Mar/2003:14:09:55 +0000] /icons/layout.gif noether.csi.cam.ac.uk 304
[26/Mar/2003:14:09:57 +0000] /alpha.html noether.csi.cam.ac.uk 200
[26/Mar/2003:14:10:00 +0000] /beta.html noether.csi.cam.ac.uk 200
Figure 7-10. The cheese.dept.cam.ac.uk log file

Common Log Format

A standard format for access log files is used by many utilities and replied on by the
web analysis programs. It is called “The Common Log Format” (CLF) and is shown
below.

CustomLog logs/chalk_log "%h %l %u %t \"%r\" %>s %b"

Figure 7-11. httpd.conf: Specifying the Common Log Format for
chalk.dept.cam.ac.uk

noether.csi.cam.ac.uk - - [26/Mar/2003:14:25:49 +0000]

"GET /index.html HTTP/1.1" 304 0
noether.csi.cam.ac.uk - - [26/Mar/2003:14:26:00 +0000]
"GET /nonesuch.html HTTP/1.1" 404 211
Figure 7-12. The Common Log Format for chalk.dept.cam.ac.uk

55
Chapter 7. Logging

The escape sequences used in the Common Log Format

%h
The client’s hostname

%l
If the web server was doing IDENT (RFCnnnn) lookups then the returned userid
would be here.

%u
If the client had authenticated as a particular user for this request the userid
would be recorded here. We discuss authentication in detail in the Section called
Access control by client identity in Chapter 10.

%t
The time of the request.

%r
The first line of the query.

%>s
The status code finally returned to the client. There is a subtle difference be-
tween %s and %>s. In most cases they are identical. In cases where the URL gets
remapped then %s gives the status code of the initial lookup and %>s the code
of the final lookup (and the code passed back to the client).

%b
The number of data bytes (i.e. excluding the headers) sent back to the client in
the case of successful completion..

Named formats
A common requirement is for all virtual hosts to log in the same format. To assist
with this it is possible to name a format definition and to then refer to the format’s
name in the CustomLog line.

LogFormat "%h %l %u %t \"%r\" %>s %b" clf

<VirtualHost *>
ServerName chalk.dept.cam.ac.uk
DocumentRoot /var/www/CHALK
CustomLog logs/chalk_log clf
</VirtualHost>

<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
CustomLog logs/cheese_log clf
</VirtualHost>
Figure 7-13. Using named log formats with virtual hosts

Logging headers
One very useful escape code is %{fubar}i which will log the value of incoming
header fubar. We could use this as %{Host}i to record the queried Host header, for
example, to check our virtual hosting was working as expected.

56
 Chapter 7. Logging

Finding bad links

Most browsers also include a query header Referer: which contains the URL of the
page that contain the link the user queried to bring them here. This can be particularly
useful to find remote pages that have bad links onto your site.

CustomLog "%{Referer}i %U %>s" referer_log

Figure 7-14. Logging referers

Those pages which have bad links pointing into your site will generate status
code 404 and can be tracked down.

Log rotation
It is one thing to create logs; it is quite another to cope with them. A log file grows
without bound unless action is taken and this can cause problems.

Problems with growing log files

• Larger files are harder to manipulate.
• File systems run out of space.
• The information you log may constitute personal data.
A solution to this generic problem of log file growth is log rotation. This involves the
regular (nightly or weekly, typically) moving of an existing log file to some other file
name and starting fresh with an empty log file. After a period the old log files get
thrown away.
Because this is a general issue, Red Hat Linux provides a general solution that can be
applied to any set of log files, not just the web server’s. There is an Apache-specific
solution (which is provided by the rotatelogs command) but we will use Red Hat
Linux’s generic solution, provided in the logrotate package.
Once each night the logrotate program reads in its configuration files telling it which
logs to rotate and how to do it. One of these instructions tells it to rotate any file in
/var/log/httpd whose name ends with log.

# rotate log files weekly

weekly

# keep 4 weeks worth of backlogs

rotate 4

# create new (empty) log files after rotating old ones

create

# RPM packages drop log rotation information into this directory

include /etc/logrotate.d
Figure 7-15. /etc/logrotate.conf: The main log rotation file

The main configuration file sets up the defaults and then reads in a directory of in-
structions for specific sets of log files from the /etc/logrotate.d directory.

/etc/logrotate.conf: commands

weekly
Each file should be rotated weekly. The log rotation job runs nightly, though, so
this can be changed to daily for a specific log file if desired.

57
Chapter 7. Logging

The three commands that specify how often rotation should take place are daily,
weekly and monthly.

rotate 4
Keep four sets of log files. The comment is slightly inaccurate; four weeks’ worth
of logs will be kept if rotation is done weekly. If rotation is done daily then this
command means that four days’ worth of logs are kept.

create
After moving the main log file from logfile to logfile.1 a new, empty
logfile should be created.

include /etc/logrotate.d
This command instructs the log rotation program to read in every file in this
directory. One of these files will correspond to the web server’s log files.

/var/log/httpd/*log {
missingok
notifempty
sharedscripts
postrotate
/bin/kill -HUP ‘cat /var/run/httpd.pid 2>/dev/null‘ 2> /dev/null || true
endscript
}
Figure 7-16. /etc/logrotate.d/httpd

The /etc/logrotate.d/httpd file (part of the httpd package), not the logrotate
package, contains the instructions specific to the web server logs.

/etc/logrotate.d/httpd: commands

/var/log/httpd/*log { ... }
This specifies that the commands within the curly brackets are to be applied to
all the files that match the expression /var/log/httpd/*log.

missingok
This is the instruction not to return an error if a particular log file is not present.

notifempty
This command instructs the system not to rotate the logs if the current main log
file is empty. See the discussion below about whether this is a good idea or not.

sharedscripts
Further down we will see a shell script that will be run after the rotatiopn has
happened. This command instructs the system to run that script only once after
all the files matching /var/log/httpd/*log have been rotated and not after
each individual rotation.

postrotate ... endscript

Following the rotation of a log file (or all of them if sharedscripts is invoked)
the commands between postrotate and endscript are run. This rotation program
runs as root so care must be taken with the commands that appear here.

58
 Chapter 7. Logging

/bin/kill -HUP ‘cat /var/run/httpd.pid 2>/dev/null‘ 2> /dev/null || true

This messy command sends a SIGHUP signal to the master web server demon.
It is equivalent to /etc/init.d/httpd reload except that it sends the SIGHUP to all
the web server processes.
There are two points which must be made about log rotation and changing its set-
tings. These are to do with the presence of editor backup files and the Data Protection
Act (1998).

Backup files
The command include /etc/logrotate.d will read in every file in that directory. So if
you edit the file httpd and leave behind both httpd and httpd~ then both these files
will be included and your log files will have the log rotation process applied twice.
Now, because of the weekly (or monthly or daily) commands the rotation shouldn’t
actually happen but it is still not certain that the right file will be applied.

Data Protection Act (1998)

The contents of the log files may constitute personal data. If, along
with any information stored within the University, they identify individu-
als then they definitely do. As a result, you must have a privacy policy,
and stick to it. For this reason, the author recommends very strongly
that you remove the notifempty command. If your policy says that per-
sonal data is only kept for 28 days and it is kept for a week longer
because one week’s log files happened to be empty then you are in
breach of your privacy policy.

Log file analysis

Red Hat Linux ships with the webalyzer which is a log file analysis package that
generates a graphical representation of the use of your web server. This can be par-
ticularly useful for convincing heads of department that web servers are actually
used.
This course focuses on the use of the Apache web server so we will not look in depth
at tuning the configuration to vary the format of the output. Instead we will simply
focus on what it does and how to configure it to work with multiple virtual hosts.
The configuration file is /etc/webalizer.conf and is set up to match the Red Hat
default file locations. As we have changed them we have to change this file too.
If you are using the default Red Hat settings then skip this section and look in
http://server/usage/ for the usage statistics.
There are two approaches to changing this configuration for a multiple-site server.
The first is to give each web site a copy of its own configuration file. This lets each
site have full control over the presentation of the usage figures. The second approach
is to have a common configuration file that lacks the per-site information and has
the missing information provided on the command line. This lets each site rely on a
centrally maintained, common presentation format.
In this course we will take the former approach for no better reason than that it’s
slightly easier to teach. This is a course about running Apache, not Webalizer.
Webalizer maintains a directory of pages and images (of usage graphs) to present
the statistics. In the default Red Hat Linux installation this is maintained in
/var/www/html/usage. We will, therefore, create /var/www/CHALK/usage and
/var/www/CHEESE/usage.

59
Chapter 7. Logging

# mkdir /var/www/CHALK/usage
# mkdir /var/www/CHEESE/usage
# chmod g+ws /var/www/CHALK/usage
# chmod g+ws /var/www/CHEESE/usage
Figure 7-17. Creating the usage directories

To keep things under the editorial control of the web site managers we will copy the
webalizer configuration files into /var/lib/webalizer and change the group and
permissions to give them control.

# cp /etc/webalizer.conf /var/lib/webalizer/chalk.conf
# cp /etc/webalizer.conf /var/lib/webalizer/cheese.conf
# chgrp chalk /var/lib/webalizer/chalk.conf
# chgrp cheese /var/lib/webalizer/cheese.conf
# chmod g+w /var/lib/webalizer/chalk.conf
# chmod g+w /var/lib/webalizer/cheese.conf
Figure 7-18. Copying the configuration files

Next, we are going to edit half a dozen items in each of the configuration files.

Parameters to change in the webalizer configuration file

LogFile /var/log/httpd/cheese_log
This specifies the log file to read the raw data from.

LogType clf
This identifies the format of the log file as common log format.

OutputDir /var/www/CHEESE/usage
This identifies the directory in which the statistics are to be compiled for presen-
tation.

HistoryName /var/lib/webalizer/cheese.hist
This specifies a “working file” the system uses. By default it will appear in the
output directory, but we don’t want it in a downloadable location.

IncrementalName /var/lib/webalizer/cheese.current
This is another “working file” the system needs.

HostName cheese.dept.cam.ac.uk
This identifies the host name used to identify the site. For a virtual host it needs
to be set.
Finally, all we need to do is to make sure that the log analysis program gets run
over each of the web sites. It is run out of a daily cron job controlled by the file
/etc/cron.daily/00webalizer.

#! /bin/bash
# update access statistics for the web site

if [ -s /var/log/httpd/access_log ] ; then
/usr/bin/webalizer
fi

exit 0
Figure 7-19. The initial /etc/cron.daily/00webalizer file

60
 Chapter 7. Logging

We need to run it multiple times for our various web sites using its command line
option to select non-standard configuration files.

#!/bin/bash

for conf in /var/lib/webalizer/*.conf

do
/usr/bin/webalizer -c ${conf}
done

exit 0
Figure 7-20. Modified cron script

61
Chapter 7. Logging

62
Chapter 8. Users’ own web pages

userdir_module
We will introduce the relevant module and the single command it provides.

Simple use
We will start with the simplest provision of users’ personal web pages by loading the
module and using it in its simplest form.

Complex use
We will then give an example of how it can be used to redirect lookups to an entirely
different system.

The principle of this chapter is to provide your users with the ability to create their
own web pages. The web pages may be located on the servers in quesion, or on a
different server altogether. As ever, there is a module that provides the extra func-
tionality. In the example below, we provides user pages in all the virtual hosts.

LoadModule userdir_module modules/mod_userdir.so

UserDir public_html
Figure 8-1. Loading and using the module

In this example, the command UserDir public_html causes any URL of

the form http://server/~bob/path/name.html to correspond to a file
~bob/public_html/path/name.html. (In this filename the expression “~bob”
means “the home directory of user bob” and is standard Unix notation. It is this that
the URL notation is based on.
The directory name public_html is not fixed and can be modified. Furthermore,
more complex mappings of username onto file name can be provided. Any instance
of “*” in the “directory name” will be replaced with the user’s ID.
If the directory name is actually a URL then, instead of the web server looking for
a local directory, it responds with an HTTP redirection, pointing the web client at a
new URL, typically on a different server.

Table 8-1. userdir_module lookups of http://server/~bob/alpha/beta.html

UserDir argument Translated path

public_html ~bob/public_html/alpha/beta.html
www ~bob/www/alpha/beta.html
/var/www/users /var/www/users/bob/alpha/beta.html
/var/www/*/web /var/www/bob/web/alpha/beta.html
http://elsewhere/users http://elsewhere/users/bob/alpha/beta.html

http://elsewhere/*/web http://elsewhere/bob/web/alpha/beta.html

http://elsewhere/~*/ http://elsewhere/~bob/alpha/beta.html

It is possible to give a sequence of targets to the UserDir command. In this case they
will be searched in turn until one provides the server with the file or directory it is

63
Chapter 8. Users’ own web pages

looking for. Only the last entry in the list is allowed to be a redirection to another
server (i.e. a URL) because when the server reaches this one it sends back the redirec-
tion to the browser and never gets to discover if the file existed at the far end.

64
Chapter 9. Delegated control

<Directory>
Applying a specialised set of commands just to a subdirectory of a web site from the
httpd.conf file.

Include
Splitting the main configuration file into components.

AccessFileName
Nominating a filename to handle control from the directory itself.

AddDescription
A commonly used command in the delegated configuration files to set the Descrip-
tion column in automatic indexes.

AllowOverride
Controlling what can be delegated

The user directory example was the first where we were passing control outside our
tidy document root. It may well be that we want a different configuration for these,
relatively uncontrolled, areas.
There are a number of aspects to this. You must decide what defaults you want
passed to these areas and what you want absolutely fixed. You also need to know
how to override the defaults where permitted.
We will start by noting how to change settings from within the httpd.conf file for
a directory tree. In our current configuration file the directory index file name is
index.html. Suppose for a subdirectory of one of our web sites we wanted to change
it to be main.html. How would we do that?

<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
CustomLog logs/cheese_log clr

<Directory /var/www/CHEESE/bestiary>
DirectoryIndex main.html index.html
</Directory>

</VirtualHost>
Figure 9-1. httpd.conf: Using the <Directory> command

The <Directory dir > ... </Directory> identifies a series of commands which
should override or enhance the general settings for a specific subdirectory,
/var/www/CHEESE/bestiary in the example given in the figure.
In the case of commands we have met, it is easy to imagine simply issuing them again
within a <Directory> block to override the previous settings. But what about turning
features on or off? A common example is to turn on or off the automatic generation
of indexes.
At the moment we can see the index of the games directory in the
cheese.dept.cam.ac.uk web site.

65
Chapter 9. Delegated control

Figure 9-2. The games directory

We can turn indexing off with the Options command as follows.

<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
CustomLog logs/cheese_log clr

<Directory /var/www/CHEESE/games>
Options -Indexes
</Directory>
...
</VirtualHost>
Figure 9-3. httpd.conf: Turning off indexes

And any future attempt to index games gives a 403, Forbidden, error.

66
 Chapter 9. Delegated control

Figure 9-4. Not the games directory

The Options command sets various parameters that basically control whether a mod-
ule’s or core facilities should be enabled or not. For the topic of delegation we will
be interested in only three parameters. The first of these is Indexes which enables
or disables autoindex_module. Note the leading minus sign. The Options command
can be used in three ways, to disable a specific option, to enable a specific option or
to set the complete set of options in one go.

Syntax summary: Setting options

Options -Indexes
Turn off automatic indexing. Leave all other options unchanged.

Options +Indexes
Turn on automatic indexing. Leave all other options unchanged.

Options Indexes
Turn on automatic indexing. Unset all other options.
So far, we haven’t really delegated control. We have allowed for variation in subdi-
rectories but we have not truly delegated the controls to people who cannot rewrite
the configuration file and tell the server to reread it. We need a means to delegate
control of a subdirectory into the subdirectory itself.

IndexIgnore "#*#" "*~" "configuration"

AccessFileName "configuration"
Figure 9-5. httpd.conf: Delegation to files in directories

The AccessFileName command names a file (or set of files) that will be looked for
within the directory being served and whose contents will be regarded as if they had
been inside a <Directory> block for that specific directory. The name of the command

67
Chapter 9. Delegated control

tells of its origins; it was used to set the access rights for a directory tree. It is, however,
a fully generic delegated configuration, not just delegated access control.
The default file name used, .htaccess, also reflects its history as a delegated access
control mechanism. It is also traditionally a “dot file” to hide it from the index listings.
It’s far better to list the file name in a IndexIgnore statement and to give it a plain file
name so the conventional Unix utilities will actually show you it’s there.
So we could copy the contents of the <Directory /var/www/CHEESE/games>
block to /var/www/CHEESE/games/configuration and the
contents of the <Directory /var/www/CHEESE/bestiary> block to
/var/www/CHEESE/bestiary/configuration.

IndexIgnore "#*#" "*~" "configuration"

AccessFileName configuration
<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
CustomLog logs/cheese_log clr
</VirtualHost>
Figure 9-6. httpd.conf: Delegated control

Options -Indexes
Figure 9-7. /var/www/CHEESE/games/configuration

DirectoryIndex main.html index.html

Figure 9-8. /var/www/CHEESE/bestiary/configuration

This puts the control of the files in the hands of the people who have access to the
directories.
However, there may be certain properties that you don’t want the users messing with.
To this end there is limited support for restricting what the users can override with
their delegated configurations. This is controlled via the AllowOverride command.
This rather unsatisfactory command allows the controllers of httpd.conf to stop the
Options command being used in the AccessFileName files, but not to specify which
options can and can’t be set. It can specify what you can do with IndexOptions but
not whether or not you can enable/disable indexes at all. It has many limitations.

Simple uses of AllowOverride

AllowOverride None
The delegated configuration files aren’t even read. Their content is entirely ig-
nored.

AllowOverride All
Any command that is syntactically legal in the delegated configuration file is
allowed to have effect.

AllowOverride Options
The delegated configuration file is allowed to run the Options command. There
is no mechanism to control which of its arguments are permitted.

68
 Chapter 9. Delegated control

AllowOverride Indexes
The delegated configuration file is allowed to run the IndexOptions command
and all the commands that modify the index. This does not permit the use of
Options [+|-]Indexes; you need AllowOverride Options for that.

69
Chapter 9. Delegated control

70
Chapter 10. Access control

Two ways
There are two ways to do access control: by the location of the client and by the
identity of the user operating the client.

Client location
There is a brief discussion of why this mechanism is fraught with difficulties caused
by proxies and the like. Then the commands to implement it are covered.

User identity
There is a discussion of the Basic and Digest protocols for user identification. Access
by user or group and user administration is then covered.

Mixed working
The mixed case of authorising passwordless access from within the institution but
requiring authentication from outside will be given in detail.

Blocked names
It is also possible to block a file from being downloaded at all based on its name.

Now we move to the topic of access control. There are fundamentally two ways of
doing this: by client location and client identity.
Client location involves specifying whether access is permitted based on the IP ad-
dress or hostname of the client (i.e. browsing) system. When a request is received by
the server the IP address of the client, browsing system is known. This address, or the
hostname associated with it, is checked against a set of rules to determine whether or
not the request should be honoured.

Proxy servers
Client location security is often used within the University for restricting
access to an institution or to the University, loosely defined as “any-
thing in cam.ac.uk”. This approach doesn’t work but is often regarded
as “good enough” to keep happy the politicians, lawyers and other peo-
ple who don’t understand technology. From the point of view of the
web administrator it also has the advantage of simplicity. The reason
it doesn’t work is that web proxies can forward a request from outside
Cambridge on to a server within Cambridge which sees the request
coming from within Cabridge and honours it. The Computing Service
has had its internal minutes cached on Google for the whole world to
read after a web proxy on the CS staff network went unnoticed.

Client identity involves challenging the user to quote some means of identifying him
or herself before permitting access to the document requested. This has the advan-
tage of dealing with proxies, but the disadvantage of requiring administration of the
userids and passwords. A common compromise is to create a single userid and pass-
word for a set of pages and pass the pair on to anyone who needs access. This has the
disadvantage that you don’t know which of your users read the pages, but often you
don’t want to know.

71
Chapter 10. Access control

Access control by client IP address

As ever, this functionality is provided by a module: access_module from library
mod_access.so.

LoadModule access_module modules/mod_access.so

Figure 10-1. Loading access_module

This module offers us three commands: Allow, Deny and Order.

These restrictions need not cover the whole server, and typically don’t. They can be
placed within a <Directory> block or delegated configuration file to restrict access
to only a subtree of the server. If you want to cover the whole web server then the
document root must be used for the <Directory> block.
The Order command takes one of two arguments: Deny,Allow and Allow,Deny. No
whitespace is allowed around the comma. While it may look like a comma-delimited
list it is not; it is just a pair of rather strange looking arguments that have a comma as
one of their characters.
If the argument is Deny,Allow (the default) then the initial state is that all access is
allowed, then all the Deny statements are processed and then they are overridden by
the Allow statements.
If the argument is Allow,Deny then then the initial state is that all access is prohibited,
then all the Allow statements are processed and then they are overridden by the
Deny,Allow statements. This is best illustrated with some examples.

<Directory /var/www/CHALK>
Order Allow,Deny
Deny from csx.cam.ac.uk
Allow from cam.ac.uk
Deny from csi.cam.ac.uk
</Directory>
Figure 10-2. A Allow,Deny example

Table 10-1. The processing of a request by client gauss.csi.cam.ac.uk

Stage Match? State

Initial All requests refused.
Allow from cam.ac.uk Rule matches. Access is allowed.
Deny from csx.cam.ac.uk Rules does not match. No change in state.
Deny from csi.cam.ac.uk Rule matches. Access is denied.
Final Access is denied.

The addresses given in the Allow and Deny statements can be specified in a variety
of ways. The examples given are for the Allow command but are equally applicable
to the Deny command.

Syntax summary: Options on the Allow command

Allow from cam.ac.uk
Access is allowed from any host whose name ends with cam.ac.uk.

Allow from 131.111.11.148

Access is allowed for queries originating from 131.111.11.148. Note that any

72
 Chapter 10. Access control

query redirected through a web proxy or cache will have the address of the web
proxy or cache.

Allow from 131.111

Access is allowed for queries originating from IP addresses whose first two bytes
are 131.111. Note that Cambridge has more networks than just this primary one.

Allow from 131.111.10.0/255.255.254

Access is allowed from any IP address which when masked by 255.255.254.0
gives 131.111.10.0.

Allow from 131.111.10.0/23

Access is allowed from any IP address whose first 23 bits form the address
131.111.10.0.
If you want to pass control of this to the delegated configuration files then you must
pass the Limit option to the AllowOverride command. This command must also
appear in a <Directory> command.

LoadModule access_module modules/mod_access.so

<Directory /var/www/CHALK>
AllowOverride Limit
</Directory>
Figure 10-3. Allowing access control to be delegated.

Access control by client identity

The alternative mechanism for restricting access to web pages is to demand a userid
and password from the user.

HTTP and userid/password

1. Browser sends request for a web page.

2. Server sends back a 401 error code and specifies a realm.
3. Browser prompts user for userid and password for the realm.
4. User quotes userid and password.
5. Browser repeats the initial request with an extra header quoting the userid and pass-
word.
6. Server sends the page if the userid/password are OK.
7. Browser sends request for another web page.
8. Server sends back a 401 error code and specifies the same realm.
9. Browser recovers the userid and password it has for that realm and repeats the
initial request with the extra header.
10. Server sends the page.

Of course things are different if the userid and password don’t grant access to the
page. There are two ways this can happen. The user and password could match but
that user, now identified, might not be allowed access to the page. In this case the
server sends back a 403, Forbidden, error code. Alternatively, the userid and pass-
word might not match, in which case the server sends back the 401 code again and
the cycle of prompting the user repeats.

73
Chapter 10. Access control

What we need to know is how to set up the server so that userids and passwords are
known to the server and certain pages are flagged as requiring user authentication.

LoadModule auth_module modules/mod_auth.so

Figure 10-4. httpd.conf: Loading the user authentication module

To start with, we will need a module: auth_module. We will then specify a mecha-
nism to identify users and finally specify policies regarding which identified users
are allowed access.
So, first we need to identify users. This comes in two parts: the first involves setting
up userids and passwords at the server end and the second involves telling the web
server to use these for identifying users.
The userids and passwords are not the same as the login IDs. Indeed, they will often not
be login IDs at all. They are maintained with a distinct file which we will need tools
to manipulate. This file is traditionally called htpasswd though we have flexibility
regarding its name and location. A server administrator must also decide whether
to have a single password file for the whole server or one per virtual host (or even
for each subtree of the virtual host he wants to restrict access to). Granting a user a
userid and password noes not autmatically assign that userid rights to access pages
(though we can configure policy so that it does). In this example, we will work with
a single userid/password file for both virtual hosts. It’s a shortcoming of the Unix
permissions model that we cannot specify that a file should be writable by members
of either one group or another. We will use a webadmin group to control access to
this file. Note that the file should not be servable by the web server.

# groupadd -r webadmin
# usermod -G chalk,cheese,webadmin rjd4
# mkdir /etc/httpd/access
# chgrp webadmin /etc/httpd/access
# chmod g+ws /etc/httpd/access
# ls -ld /etc/httpd/access
drwxrwsr-x 2 root webadmin 4096 Apr 14 11:26 /etc/httpd/access
# touch /etc/httpd/access/passwd
# chmod g+w /etc/httpd/access/passwd
# ls -l /etc/httpd/access/passwd
-rw-rw-r-- 1 root webadmin 0 Apr 14 11:39 /etc/httpd/access/passwd
Figure 10-5. Creating a userid/password file and adding users

We make the directory writable rather than just the individual files to make life easier
for programs that move files about within directories for backing up.

$ htpasswd -m /etc/httpd/access/passwd rjd4

New password: password
Re-type new password: password
Adding password for user rjd4
$ cat /etc/httpd/access/passwd
rjd4:$apr1$kEDyP/..$n0DCjezTD.T.C.1s3td6..
Figure 10-6. Setting up users in the password file

htpasswd’s -m option causes the password file to use an MD5 password encod-
ing for the password. This is better than the traditional (and default) crypt algo-
rithm. This makes the password much harder to reverse engineer from the file but
all userid/password schemes are vulnerable to disctionary attacks and it is impor-
tant that the password file not be downloaded to make this attack much harder.

74
 Chapter 10. Access control

Now that we have a way to identify users we need to specify policies. As with ac-
cess_module the restrictions on access can only be specified in a <Directory> block
or in a delegated configuration file.
The simplest policy, called “valid user” is to permit access to any user who can au-
thenticate against the web password file.

<Directory /var/www/CHALK>
AuthType Basic
AuthName "Restricted area"
AuthUserFile /etc/httpd/access/passwd
Require valid-user
</Directory>
Figure 10-7. httpd.conf: Implementing the “valid user” policy

Syntax summary: implementing the “valid user” policy

<Directory /var/www/CHALK>...</Directory>
This is the standard block for restricting a set of commands to a directory tree.
The commands in this block could appear in a delegated configuration file.

AuthType Basic
This defines the protocol used for the exchange of userid and password. Every
browser supports this protocol, but it does send passwords in plain text. A supe-
rior protocol, called “Digest” exists and is supported by modern browsers. See
the Section called Variations on a theme of user identification for details.

AuthName "Restricted area"

This identifies the realm applying to the files in the directory tree. This string
appears in the challenge for the userid and the password and is used by the
browser to work out which previously given userid and password it should send
without having to prompt the user again.

AuthUserFile /etc/httpd/access/passwd
This identifies the file used to contain userids and passwords. This cannot be the
system /etc/passwd file!

Require valid-user
This specifies the policy. Any user validated against the password file may access
the pages.
Given this setup (and a reload of the server’s configuration file) we can see the effect
it has on our web server. Our attempt to access the index.html page results in a
challenge for userid and password.

75
Chapter 10. Access control

Figure 10-8. The userid and password challenge

Note that the prompt contains the phrase “Restricted area”. That text comes directly
from the AuthName command. If we fill in any valid userid and password from the
/etc/httpd/access/passwd file we can proceed.

Figure 10-9. Entering the userid and password

Next we will consider other policies. We will assume that we have created three ad-
ditional web userids: tom, dick and harry.

<Directory /var/www/CHEESE/games>
AuthType Basic
AuthName "Cheese lovers only"
AuthUserFile /etc/httpd/access/passwd
Require user tom dick
</Directory>
Figure 10-10. httpd.conf: Restricting access to /var/www/CHEESE/games to users
tom and dick.

The Require user tom dick statement replaces the “valid user” policy with a “one of
these users” policy.
If you plan to use certain collections of users repeatedly for access control this scheme
can be taken further and groups of users can be defined. We can then specify that the
validated user be one of a series of groups.
First we must define our groups. We will create a groups file this time by hand be-
cause there are no tools analogous to htpasswd to manage the files for us.

stilton: tom rjd4

cheddar: tom dick
Figure 10-11. The /etc/httpd/access/group file

We can change from a user list to a group list by specifying which group file to use
and which groups are permitted access.

76
 Chapter 10. Access control

<Directory /var/www/CHEESE/games>
AuthType Basic
AuthName "Cheese lovers only"
AuthUserFile /etc/httpd/access/passwd
AuthGroupFile /etc/httpd/access/group
Require group stilton cheddar
</Directory>
Figure 10-12. httpd.conf: Restricting access to /var/www/CHEESE/games to groups
cheddar and stilton

Syntax summary: Require

Require
The Require command specifies the policy of who is allowed access once identi-
fication is complete.

Require valid-user
Any authenticated user may have access to the pages.

Require user user1 user2 user3 ...

Only one of the listed users may have access to the pages.

Require group group1 group2 group3 ...

Any user in one or more of the listed groups may have access to the pages.
If we wanted to delegate policy regarding access control by this mechanism we must
allow the override with AllowOverride AuthConfig.

Variations on a theme of user identification

What we described in the previous section is a way to provide user authenticated
access control. We used the Basic protocol and simple text files to store the userids,
passwords and groups.
The Basic protocol can be replaced with the Digest protocol. This comes from module
auth_digest_module from mod_auth_digest.so.

LoadModule auth_digest_module modules/mod_auth_digest.so

<Directory /var/www/CHEESE/games>
AuthType Digest
AuthName "Cheese lovers only"
AuthDigestDomain /games/
AuthDigestFile /etc/httpd/access/digest_pw
AuthDigestGroupFile /etc/httpd/access/group
Require group cheddar stilton
</Directory>
Figure 10-13. httpd.conf: The equivalent commands for the Digest protocol

The password file is replaced with one with a different structure, but the group file is
the same as it was before.

77
Chapter 10. Access control

$ touch /etc/httpd/access/digest_pw
$ htdigest /etc/httpd/access/digest_pw "Cheese lovers only" rjd4
Adding user rjd4 in realm Cheese lovers only
New password: password
Re-type new password: password
Figure 10-14. Adding a user to a digest password file

The other issue we mentioned was that text files were used to hold the users, pass-
words and groups. For a small number of users this is fine but if your users reach
into the thousands you may want to consider alternatives that are faster to search.
Alternatively, you may already have an LDAP authentication mechanism and want
to use that. A series of other modules exist for providing Basic authentication with
passwords and groups held in other formats.

Table 10-2. Various storage formats for Basic authentication

auth_module Text files. This is the module we have reviewed extensively

above.
auth_anon_module This acts in a manner analogous to anonymous FTP. It uses the
userid anonymous and requests an email address for the
password. This can be logged.
auth_dbm_module This uses DBM files rather than text files. These are much
faster to look up in once the number of ids stored becomes
large.
auth_ldap_module This uses LDAP to authenticate the user. This can be used to
tie in with an existing LDAP service to stop your users
needing to know yet another password. Unless you specify
the use of LDAPS in the configuration files, the passwords will
travel across the network in plain text from the web server to
the LDAP server.
auth_mysql_module This module checks the userids and groups against a MySQL
database. This is not a standard Apache module but requires
an additional software package.
auth_pgsql_module As above, but using a PostgreSQL database.

Mix and match: Location and Authentication

This brief section shows how the two mechanisms for controlling access, location and
identification, interoperate. Specifically, there is a common desire in the University
to grant passwordless access from within the department or cam.ac.uk domain and
passwordful access otherwise.
Suppose we wanted http://cheese.dept.cam.ac.uk/games/ accessible from
dept.cam.ac.uk without a password and with a password from elsewhere.

<Directory /var/www/CHEESE/games>
Order Allow,Deny
Allow from csi.cam.ac.uk

AuthType Digest
AuthName "Cheese lovers only"
AuthDigestDomain /games/
AuthDigestFile /etc/httpd/access/digest_pw
AuthDigestGroupFile /etc/httpd/access/group
Require group cheddar stilton
78
 Chapter 10. Access control

Satisfy any
</Directory>
Figure 10-15. Mixed restrictions

The two worlds of access control are joined by the Satisfy command. This has two
possible options: Any and All. Satisfy Any requires the request to satisfy either the
location requirement or the authentication requirement. Satisfy All would require it
to satisfy both.

Blocking access based on a file’s name

There is one last aspect of access control we must consider. We have stopped certain
files being listed in indexes in the Section called Automatic indexing of directories in
Chapter 6 but we warned that this did not stop the files being downloaded if the
client could guess the name. This section will demonstrate how to block downloads
of files matching certain expressions in the same way as the IndexIgnore command
stops files matching those patterns being listed.
We can restrict certain commands to files that match regular expressions with the
<FilesMatch> ... </FilesMatch> directive. We can put a simple denial of all access
in this block.
In an ideal world, IndexIgnore and <FilesMatch> would accept the same syntax
for describing their files. Unfortunately they don’t, and this is a serious flaw in the
Apache Software Foundation’s way of handling their modules. IndexIgnore uses
shell-style wildcards, formally known as globbing, and <FilesMatch> uses sed-style
regular expressions.
Our current example configuration file has the line
IndexIgnore "#*#" "*~" "configuration"

and the equivalent <FilesMatch> regular expression is

(^#.*#$|.*~$|^configuration$)

IndexIgnore "#*#" "*~" "configuration"

...
<FilesMatch (^#.*#$|.*~$|^configuration$)>
Order allow,deny
Deny from All
</FilesMatch>
Figure 10-16. Blocking access to the files ignored in the index

79
Chapter 10. Access control

80
Chapter 11. Conclusion

What’s next?
University Computing Service courses that have this course as a prerequisite.

Tidying up
Some re-ordering of the final configuration file.

We have illustrated a number of facilities in the Apache 2 web server which can be
used to create a web server serving multiple web sites.
The configuration file we have built as we go along is syntactically valid, but reflects
its didactic origins. Our final act will be to tidy it up.
The first thing typically done is to move all the LoadModule commands to a block
near the start of the file. That allows us to use all their commands in whatever order
we want further down the file.
The next thing we would do is to reorder the commands to exploit this freedom. In
our case the only major shuffle will be to put the IndexIgnore statement next to the
FilesMatch block that cover the same files.
# These are the absolute basics to launch the web server.
Listen 80
User apache
Group apache
ServerRoot /etc/httpd

# Turn off all options for didactic reasons.

Options None

# Follow symbolic links

Options +FollowSymLinks

# Load the modules needed for this file

LoadModule mime_module modules/mod_mime.so
LoadModule dir_module modules/mod_dir.so
LoadModule autoindex_module modules/mod_autoindex.so
LoadModule alias_module modules/mod_alias.so
LoadModule log_config_module modules/mod_log_config.so
LoadModule userdir_module modules/mod_userdir.so
LoadModule access_module modules/mod_access.so
LoadModule auth_digest_module modules/mod_auth_digest.so

# Set up MIME type recognition by file name extension

TypesConfig /etc/mime.types

# Enable default documents for directory queries

DirectoryIndex index.html

# Enable automatic indexing of directories

Options +Indexes

# Make the indexes "fancy" and read HTML pages’ titles for descriptions
# Don’t show the size and timestamp columns.
# Let the name and description columns be as wide as they need to be
# Put folders first
# Use default sizes for icons
# Arrange that header files should have their own HTML preamble
IndexOptions FancyIndexing ScanHTMLTitles SuppressSize SuppressLastModified NameWidt

81
Chapter 11. Conclusion

# Put a header file above the listing

HeaderName HEADER.html

# Set up aliasing
Alias /icons/ /var/www/icons/

# Set up icons
AddIconByType /icons/layout.gif text/html
AddIconByType /icons/text.gif text/plain
AddIconByType /icons/generic.gif text/*

AddIconByType /icons/image2.gif image/*

AddIconByType /icons/sound1.gif audio/*
AddIconByType /icons/movie.gif video/*

AddIconByType /icons/ps.gif application/postscript

AddIconByType /icons/pdf.gif application/pdf

DefaultIcon /icons/ball.gray.gif

AddAltByType "HTML file" text/html

AddAltByType "Plain text" text/plain
AddAltByType "Text" text/*

AddAltByType "Static image" image/*

AddAltByType "Audio" audio/*
AddAltByType "Video" video/*

AddAltByType "PostScript" application/postscript

AddAltByType "PDF" application/pdf

AddIcon /icons/dir.gif "^^DIRECTORY^^"

AddIcon /icons/back.gif ".."

AddAlt "Directory" "^^DIRECTORY^^"

AddAlt "Up" ".."

# Suppress backup and working files from indexes

IndexIgnore "#*#" "*~" "configuration"

# Delegate control via "configuration" files

AccessFileName "configuration"

<FilesMatch (^#.*#$|.*~$|^configuration$)>
Order allow,deny
Deny from All
</FilesMatch>

# Set the error logging level to "info": informational messages.

LogLevel info

# Specify the error log file.

ErrorLog /var/log/httpd/error.log

# Set up access logging

# Split the log files between the two virtual hosts
# Define the Common Log Format for both of them
LogFormat "%h %l %u %t \"%r\" %>s %b" clf

# We want hosts’ names rather than addresss in the logs

HostnameLookups On

# Users’ own web pages

UserDir public_html

# Set up name-based virtual hosting on all interfaces.

82
 Chapter 11. Conclusion

NameVirtualHost *

<VirtualHost *>
ServerName chalk.dept.cam.ac.uk
DocumentRoot /var/www/CHALK
CustomLog logs/chalk.log clf
</VirtualHost>

<VirtualHost *>
ServerName cheese.dept.cam.ac.uk
DocumentRoot /var/www/CHEESE
CustomLog logs/cheese_log clf

<Directory /var/www/CHEESE/bestiary>
DirectoryIndex main.html index.html
</Directory>

<Directory /var/www/CHEESE/games>
Order Allow,Deny
Allow from csi.cam.ac.uk

AuthType Digest
AuthName "Cheese lovers only"
AuthDigestDomain /games/
AuthDigestFile /etc/httpd/access/digest_pw
AuthDigestGroupFile /etc/httpd/access/group
Require group cheddar stilton

Satisfy any
</Directory>

</VirtualHost>

What’s next?
Having completed this course you are now in a position to follow up by adding
modules that provide for extra facilities. The computing service has two follow-on
courses from this one that build on this foundation.

Follow-on web server courses

Web Server Management: Securing Access to Web Servers
This course introduces the use of the HTTPS (secure http) protocol used to pro-
tect communication between web browsers and web servers. This additional se-
curity is particularly appropriate when sensitive information is being transmit-
ted (passwords, credit card numbers, personal details, etc) or when the identity
of an end-user needs to be established securely.
The course is presented from the point of view of a web server administrator
who wishes to configure servers to support such communication. It provides
an overview of the https protocol and of related cryptographic components, in-
cluding public key and symmetric key encryption, message digests and digital
certificates as they relate to HTTPS. It also describes how to obtain certificates
for webservers, and demonstrates the configuration of an Apache server to use
HTTPS.

83
Chapter 11. Conclusion

CGI Scripting for Programmers: Introduction

The Common Gateway Interface (CGI) underlies much of the modern web. It pro-
vides the most popular way by which web servers can respond to browser re-
quests by invoking programs and using the resulting output as their response.
CGI programs can make decisions based on information contained in browser
requests, and so can be used for vaious tasks, including dynamic page genera-
tion, processing fill-in forms, interfacing to databases, implimenting ’shopping
carts’, etc.
This two-afternoon course covers the CGI itself, various aspects of HTML and
HTTP that are directly relevant to CGI programmers, and general CGI program-
ming issues including security. CGI programs can be written in a variety of lan-
guages, and their use is supported by most web servers. This course uses the
Perl scripting language to develop examples for use with an Apache web server
running under Linux. However the vast majority of the material covered is ap-
plicable to other languages and servers, and should be readily understandable
by anyone with programming experience. Some of the examples may be suitable
for general use.

84
Appendix A. Apache modules
This lists the modules shipped with Red Hat Linux’s packages.

Table A-1. Modules shipped as part of the base httpd package.

Library Module name Description

mod_access.so access_module Access control by browser
hostname.
mod_actions.so actions_module Run specific CGI programs
according to the MIME content
type of the object served.
mod_alias.so alias_module Override the DocumentRoot
directive for specific URLs.
mod_asis.so asis_module Allows the HTTP headers to be
in the file, rather than generated
automatically by the web server.

mod_auth.so auth_module User authentication for access

control usng plain text files..
mod_auth_anon.so auth_anon_module Allows anonymous user access
and logs the password given.
mod_auth_dbm.so auth_dbm_module Just like auth_module but it uses
DBM files rather than text files.
mod_auth_digest.so auth_digest_module Similar to auth_module but
instead of using a plain text
authentication scheme, it uses a
cryptographic one.
mod_autoindex.so autoindex_module Automatically generates
directory listings.
mod_cern_meta.so cern_meta_module An out-of-date way to specify
some headers.
mod_cgi.so cgi_module Run CGI programs.
mod_dav.so dav_module Distributed Authoring and
Versioning functionality. (A
standard for remote authoring
and uploading.)
mod_dav_fs.so dav_fs_module A file system for DAV.
mod_deflate.so deflate_module Compress content prior to
serving it.
mod_dir.so dir_module Supports the use of index.html
files for directory lookups.
mod_env.so env_module Changes the environment that
CGI program are run in.
mod_expires.so expires_module Autogenerates the Expires:
header according to user rules.
mod_headers.so headers_module More general control of HTTP
headers.
mod_imap.so imap_module Server-side image maps.
mod_include.so include_module Server-side includes.
85
Appendix A. Apache modules

Library Module name Description

mod_info.so info_module Lets the server report on its
configuration via a web request.
mod_log_config.so log_config_module Configurable logging of requests
and reponses.
mod_mime.so mime_module Determines MIME types based
on file names.
mod_mime_magic.so mime_magic_module Determines MIME types based
on file contents.
mod_negotiation.so negotiation_module Provides for conent negotiation
between server and client.
mod_proxy.so proxy_module Lets your web server be a proxy.
Typically it needs additional
modules for specific protocols.
mod_proxy_connect.so proxy_connect_module Lets a proxying server handle
CONNECT requests.
mod_proxy_ftp.so proxy_ftp_module Lets a proxying server handle
FTP queries.
mod_proxy_http.so proxy_http_module Lets a proxying server handle
HTTP queries.
mod_rewrite.so rewrite_module Allows for very complex
rewriting of URLs before
responding to them.
mod_setenvif.so setenvif_module Sets the environment for CGI
programs based on properties of
the request.
mod_speling.so speling_module Attempts to correct misspelled
URLs.
mod_status.so status_module Provides information about the
server’s current status via a web
request.
mod_suexec.so suexec_module Allows CGI scripts to run as a
user other than apache.
mod_unique_id.so unique_id_module Provides a unique key in the
environemnt for each request.
mod_userdir.so userdir_module Allows for user’s to have web
pages from their home
directories.
mod_usertrack.so usertrack_module Provision of cookies.
mod_vhost_alias.so vhost_alias_module Allows for handling enormous
numbers of virtual hosts without
having to change the
configuration each time.

A number of other modules are available in other Red Hat Linux packages that de-
pend on the httpd package. Typically the package is named after the library. These
are not sup[ported or maintained by the Apache group and there is no guarantee
that they will not disappear between version of Red Hat Linux. The truly brave may
care to wander through the Red Hat Linux “contributed” package sets for packages
of Apache modules that aren’t provided by Red Hat at all. caveat administrator.

86
 Appendix A. Apache modules

Table A-2. Modules shipped as part of other packages.

Package Library Module name

mod_auth_mysql mod_auth_mysql.so mysql_auth_module
mod_auth_pgsql mod_auth_pgsql.so auth_pgsql_module
mod_perl mod_perl.so perl_module
mod_python mod_python.so python_module
mod_ssl mod_ssl.so ssl_module
php libphp4.so php4_module

87
Appendix A. Apache modules

88
Appendix B. Reference information for logging

Table B-1. Escape sequences for custom logs

%% How to get “%” in the log line. Why would you want to?
%a Client IP address
%A Server IP address. Recall that you may be running different virtual
hosts on different IP addresses.
%B Number of data bytes sent back. (i.e. excluding headers)
%b As for %B except that if the number is 0 then “-” is inserted
instead.
%{fubar}C The value of cookie fubar.
%D The number of microseconds it took to serve the query. See %T
below for a less accurate representation.
%{fubar}e The value of environment variable fubar when the query was
processed.
%f The name of the file whose contents were ultimately served back
to the client.
%H The request protocol. (Typically HTTP or HTTPS.)
%{fubar}i Value of the fubar header on the input query. See also %o below.
%l The remote userid, if provided by RFCnnnn.
%m The request method. Typically “GET” for our queries, but
occasionally “HEAD” if the browser is smart. It may be “POST”
for some CGI programs uploading data.
%{fubar}n A record of a “note” passed from one module to another. Not of
interest at our level.
%{fubar}o The value of header fubar in the outgoingt reponse headers. See
also %i above.
%p The port number of the server. Typically 80.
%P The process ID of the child that serviced the query. Typically only
of use for debugging and trouble-shooting.
%q The query string component of the URL.
%r The first line of the query.
%>s The status code passed back to the client.
%t The time of the request in standard format.
%{format}t The time of the query in the format specified. See the manual page
for strftime for details of the format.
%T The time taken to service the query in seconds. See %D above for
more accuracy.
%u The userid used to authenticate to this page, if necessary.
%U The URL requested without the server name and protocol
elements and without any trailing query string.
%v The server name for the virtual host that was given the query.

HTTP (RFC 26161) is a very subtle protocol with much more happening than you
might expect from the simple stuff we have been covering. The following table lists all
the status codes it has and which might find themselves in your log files. In paractice
you will only see a tiny subset of them.
89
Appendix B. Reference information for logging

Table B-2. HTTP status codes

100 Continue
101 Switching protocols
200 OK
201 Created
202 Accepted
203 Nonauthoritative information
204 No content
205 Reset content
206 Partial content
300 Multiple choices
301 Moved permanently
302 Found
303 See other
304 Not modified
305 Use proxy
307 Temporary redirect
400 Bad request
401 Unauthorized
402 Payment required
403 Forbidden
404 Not found
405 Method not allowed
406 Not acceptable
407 Proxy authentication required
408 Request time-out
409 Conflict
410 Gone
411 Length required
412 Precondition failed
413 Request entity too large
414 Request URI too large
415 Unsupported media type
416 Requested range not satisfiable
417 Expectation failed
500 Internal server error
501 Not implemented
502 Bad gateway
503 Service unavailable
504 Gateway timed out
505 HTTP version unsupported

90
 Appendix B. Reference information for logging

Notes
1. http://www.w3.org/Protocols/rfc2616/rfc2616.html

91
Appendix B. Reference information for logging

92