[Chapter 12] 12.2 Installing and Configuring a Web Server

12.2 Installing and Configuring a Web Server

Installing and configuring a web server is not much more difficult than installing an FTP server. Once your web server is up and running, other Internet users can view documents you publish on your Linux system.

12.2.1 Configuration

Configuring a web server can be as easy or as difficult as you choose. Like other web servers, Apache provides seemingly countless options. Fortunately, Debian GNU/Linux automatically configures Apache when you install it. However, the configuration options selected by the install scripts may not suit your needs. In that case, you can modify the configuration files.

Apache's configuration files reside in the directory /etc/apache. For historical reasons that no longer apply, Apache has three configuration files:

access.conf: Specifies what hosts and users are allowed access to what documents and services
httpd.conf: Specifies options that govern the operation of the httpd daemon
srm.conf: Specifies how your server's documents and organized and formatted

Currently, you can place Apache configuration commands in any of these files. However, each of these files must exist, even if it is empty; otherwise, the httpd daemon will refuse to run. As distributed, the files contain a default configuration. Before starting the web server, you should revise the ServerName option of the httpd.conf file. The three following subsections describe other options that you may wish to specify. You can scan them to see what options are available and specify options that interest you. A more complete description of the options is available in Apache's online documentation. Also, the Apache web site ( http://www.apache.org/) provides a tutorial on Apache configuration. To change an option, simply open the related file by using your favorite text editor, change the file as you wish, and save the file.

The subsections assume some familiarity with HTML and web servers. If you find that some options are obscure, don't fret; your web server will serve ordinary HTML pages even if you set no options other than ServerName.

12.2.1.1 The access.conf file

The access.conf file specifies a default set of permissions that govern access to documents and services. It then specifies sets of permissions that override the default permissions for particular documents and services. The usual practice is to specify a quite restrictive set of default permissions, relaxing these permissions to provide access to particular documents and services.

The file contains a mixture of comments (lines beginning with #) and directives. Comments are ignored by the server.

The default permissions are specified as follows:

<Directory />
Options None
AllowOverride None
</Directory>

The paired tags <Directory> and </Directory> enclose a list of options that pertain to the / directory, the directory specified in the <Directory> tag. The options are:

Options None, which specifies that no special server features are enabled for the specified directory or its subdirectories.
AllowOverride None, which specifies that access specifications cannot be overridden by an .htaccess file.

Table 12.2 describes special server features that are available.

Table 12.2: Special Server Features
Option	Description
`ExecCGI`	Execution of CGI scripts is permitted in this directory.
`FollowSymLinks`	The server will follow symbolic links in this directory.
`Includes`	Server-side includes are permitted.
`IncludesNOEXEC`	Server-side includes, except `#exec` and `#include`, are permitted in this directory.
`Indexes`	If the directory contains no index file (for example, index.html), the server will prepare a formatted index.
`MultiViews`	Content-negotiated MultiViews are permitted in this directory. MultiViews permit, for example, a client browser to select a document in a particular language from a set of documents.
`SymLinksIfOwnerMatch`	The server will follow symbolic links for which the target file or directory has the same owner as the link.

Unless the specification for a directory specifies AllowOverride None, you can override the specified options by placing an .htaccess file in the directory or one of its subdirectories. The .htaccess file can contain specifications of the same sort as the access.conf file; the server applies the specifications in the .htaccess file in preference to those specified in the access.conf file.

After the restrictive default specifications come some more relaxed specifications:

<Directory /home/httpd/html>
Options Indexes Includes FollowSymLinks
AllowOverride None
order allow,deny
allow from all
</Directory>

These specifications apply to the directory /home/httpd/html and its subdirectories. Here, the Indexes, Includes, and FollowSymLinks options are specified. As for the root directory, use of .htaccess files is forbidden, via AllowOverride None. Here, unlike the specification for the root directory, the hosts allowed to access documents are services are specified. The order allow,deny directive specifies that any deny directives will be applied after any allow directives, and will therefore take precedence. No deny directives appear in this specification; the allow from all directive permits any host to access documents and services within the /home/httpd/html directory and its subdirectories.

Another specification allows execution of CGI scripts within the /home/httpd/cgi-bin directory and its subdirectories:

<Directory /home/httpd/cgi-bin>
AllowOverride None
Options ExecCGI
</Directory>

The final specification in the default configuration lets the local host access HTML documents within the /usr/doc directory and its subdirectories:

<Directory /usr/doc>
order deny,allow
deny from all
allow from localhost
Options Indexes FollowSymLinks
</Directory>

12.2.1.2 The srm.conf File

The srm.conf file specifies the organization and format of documents provided by your web server. As was the case with the access.conf file, you don't need to make any changes to the srm.conf file, though you may wish to do so.

The DocumentRoot directive specifies the directory that contains your HTML files. When a web client accesses the root directory, the server actually fetches files from the directory specified as DocumentRoot:

DocumentRoot /home/httpd/html

The UserDir directive specifies the name of the subdirectory that the server appends to a user's home directory when a client makes a ~user request:

UserDir public_html

This directive specifies that a reference to ~user will be translated to a reference to /home/user/public_html.

The DirectoryIndex directive specifies the name of the file (or names of the files) used a directory indexes:

DirectoryIndex index.html index.shtml index.cgi

The FancyIndexing directive specifies whether icons are used to produce fancy directory indexes:

FancyIndexing on

The AddIcon and AddIconByType directives associate icons with files of given types:

AddIconByType (TXT,/icons/text.gif) text/*
AddIcon /icons/binary.gif .bin .exe

The default configuration includes many such directives. If you wish to add support for a new type of file, you may want to add a directive associating an icon with the new file type.

The DefaultIcon directive specifies the icon used for file types not explicitly associated with an icon:

DefaultIcon /icons/unknown.gif

The ReadmeName directive specifies the name of a file used by the server to produce readme entries:

ReadmeName README

The server will first look for the file README.html and then for the file README.

The similar HeaderName directive specifies the name of a file that the server will prepend to a generated index:

HeaderName HEADER

The IndexIgnore directive specifies a set of file names that should not be included in a generated index. These are often specified by using wildcard characters:

IndexIgnore .??* *~ *# HEADER* README* RCS

The AccessFileName directive specifies the name of the file that, if present, overrides access control specifications for a directory:

AccessFileName .htaccess

The TypesConfigdirective identifies the mime.types file, which describes multimedia files:

TypesConfig /etc/mime.types

The DefaultTypedirective specifies the default MIME type for documents:

DefaultType text/plain

The AddEncoding directive instructs compatible browsers to uncompress information as it's downloaded:

AddEncoding x-compress Z
AddEncoding x-gzip gz

The Redirect directive lets you provide a forwarding address for documents that have moved. The default configuration includes no Redirects, which have this simple form:

Redirect 
old-URL new-URL

The Alias directive lets you refer to a directory by using an alias. For example, the following directive provides a more convenient way of referring to the /home/httpd/icons directory:

Alias /icons/ /home/httpd/icons/

The ScriptAlias directive lets users refer to the CGI directory as simply /cgi-bin/:

ScriptAlias /cgi-bin/ /home/httpd/cgi-bin/

The following sets of specifications work around problems with several browsers. First, two directives that disable keepalive for browsers that do not support it:

BrowserMatch "Mozilla/2" nokeepalive
BrowserMatch "MSIE 4\.0b2;" nokeepalive downgrade-1.0 force-response-1.0

The final directives force use of HTTP version 1.0 responses for several browsers that do not support HTTP 1.1:

BrowserMatch "RealPlayer 4\.0" force-response-1.0
BrowserMatch "Java/1\.0" force-response-1.0
BrowserMatch "JDK/1\.0" force-response-1.0

12.2.1.3 The httpd.conf File

The httpd.conf file specifies options related to the httpd daemon. You should specify the ServerName option before starting your web server.

The ServerType directory specifies whether the web server is started via inetd or standalone:

ServerType standalone

The port directive specifies the port on which the web server listens for client requests:

Port 80

The HostnameLookups directive specifies whether clients are logged by IP address (off) or hostname (on):

HostnameLookups off

The User and Group directives specify the userid and group under which the httpd daemon runs. The daemon initially runs as root and then switches to the specified userid and group. The default configuration specifies the userid as nobody, a standard Unix userid that has very limited permissions. A user or process running as nobody can access files only in ways permitted to all users; generally, this means the user or process cannot modify files. The group nobody has similarly constrained privileges:

User nobody
Group nobody

The ServerAdmin directive specifies the email address of the server administrator:

ServerAdmin root@localhost

The ServerRoot directive specifies the directory that contains the configuration, error, and log files:

ServerRoot /etc/httpd

The BindAddress directive provides for virtual hosts. It specifies the IP address to which the server should listen. It is normally disabled by a comment token:

#BindAddress *

The ErrorLog directive specifies the location of the error log file:

ErrorLog logs/error_log

The LogLevel directive specifies the verbosity of the server log. Possible values include debug, info, notice, warn, error, crit, alert, and emerg:

LogLevel warn

The LogFormat directive specifies format names that can be used with the CustomLog directive:

LogFormat "%h %l %u %t \"%r\" %>s %b \"%{Referer}i\" \"%{User-Agent}i\""
combined LogFormat "%h %l %u %t \"%r\" %>s %b" common
LogFormat "%{Referer}i -> %U" referer
LogFormat "%{User-agent}i" agent
CustomLog logs/access_log common

For more information about LogFormat and CustomLog, see the Apache documentation.

The PidFile directive specifies the file in which the server should log its process id:

PidFile /var/run/httpd.pid

The ScoreBoardFile directive specifies the name of the file used to store internal server process data:

ScoreBoardFile /var/run/httpd.scoreboard

The LockFile directive specifies the file used to providing locking. You'll need to change this option only if you NFS mount the directory used by the server for logging:

#LockFile /var/lock/httpd.lock

The ServerName directive specifies the hostname of the system on which your server runs. Depending on your network configuration, you may not need to specify this directive:

ServerName host.domain.com

The UseCanonicalName directive specifies whether the server will return a canonical URL formed from the ServerName and Port directives (on) or the hostname and port supplied by the client (off):

UseCanonicalName on

The CacheNegotiatedDocs directive instructs browsers not to cache documents; it is usually disabled by prefixing it with a comment token (#):

#CacheNegotiatedDocs

The Timeout directive specifies the maximum number of seconds the server will wait for certain responses, such as the next packet in a sequence of TCP packets:

Timeout 300

The KeepAlive directive specifies that connections are persistent; that is, that a client can make multiple requests per connection:

KeepAlive On

The MaxKeepAliveRequests directive specifies the maximum number of requests permitted during a persistent connection:

MaxKeepAliveRequests 100

The value 0 denotes an unlimited number of requests.

The KeepAliveTimeout directive specifies the maximum number of seconds during which the server will wait for the next request:

KeepAliveTimeout 15

The MinSpareServers and MaxSpareServers directives respectively fix the minimum and maximum number of spare server processes the server will create. Having an available server expedites handling of an incoming request:

MinSpareServers 8
MaxSpareServers 20

The StartServers directive specifies the initial number of server processes:

StartServers 10

The MaxClients directive specifies the maximum number of simultaneous server processes. When this number is reached, requests from additional clients are locked out:

MaxClients 150

The default values of these options are higher than necessary for most workstation users. Your system will perform more efficiently (and perhaps more reliably) if you change the directives to specify the following values:

MinSpareServers 3
MaxSpareServers 6
StartServers 3
MaxClients 20

The MaxRequestsPerChild directive specifies the number of requests a child process can handle before expiring. This ensures that processes are periodically recreated, minimizing problems due to software errors such as memory leaks:

MaxRequestsPerChild 100

The ProxyRequests directive specifies whether the proxy server is enabled; it is normally disabled:

#ProxyRequests On

When the proxy server is active, the following directives specify various caching options. They are normally disabled:

#CacheRoot /var/cache/httpd
#CacheSize 5
#CacheGcInterval 4
#CacheMaxExpire 24
#CacheLastModifiedFactor 0.1
#CacheDefaultExpire 1
#NoCache a_domain.com another_domain.edu joes.garage_sale.com

The Listen directive lets your bind Apache to a specific IP address or port, in addition to the default IP address and port. It is generally disabled:

#Listen 3000
#Listen 12.34.56.78:80

The <VirtualHost> and </VirtualHost> tags enclose a series of options that establish a virtual host, useful if your system has multiple IP addresses. The options can include any of the options described in this subsection. The tags and options are normally disabled:

#<VirtualHost host.some_domain.com>
#ServerAdmin webmaster@host.some_domain.com
#DocumentRoot /www/docs/host.some_domain.com
#ServerName host.some_domain.com
#ErrorLog logs/host.some_domain.com-error_log
#TransferLog logs/host.some_domain.com-access_log
#</VirtualHost>

12.2.2 Startup and Use

The install script will automatically start your web server as soon as you install it. If you installed the lynx browser, you can use it to test your web server. Issue the command:

lynx http://localhost

You should see a screen that resembles Figure 12.1.

Figure 12.1: The Apache start page viewed by lynx

If you prefer, you can view the start page by using Netscape Navigator, which supports the graphics embedded in the page. The result should resemble Figure 12.2.

Figure 12.2: The Apache start page viewed by Netscape Navigator

Once you can access your web server locally, try accessing it from a remote computer. This should be as simple as forming a URL that includes the fully qualified hostname of your system (that is, the host and domain names); for example, http://mysystem.mydomain.


	Learning Debian GNU/Linux By Bill McCarty 1st Edition September 1999 1-56592-705-2, Order Number: 7052 360 pages, $34.95 , Includes CD-ROM


12.1 An FTP Server		12.3 Using a Mail Server

Learning Debian GNU/Linux