mod_gzip pour Apache

mod_gzip permet de compresser des elements renvoyé par apache.

Une étude menée sur un des sites que je “gère” a revelé que 22% du traffic était du Javascript et que 17% était du CSS, afin de réduire les couts de bande passante, nous avons dont retenu mod_gzip pour compresser ces elements.

Exemple de configuration

<source lang=apache>

<IfModule mod_gzip.c>
	mod_gzip_on Yes
	mod_gzip_can_negotiate Yes
	mod_gzip_static_suffix .gz
	AddEncoding gzip .gz
	mod_gzip_update_static No
	mod_gzip_command_version '/mod_gzip_status'
	mod_gzip_temp_dir /tmp
	mod_gzip_keep_workfiles No
	mod_gzip_minimum_file_size 500
	mod_gzip_maximum_file_size 1000000
	mod_gzip_maximum_inmem_size 60000
	mod_gzip_min_http 1000
	mod_gzip_handle_methods GET POST
	mod_gzip_item_include mime ^text/css$
	mod_gzip_item_include mime ^text/javascript$
	mod_gzip_item_include mime ^application/x-javascript$
	mod_gzip_dechunk Yes
	mod_gzip_add_header_count Yes
	mod_gzip_send_vary Yes
</IfModule>

</source>

Documentation officielle

Source : http://schroepl.net/projekte/mod_gzip/

#######################################
### Apache configuration directives ###
###   for mod_gzip 1.3.26.1a        ###
#######################################

##########################
### loading the module ###
##########################

# ---------------------------------------------------------------------
# load DLL / Win32:
# LoadModule gzip_module modules/ApacheModuleGzip.dll
#
# load DSO / UNIX:
# LoadModule gzip_module modules/mod_gzip.so
#
# (none of both if module has been compiled in statically;
#  the exact file name may depend upon the exact compilation method used
#  for this module)
# ---------------------------------------------------------------------

  <IfModule mod_gzip.c>

########################
### responsibilities ###
########################

# ---------------------------------------------------------------------
# use mod_gzip at all?
  mod_gzip_on                   Yes
# (you can especially enable mod_gzip inside the central server
#  configuration but disable it inside some directories ot virtual
#  hosts by using this directive.)
# ---------------------------------------------------------------------

######################################
### statically precompressed files ###
######################################

# ---------------------------------------------------------------------
# let mod_gzip perform 'partial content negotiation'?
  mod_gzip_can_negotiate        Yes
# (if this option is active and a static file is to be served in com-
#  pressed for, then mod_gzip will look for a static precompressed
#  version of this file with a defined additional extension - see next
#  directive - which would be delivered with priority. This would allow
#  for avoiding to repeatedly compress the same static file and thus
#  saving CPU time.
#  No dynamic caching of this file is provided; currently the user
#  himself is responsible for creating and updating the precompressed
#  file's content.
#  From version 1.3.19.2a mod_gzip automatically recognizes whether
#  a statically precompressed file is older than its uncompressed
#  original and in this case will serve the content of the original
#  file in uncompressed form - as to rather serve correct data than
#  outdated ones ...)
# ---------------------------------------------------------------------
# extension (suffix) for statically precompressed files
  mod_gzip_static_suffix        .gz
  AddEncoding              gzip .gz
# (effect: see previous directive; this string will be appended to the
#  name of the original file.
#  be sure to configure the encoding 'gzip' for this extension as well,
#  because mod_gzip doesn't serve the content itself but simply generates
#  an Apache internal redirection to this URL. Therefore the remaining
#  Apache configuration is responsible for setting the 'Content-Encoding'
#  header properly ...
#  prior to version 1.3.19.2a this value was not configurable.)
# ---------------------------------------------------------------------
# automatic updates for statically precompressed files
  mod_gzip_update_static        No
# (if set to 'Yes', this directive (being new in version 1.3.26.1a) would
# cause mod_gzip to automatically update an outdated version of any
# statically precompressed file during the request, i. e. compress the
# originally requested file and overwrite the precompressed variant
# file with it!
# for each automatic update of this type, mod_gzip will write a message
# of the severity 'notice' into the Apache error_log.
# while doing so, mod_gzip will directly read the original file's content.
# therefore this content cannot be interpreted by any other Apache module
# during the request. this might possibly not be what you want - hopefully
# it will be what most users want, because it works fast this way.
# use this configuration with a lot of care, and be sure that you don't
# inadvertantly cause valuable files within the URL tree to be overwritten.
# this isn't a feature to be used for mass hosting servers, especially
# because mod_gzip might experience access control problems there - the
# userid the Apache processes are running under need to have write access
# to the precompressed files of all users, which may not automatically be
# the case.)
# [mod_gzip error handling in this situation??? what will be served?]
# ---------------------------------------------------------------------

###################
### bureaucracy ###
###################

# ---------------------------------------------------------------------
# display status for mod_gzip
  mod_gzip_command_version      '/mod_gzip_status'
# (defines an URL to display the status of mod_gzip; can be specified
# individually for each installation and protected against access via
# <Location> section for privacy reasons)
# ---------------------------------------------------------------------
# The status display will look like this:
#       mod_gzip is available...
#       mod_gzip_version = 1.3.26.1a
#       mod_gzip_on = Yes/No
# and thus will provide information about
# - mod_gzip being installed at the server and working correctly,
# - which version has been installed and
# - whether mod_gzip has been set 'active' for this Location
#   (-> mod_gzip_on)
# ---------------------------------------------------------------------

#######################
### data management ###
#######################

# ---------------------------------------------------------------------
# Working directory for temporary files and the compression cache
# if not specified, the following default values are used:
# [Win32=c:\temp], [UNIX=/tmp]
# mod_gzip_temp_dir             /tmp
# (This directory must already exist and the userid being used for
#  running the Apache server must have read and write access to this
#  directory.
#  Unlike other Apache directives an absolute path name must be specified
#  here; a relative value will not be interpreted relatively to ServerRoot.
#  This pastname must NOT be terminated with '/'.
#  For maximum performance this directory should be located on a RAM disk,
#  if the file system isn't already being cached efficiently
# ---------------------------------------------------------------------
# Save temporary work files [Yes, No]
  mod_gzip_keep_workfiles       No
# (one file per HTTP request - set to 'yes' for debugging purpose only!)
# ---------------------------------------------------------------------

##################
### file sizes ###
##################

# ---------------------------------------------------------------------
# minimum size (in bytes) for files to be compressed
  mod_gzip_minimum_file_size    500
# (for very small files compression will produce only small absolute gains
#  [you will still save about 50% of the content, but some additional
#  500 bytes of HTTP and TCP headers will always remain uncompressed],
#  but still produce CPU load for both client and server.
#  mod_gzip will automatically set smaller values than 300 bytes for
#  this directive to exactly this value 300.)
# ---------------------------------------------------------------------
# maximum size (in bytes) for files to be compressed
  mod_gzip_maximum_file_size    500000
# (for very large files compression may eventually take rather long and
#  thus delay the start of the transmission.
#  Furthermode a limitation at this point prevents the server from
#  producing output of unlimited size in case of some endless loop
#  inside a CGI script - or even trying to compress streaming data -
#  which might otherwise cause the creation of a temporary file of
#  any size and even fill up the whole hard disk.
#  On the other hand, compression will have a much more perceivable
#  subjective effect for large files ... so be sure to fine-tune this
#  according to your requirements.)
# ---------------------------------------------------------------------
# maximum size (in bytes) for files to be compressed in memory
  mod_gzip_maximum_inmem_size   60000
# (larger files will be compressed into the temp file directory; adapt
#  this value to your server's available main memory.
#  In mod_gzip 1.3.19.x larger values will automatically be limited to
#  60000 because some operating systems are said to have problems
#  allocating more than 64 kb of memory at a time.
# ---------------------------------------------------------------------

####################
### requirements ###
####################

# (see chapter about caching for problems when using these directives.)
# ---------------------------------------------------------------------
# Required HTTP version of the client
# Possible values: 1000 = HTTP/1.0, 1001 = HTTP/1.1, ...
# This directive uses the same numeric protocol values as Apache does
# internally
  mod_gzip_min_http             1000
# (By using this directive you may exclude old browsers, search engines
#  etc. from the compression procedure: if the user agent doesn't
#  declare itself capable of understanding at least the HTTP level
#  specified here, only uncompressed data will be delivered - no matter
#  what else it claims to be able to. The value of '1001' will especially
#  exclude Netscape 4.x. and a lot of proxy servers.)
# ---------------------------------------------------------------------
# HTTP methods to be handled
# Possible values: 'GET', 'POST' or a list of both values.
  mod_gzip_handle_methods        GET POST
# (By using this directive you may particularly exclude POST requests
#  from the compression procedure. There are known cases where the
#  handling of these requests by previous mod_gzip versions could cause
#  problems.
#  Before version 1.3.19.2a this value was not configurable.)
# ---------------------------------------------------------------------

###############
### filters ###
###############

# ---------------------------------------------------------------------
# which files are to be compressed?
#
# The order of processing during each of both phases is not important,
# but to trigger the compression of a request's content this request
# a) must match at least one include rule in each of both phases and
# b) must not match an exclude rule in any of both phases.
# These rules are not minimal, they are meant to serve as example only.
#
# Note that all parameter values of the directives in this section are
# evaluated as regular expressions, and not in a case-sensitive way.
# ---------------------------------------------------------------------
# phase 1: (reqheader, uri, file, handler)
# ========================================
# NO:   special broken browsers which request for gzipped content
#       but then aren't able to handle it correctly
  mod_gzip_item_exclude         reqheader  "User-agent: Mozilla/4.0[678]"
# From version 1.3.19.2a on I advise against using filters
# for User-agents, as this will cause HTTP-Headers 'Vary: User-Agent'
# to be generated, thus making life more difficult for proxy servers.
#
# JA:   HTML-Dokumente
  mod_gzip_item_include         file       \.html$
#
# NO:   include files / JavaScript & CSS (due to Netscape4 bugs)
  mod_gzip_item_exclude         file       \.js$
  mod_gzip_item_exclude         file       \.css$
#
# YES:  CGI scripts
  mod_gzip_item_include         file       \.pl$
  mod_gzip_item_include         handler    ^cgi-script$
#
# phase 2: (mime, rspheader)
# ===========================
# YES:  normal HTML files, normal text files, Apache directory listings
  mod_gzip_item_include         mime       ^text/html$
  mod_gzip_item_include         mime       ^text/plain$
  mod_gzip_item_include         mime       ^httpd/unix-directory$
#
# NO:   images (GIF etc., will rarely ever save anything)
  mod_gzip_item_exclude         mime       ^image/
# ---------------------------------------------------------------------
# In fact mod_gzip is checking only the first 4 characters of the 1st
# operand (in case of uri even the first 2 characters only, as to
# allow for values like url).
# ---------------------------------------------------------------------
# The table for mod_gzip_item rules (include and exclude) cannot contain
# more than 256 entries; when this number is exceeded mod_gzip will
# output the message "mod_gzip: ERROR: Item index is full"
# and report a configuration error to the Apache server.
# ---------------------------------------------------------------------
# The directive values described here are meant to describe the requests
# elected for compression most exactly.
# Especially for the mime rules it has to be made clear that the HTTP
# header 'Content-Type' (that will be checked by mod_gzip for this rule)
# in some cases may contain not only a MIME type but additionally a
# character set description (charset) as well.
# If this is the case for the requests to be handled then you need to
# remove the '$' char at the end of the corresponding value so that now
# only the prefix of this value will be tested for matching.
# ---------------------------------------------------------------------

##########################
### transfer encodings ###
##########################

# ---------------------------------------------------------------------
# Allow mod_gzip to eliminate the HTTP header
#    'Transfer-encoding: chunked'
# and join the chunks to one (compressable) packet
  mod_gzip_dechunk              Yes
# (this is required for handling several types of dynamically generated
# contents, especially for CGI and SSI pages, but also for pages produced
# by some Java Servlet interpreters.
# ---------------------------------------------------------------------

###############
### logging ###
###############

# ---------------------------------------------------------------------
# Extended log format (for testing the compression effect)
  LogFormat                     "%h %l %u %t \"%V %r\" %<s %b mod_gzip: %{mod_gzip_result}n In:%{mod_gzip_input_size}n -< Out:%{mod_gzip_output_size}n = %{mod_gzip_compression_ratio}n pct." common_with_mod_gzip_info2
# ---------------------------------------------------------------------
# Create additional log file
  CustomLog                     logs/mod_gzip.log common_with_mod_gzip_info2
# (surely you can redefine your normal log file format, but you mal well
#  keep its format standard compatible for evaluation by standard web
#  analysis tools. So we just create another log file.)
# ---------------------------------------------------------------------
# Volume computation of the delivered files inside the Apache access_log:
# count HTTP header size (in bytes) as part of total output size
  mod_gzip_add_header_count     Yes
# (This will be more than the pure document content, but it will more
#  realistically describe the total output traffic of the HTTP request)
# ---------------------------------------------------------------------


###############
### proxies ###
###############

# ---------------------------------------------------------------------
# sending a 'Vary' HTTP header
  mod_gzip_send_vary            On
# (see chapter about caching for this directive.)
#  don't change this unless you absolutely know what you are doing!
# ---------------------------------------------------------------------
  </IfModule>
wiki/mod_gzip.txt · Dernière modification: Friday 03 July 2009 par kathryl
Flux RSS du Blog Driven by DokuWiki Gentoo Powered Valid XHTML1.0 Powered by Apache PHP Powered Coffee Powered