Goal:

• Learn how to configure apache as
  • web server.
  • cache server
  • reverseproxy
• Learn how to form a single cluster using reverseproxy.
• Learn how to use httperf to evaluate the web server performance.

Assignment Date: 1/25/2005
Due Day: 2/1/2005
Description:

• In this exercise, we will learn how to configure apache to run as a web server.
• On our CS Unix machines, there is a httpd program in /usr/sbin/. It is the Apache HyperText Transfer Protocol (HTTP) server program. It is designed to be run as a standalone
  daemon process. Read man page with "man httpd" for more info.
  

  SYNOPSIS
  httpd [ -X ] [ -R libexecdir ] [ -d serverroot ] [ -f con-fig ] [ -C directive ] [ -c directive ] [ -D parameter ]
  

• If "/usr/sbin/httpd -d /users/server/students/cs526/public_html/apache/ws1 -X" is run, the httpd program will treat /users/server/students/cs526/public_html/apache/ws1 as serverroot directory and read the directives in the conf/httpd.conf file for configuring the apache server. -X is for debugging purpose, it indicates to the httpd daemon that do not fork spare child process for improving performance. Just run a single standalone process.
• The Apache provides a rich set of directives for configuring/customizing the web server. A subset of directives are realized by an Apache module. See http://blanca.uccs.edu/manual/mod/directives.html for the set of runtime directives.
• An example of httpd.conf is
  
  LoadModule log_config_module /etc/httpd/modules/mod_log_config.so
  LoadModule access_module /etc/httpd/modules/mod_access.so

  User cs526
  ServerName sanluis.uccs.edu
  Port 8388

  DocumentRoot /users/server/students/cs526/public_html/apache/ws1/htdocs

  ErrorLog logs/error_log
  TransferLog logs/access_log
  PidFile logs/httpd.pid

  
  
• The first directive LoadModule tells the httpd program to dynamically load in /etc/httpd/modules/mod_log_config.so as dynamic shared object (very much like dll). The log_module code deals with creating error log entries in the log files.
• Access_module deal with creating access log entries in the log files.
• User directives tells httpd to run itself as cs526 account.
• ServerName tells httpd the domain to use when displaying error msg.
• Port directive tells httpd to listen to port 8388 for incoming http requests. We need to enter http://sanluis.ccs.edu:8388/ to access this webserver.
• DocumentRoot specifies the actual path to the default web document directory.
• EorrLog specifies the name of the file for saving error log records.
• TransferLog specifies the name of the file for saving the access records.
• PidFile specifies the names fo the files for saving the process id of the httpd server. It can be used by system admin to restart or stop the httpd server.

Unfortunately, the cache module was removed from the apache 1.3 in apache 2.0. I just recompiled apache 1.3.28. To simplify the task, instead of using the tempate file in ~cs526/public_html/apache we will use ~cs526/public_html/apache1.3

• For exercise1, we would like you to copy the directory ~cs526/public_html/apache1.3 to your ~<login>/public_html directory, using "cp -r ~cs526/public_html/apache1.3 ~<login>/public_html.
• login to blanca.
• cd public_html/apache/ws1/conf
• edit httpd.conf file by replacing all LOGIN instances with your own login name on CS Unix machines. In other words,
  • Replace ccount on User directive with your own account.
  • Replace DocumentRoot and point to your own htdocs directory.
  • Replace ErrorLog, TransferLog, and PidFile with your own directory.
• Replace the port number with 8<last 3 digits of your SID> in the Port directive.
• "cd ../htdocs" to web server default document directory.
• Edit the index.html in htdocs directory with your login name.
• Edit the cmd file substitute the "cs526/public_htlm" string there with "<your loginname>/public_html".
• Run the apache web server with "./cmd
  which is
  "/users/server/students/cs526/apache_1.3.28/src/httpd -d /users/server/students/LOGIN/public_html/apache1.3/ws1 -X"
• Go to any browser and enter "http://blanca.uccs.edu:8<last 3 digits of your SID>/index.html" and see if the web page shows up.
• Note that we need add index.html in the url because we did not specify use DirectoryIndex directive to set the index.html.
• Check the logs in /users/server/students/<yourloginname>/public_html/apache1.3/ws1/log/access_log and see the access records is created.
• Email me the url of the access_log file.
• Login to sanluis.
• cd public_html/apache1.3/ws2
• Repeat the above steps for sanluis web site.
• Edit the cmd file substitute the "cs526/public_htlm" string there with "<your loginname>/public_html".
• Run the apache web server with "./cmd
  which is
  "/users/server/students/cs526/apache_1.3.28/src/httpd -d /users/server/students/<your loginname>/public_html/apache1.3/ws2 -X"
• cp -r ws1 ws3 to create ws3 directory.
• login to shavano. Edit the httpd.conf file and web page files in ws3/conf/htdocs directory to reflect the machine shavano which will serve the web page. Create a subdirectory personnel and a personnel web page with "this is shavano personnel web page " as content in the htdocs directory.
• Run the web server and leave some access records.

• In your apache1.3 directory, there is a proxy subdirectory.
• login to shavano and cd to the proxy directory.
• create a cache direcotry with "mkdir cache"
• Edit the conf/httpd.conf with your User account and Port number.
  
  LoadModule proxy_module /users/server/students/cs526/apache_1.3.28/src/modules/proxy/libproxy.so
  LoadModule config_log_module /users/server/students/cs526/apache_1.3.28/src/modules/standard/mod_log_config.so
  LoadModule access_module /users/server/students/cs526/apache_1.3.28/src/modules/standard/mod_access.so

  User LOGIN
  ServerName shavano.uccs.edu
  ProxyRequests on
  CacheRoot /users/server/students/LOGIN/public_html/apache1.3/proxy/cache
  CacheSize 10000000
  Port 8388

  
  ErrorLog /users/server/students/LOGIN/public_html/apache1.3/proxy/logs/error_log
  TransferLog /users/server/students/LOGIN/public_html/apache1.3/proxy/logs/access_log
  PidFile /users/server/students/LOGIN/public_html/apache1.3/proxy/logs/httpd.pid
  

  
  
  
• Add CacheDirLevels and CacheDirLength directive to the httpd.conf and change the default settings to 2 and 3 respectively. See directive info at http://httpd.apache.org/docs/mod/directives.html
• Remove the cache directory each time your start another run. There is .time timestamp created that seems to prohibit the reuse of the directory for a different run. Bonus exercise: figer out how to allow the re-use of the cache content in different run. Only the first submission wins the bonus.
• Run the cache server with "./cmd" make sure to change the web site directory.
• Setup your browser to use the proxy server as follows:
  • On Internet Explorer,
    • Select Tool | Internet option...
    • Select Connections tab and click the LAN Settings button.
  • Check "Use a proxy server for your LAN" and enter "shavano.uccs.edu" as address and "8<last3digts of your SID>" for the port number. This is port number your cache server is listening.
  • Click OK.
  • On Netscape/mosaic browser,
    • Select Edit | Preference, Preference window appears.
  • Double click on the "advanced menu" in the catalog section, it opens out more choices.
  • Select the Proxies menuitem Click manual proxy configuration
  • Enter shavano.uccs.edu for HTTP Proxy:
  • Enter 8<last 3digits of SID> for Port:
  • Click OK.
• Enter the url to your personal web page for CS526 in the address box of your IE browser.
• Locate your personal web page in the cache directory.
• Report the path to that <login>.html file and the first line and the HTTP header section of the cached file. Verify that the path to cached document indeed contained the subdirectory name followed the specification of CacheDirLength directive.
• Try download a large file such as http://apache.towardex.com/httpd/httpd-2.0.52.tar.bz2 twice. The second download should be quicker due to caching.
• After the exercise, do not forget to uncheck the "Use a proxy server for your LAN" Internet option. Otherwise you IE will not work. If you use netscape, select Proxies menuitem in Edit | Preference window and click on "Direct connection to the internet" option. Remove the large file(s) from cache afterwards.
• Question 2.1. Why we put cached documents in directories of different level as opposed to put them all in a single cache directory?
• Question 2.2. Why the cached document is saved with a file name generated by applying hash on the url?
• Question 2.3. One of the HTTP headers in the cached document file is the last modified. How the cache server uses the value of this header? (Hint: http://cs.uccs.edu/~cs522/hw/hwF2004/hw.htm#hw1 exercise 2.4; http://www.w3.org/Protocols/rfc2616/rfc2616-sec13.html#sec13).

• An apache web site template is availabe in apache/reverseproxy directory.
• login to crestone and cd to reverseproxy directory.
• Edit the conf/httpd.conf with your User account and Port number.
• LoadModule proxy_module /users/server/students/cs526/apache_1.3.28/src/modules/proxy/libproxy.so
  LoadModule config_log_module /users/server/students/cs526/apache_1.3.28/src/modules/standard/mod_log_config.so
  LoadModule access_module /users/server/students/cs526/apache_1.3.28/src/modules/standard/mod_access.so
  

  User LOGIN
  ServerName crestone.uccs.edu
  #ProxyRequests on
  Listen 8388

  DocumentRoot /users/server/students/LOGIN/public_html/apache1.3/reverseproxy/htdocs/

  ProxyPass /eng/ http://blanca.uccs.edu:8388/eng/
  ProxyPassReverse /eng/ http://blanca.uccs.edu:8388/eng/

  ProxyPass /sales/ http://sanluis.uccs.edu:8388/sales/
  ProxyPassReverse /sales/ http://sanluis.uccs.edu:8388/sales/

  ErrorLog logs/error_log
  TransferLog logs/access_log
  PidFile logs/httpd.pid
  

  
  
• ProxyPass directive tells the httpd to pass any url request with /eng as prefix to blanca web server.
• ProxyPassReverse tells the httpd to modify the Location header value return from the domain name of the real web server to crestone.uccs.edu. Therefore the client will not see replies from different web servers. They are all from crestone.uccs.edu.
• Add shavano as a web server where crestone reverseproxy server will forward all web pages with /personnel as prefix in the url.
• Run blanca, sanluis, and shavano web servers according to Exercise 1.
• Run crestone as reverseproxy server.
• Access http://crestone.uccs.edu:8<last3digitsofSS#>/eng/eng.html
• Access http://crestone.uccs.edu:8<last3digitsofSS#>/sales/sales.html
• Access http://crestone.uccs.edu:8<last3digitsofSS#>/personnel/personnel.
• See if the return web pages are indeed from the right web server.

Exercise 4. Evaluate Web Server Performance using httperf.

Exercise 4b: Evaluate Web Server Performance using WebBench (optional)

• Use WebBench on Ardor.uccs.edu to evaluate the performance of the Apache web server on a CS Unix Machine and win2000 server on wait.uccs.edu.
• Setup an apache web server on a Unix machine as Exercise1
  • You need to copy the webbench tar file 3wtree.tar.gz (7.29MB) from ~cs526/public_html/webbench directory to the htdocs (document root) directory of your web server.
  • tar xvzf 3wtree.tar.gz to restore the directory wbtree there.
  • Check if one of the data file http://blanca.uccs.edu:8<3digits>/wbtree/11426_1.htm is there. Assume blanca is where your web server runs.
  • Note that for IIS on wait.uccs.edu. I have set up the wbtree data files already.
  • For window system, make sure you download the 3wtree.exe winzip file 20.4MB). The Unix file format .tar.gz and winzip file format are different. If you download the incorrect file, you will get 404 error.
  • The location for download the web bench http://www.etestinglabs.com/benchmarks/webbench/webbench.asp?visitor=
• Setup WebBench Controller and client. Using the static_mt test suite. With Ramp Up time=3 seconds, Ramp Down time=3 seconds, and length=30 seconds. We would like to shorten the testing period since we are time share both wait.uccs.edu web server and webbench machine ardor.uccs.edu.
• See instruction on installing webbench and configuration and running the test suite.
  http://cs.uccs.edu/~cs526/webbench/webbench.htm
• You have the choice of running the tests at home or use the ardor.uccs.edu.
• Ardor.uccs.edu has already installed with webbench. You can access them in start | all programs | pc magazine benchmarks. Login with cs526 as username.
• To set the shorter Ramp Up/Down times and length, first follow the steps in http://cs.uccs.edu/~cs526/webbench/webbench.htm#runWebbench
• In Edit Mix Defintion step, enter those new Ramp Up/Down times and length parameters as below
  
• Make sure you change the port and server IP address in the mix definition workload dialog box. http://cs.uccs.edu/~cs526/webbench/webbench.htm#portIPaddr
• Then click on "Copy Across" button on the right. It will bring up the Duplicate. Select the Ramp Up/Down times, length, www port, and www server options we just changed and click OK so that the changes will apply to all test runs (1,4,8...,60 clients)
  
• After the test, to see the test results, select results | view results ... and choose the test suits static_mt.tlg which we use. A dialog box similar to below will appear.
  
  Make sure all worksheets options are selected and "Save Workbook with Results Name" is selected.
  You can edit the Client/Server configuration to reflect their OS version, CPU, memory, and network configuration. Or just enter those info in the comments.
  Select View to bring up the Excell spreadsheet.
• You will see a table summarized the mix names and the measured response time (requests/seconds) and throughtput (bytes/sec) numbers. It also shows the plots of those two performance parameters over different number of clients.
• Save the Excell spreadsheet results for your exercise 4.
• Repeat the same thing for the IIS server running on wait.uccs.edu.
• Compare the performance of the two web servers.
  Hint: You can combine the two related plots to highlight the comparison.
• In your hw1 of submission, include spreadsheets (or a url link to them) and your assessment of the two web severs.

Create a homework web page hw1.html with the above answers