Miles to go …

TOTD #81 explained how to install/configure nginx for load-balancing/front-ending a cluster of Rails application running on GlassFish Gem. Another popular approach in the Rails community is to use Apache HTTPD + mod_proxy_balancer. A user asked the exact details of this setup on the GlassFish Gem Forum. This Tip Of The Day (TOTD) will clearly explain the steps.

1. Create a simple Rails scaffold and run this application using GlassFish Gem on 3 separate ports as explained in TOTD #81.
2. Setup and configure HTTPD and mod_proxy_balancer
   1. Setup and install Apache HTTPD as explained here. I believe mod_proxy_balancer and other related modules comes pre-bundled with HTTPD, at least that’s what I observed with Mac OS X 10.5.7. Make sure that the “mod_proxy_balancer” module is enabled by verifying the following line is uncommented in “/etc/apache2/httpd.conf”:

      LoadModule proxy_balancer_module libexec/apache2/mod_proxy_balancer.so

      Please note another similar file exists in “/etc/httpd/httpd.conf” but ignore that one.

   2. Setup a mod_proxy_balancer cluster by adding the following fragment in “httpd.conf” as:
      

      <Proxy balancer://glassfishgem> BalancerMember http://localhost:3000 BalancerMember http://localhost:3001 BalancerMember http://localhost:3002 </Proxy>

      The port numbers must exactly match with those used in the first step.

   3. Specify the ProxyPass directives to map the cluster to a local path as:
      

      ProxyPass / balancer://glassfishgem/ CustomLog /var/log/glassfishgem.log/apache_access_log combined

      The “/” at the end of “balancer://glassfishgem” is very important to ensure that all the files are resolved correctly.

   4. Optionally, the following directive can be added to view the access log:
      

      CustomLog /var/log/glassfishgem.log/apache_access_log combined

      Make sure to create the directory specified in “CustomLog” directive.

3. Now the application is accessible at “http://localhost/runlogs”. If a new GlassFish instance is started then update the <Proxy> directive and restart your HTTPD as “sudo httpd -k restart”. Dynamic update of BalancerMembers can be configured as explained here.

TOTD #81 started the Rails application in root context. You can alternatively start the application in a non-root context as:

~/tools/jruby/rails/runner >../../bin/jruby -S glassfish -e production -c myapp Starting GlassFish server at: 10.0.177.178:3000 in production environment… Writing log messages to: /Users/arungupta/tools/jruby-1.3.0/rails/runner/log/production.log. Press Ctrl+C to stop. . . . ~/tools/jruby/rails/runner >../../bin/jruby -S glassfish -e production -c myapp -p 3001 Starting GlassFish server at: 10.0.177.178:3001 in production environment… Writing log messages to: /Users/arungupta/tools/jruby-1.3.0/rails/runner/log/production.log. Press Ctrl+C to stop. . . . ~/tools/jruby/rails/runner >../../bin/jruby -S glassfish -e production -c myapp -p 3002 Starting GlassFish server at: 10.0.177.178:3002 in production environment… Writing log messages to: /Users/arungupta/tools/jruby-1.3.0/rails/runner/log/production.log. Press Ctrl+C to stop.

and then the ProxyPass directive will change to:

ProxyPass /myapp/ balancer://glassfishgem/myapp/

The changes are highlighted in bold. And the application is now accessible at “http://localhost/myapp/runlogs”.

After discussing on Apache HTTP Server forum, the BalancerMember host/port can be printd in the log file using a custom log format. So add the following log format to “/etc/apache2/httpd.conf”:

LogFormat “%h %l %u %t \”%r\” %>s %b \”%{Referer}i\” \”%{User-agent}i\” \”%{BALANCER_WORKER_NAME}e\”" custom

And change the format from the default “combined” to the newly defined “custom” format as:

CustomLog /var/log/glassfishgem.com/apache_access_log custom

Three subsequent invocations of “http://localhost/runlogs” then prints the following log entries:

::1 – - [17/Jun/2009:10:53:53 -0700] “GET /runlogs HTTP/1.1″ 304 – “-” “Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.0.11) Gecko/2009060214 Firefox/3.0.11″ “http://localhost:3002″ ::1 – - [17/Jun/2009:10:54:04 -0700] “GET /runlogs HTTP/1.1″ 200 621 “-” “Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.0.11) Gecko/2009060214 Firefox/3.0.11″ “http://localhost:3000″ ::1 – - [17/Jun/2009:10:54:05 -0700] “GET /runlogs HTTP/1.1″ 304 – “-” “Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.0.11) Gecko/2009060214 Firefox/3.0.11″ “http://localhost:3001″

As evident from the last fragment of each log line, the load is distributed amongst three GlassFish Gem instances. More details on load balancer algorithm are available here.

Feel free to d
rop a comment on this blog if you are using GlassFish in production for your Rails applications. Several stories are already available at rubyonrails+glassfish+stories.

Technorati: glassfish rubyonrails apache httpd mod_proxy_balancer loadbalancing clustering

Related posts:

1. TOTD #81: How to use nginx to load balance a cluster of GlassFish Gem ?
2. Track your running miles using JRuby, Ruby-on-Rails, GlassFish, NetBeans, MySQL, and YUI Charts
3. TOTD #37: SQLite3 with Ruby-on-Rails on GlassFish Gem
4. TOTD #104: Popular Ruby-on-Rails applications on GlassFish v3 – Redmine, Typo, Substruct
5. TOTD #44: JDBC Connection Pooling for Rails on GlassFish v3