Sriram,

If you would like some google indexed webspace to post this howto, let
me know.

Zack

Sriram Narayanan wrote:
> Hi all:
>
> I just got this working, so I thought of posting a small howto for the
> list :)
>
> ================================================
>
> An apache instance (httpd) that serves http://localhost/examples/
> where /examples is served by two Tomcat instances (if one is
> down/unavailable, then httpd decides to redirect requests to the
> second tomcat instances.
>
> What I've got working so far:
> - Session replication between the various Tomcat instances in the
> cluster.
> - In case one instance is not available, then apache sends new
> requests to the next instance.
> - When and instance comes back, it joins the cluster, receives session
> replication, and is ready to take over when required.
>
> What else is possible:
> - Load balance based on the number of requests already sent to a
> Tomcat instance
> - Configure timeouts, etc
> - More information at
> http://tomcat.apache.org/connectors-doc/reference/workers.html
>
> What's not possible:
> - If the first instance dies, then the second instance carries on
> processing from that point.
>
> To try this yourself:
> 1. Download and install Apache Tomcat 6.0.9
>
> 2. In server.xml, uncomment the tag for clustering
>
> 3. Edit the webapps/example/WEB-INF/web.xml to include the
> distributable tag
>   <distributable/>
>
> 4. Create a copy of this Tomcat folder (say, Tomcat2)
>
> 5. Edit the server.xml in Tomcat2. Change all the port numbers. e.g.
> change the AJP port number to 8010
>
> 6. Download the file
> http://mirrors.24-7-solutions.net/pub/apache/tomcat/tomcat-connectors/jk/binaries/win32/jk-1.2.20/mod_jk-apache-2.2.3.so
>
> and save it as mod_jk.so at apache/modules folder.
>
> 7. Place the following contents in the file apache/conf/worker.properties
>   # The advanced router LB worker
>     worker.list=router,worker2,worker1
>  
>     # Define a worker using ajp13
>     worker.worker1.port=8009
>     worker.worker1.host=localhost
>     worker.worker1.type=ajp13
>     worker.worker1.lbfactor=1
>     # Define prefered failover node for worker1
>     worker.worker1.redirect=worker2
>  
>     # Define another worker using ajp13
>     worker.worker2.port=8010
>     worker.worker2.host=localhost
>     worker.worker2.type=ajp13
>     worker.worker2.lbfactor=1
>     worker.worker2.mount=/examples/*
>     # Disable worker2 for all requests except failover
>     worker.worker2.activation=disabled
>  
>     # Define the LB worker
>     worker.router.type=lb
>     worker.router.balance_workers=worker1,worker2
>  
>     worker.list=jkstatus
>     # Define a 'jkstatus' worker using status
> worker.jkstatus.type=status
>
> 8. Edit apache/conf/httpd.conf to include the following:
> LoadModule   jk_module modules/mod_jk.so
>
> #AddModule   mod_jk.c
>
> JkWorkersFile conf/worker.properties
>
> JkShmFile   mod_jk.shm
> JkLogFile   mod_jk.log
>
> JkLogLevel   info
> JkLogStampFormat "[%a %b %d %H:%M:%S %Y] "
>
> JkMount /jkmanager/* jkstatus
>
> JkMount /examples/* router
>
> 9. Start the two Tomcat instances
>   You should see some replication related messages in the console
>
>   Feb 13, 2007 8:01:34 PM
> org.apache.catalina.ha.tcp.SimpleTcpCluster memberAdded
>   INFO: Replication member
> added:org.apache.catalina.tribes.membership.MemberImpl[
>   tcp://Ramn.corporate.thoughtworks.com:4001,Ramn.corporate.thoughtworks.com,4001,
>
>    alive=1016,id={99 34 -19 82 10 -107 64 51 -119 77 9 -117 -91 9 24 -9
> }, payload
> ={}, command={}, domain={}, ]
>
> INFO: Manager [localhost#/examples]; session state send at 2/13/07
> 8:01 PM recei
> ved in 125 ms.
>
>
> 10. Restart the httpd web server
>
> Testing the session replication
> 1. Browse to http://localhost/examples/
>
> 2. Try the Session examples for the servlet.
>
> 3. Set a variable into the session with a value
>
> 4. Stop the first Tomcat instance
>
> 5. Set another variable into the session.
>   This time, the second Tomcat instance will serve the content
>   You'll see that the first variable is still in the session. This is a
> sign that replication has been successful.
>
> 6. Restart the first instance, wait for it to enter the cluster and
> for the replication to happen
> INFO: Manager [localhost#/examples]; session state send at 2/13/07
> 8:01 PM recei
> ved in 125 ms.
>
> 7. Stop the second Tomcat instance.
>
> 8. Set another variable into the session.
>
> 9. You'll see that all the variables are still available.
>
> -- Sriram
>
> ---------------------------------------------------------------------
> To start a new topic, e-mail: users@(protected)
> To unsubscribe, e-mail: users-unsubscribe@(protected)
> For additional commands, e-mail: users-help@(protected)
>
>


---------------------------------------------------------------------
To start a new topic, e-mail: users@(protected)
To unsubscribe, e-mail: users-unsubscribe@(protected)
For additional commands, e-mail: users-help@(protected)