Multicast based Clustering With deployment Systematization

In current competitive enterprise environment, integration is playing very important role. Using integration systems are not going to improve the environment, Unless Having right pattern and solution. Having said that, making integration platform in a high availability manner is make more meaningful to scale your enterprise.

In this blog, I am explaining how to make the Carbon 4.2.0 based products in high available manner and how to synchronize the deployment a cross the same products. Furthermore, in this example I have taken WSO2 DSS as product to experiment this feature. You can download this in official website of the WSO2 for free of charge!

Lets have look at the illustrated deployment pattern, I have deployed DSS in 3 diffident machines(Nodes). Important to note is, i have not create Master/Salve setup. Each and every node are in peer to peer way.  All the servers are load balanced by external hardware based "F5 Load Balancer  and inside the Demilitarized Zone(DMZ).

Deployment Diagram

Lets take a step into configuration to understand, How to implement above scenario! 

Step 1 - Downloaded a product into any one of the box and started and shutdown to testing purpose.

Step 2 - Since I have 3 different machine to configure and i wont to use the my domain name instant of localhost/127.0.0.1!!! There for  i used a simple script to replace the values with in "conf" directory.

example - find ./ -type f -exec sed -i 's/localhost/box1.vanji.com/g' {} \;

Step 3 - Configure deployment synchronization in  /repository/conf/carbon.xml

 To understand further on SVN based Deployment synchronization, refer  [1]

Step 4 - Configure registry to maintain the metadata across the all three nodes. Refer further on [2]

Step 5 - Configure User Store to cater external user store. Refer further on [3]

Step 6 - Configure /repository/conf/axis2/axis2.xml to cluster the system.
This step is very important to understand! We are enabling the clustering in multicast mode! Let see the config .

• In Line 2, I have enabled the clustering
• In Line 5, I have mentioned "membershipScheme" as multicast.
• In line 10, I have mentioned "localMemberHost" as name of my current machine "box1.vanji.com"
• And more importantly, i have mentioned my other members detail (Line 20)

Note
I have removed the comment in the configuration to compact the config file!


Step 7 - Just Start the server and observe is there any expedition or error due to the configuration. If there is anything resolve the issues. If there is no miss configuration, You will see the carbon log as above.

 Server is waiting for other nodes to start to make the clustering.
So now stop this sever for while and move to step 8.

Step 8 -  Take the copy of this server and copy to other servers and run the find and replace script to change the name.
Example:- find ./ -type f -exec sed -i 's/box1.vanji.com/box2.vanji.com/g' {} \;

Step 9 - Go to /repository/conf/axis2/axis2.xml and make sure Step 6 followed or not in each and every nodes and make sure Members are defined properly!

Step 10 - Time to Start the servers and enjoy the High Available WSO2 Carbon 4.2.0 based server.

[1] http://docs.wso2.org/display/DSS300/Configuring+the+SVN-Based+Deployment+Synchronizer

[2] http://docs.wso2.org/display/ESB470/Governance+Partition+in+a+Remote+Registry

[3] http://docs.wso2.org/display/DSS310/Configuring+External+User+Stores