If you use EJB's it will be simple, just mark your Beans as Clustered, JBoss will do (most) of the rest if you use the 'all' profile.
If you talk about http stuff you will need a loadbalancer in front of the JBoss cluster (i.e. Apache) for load-balancing and failover.
So you should provide a bit more information that we are able to help.
Thank you for your reply. I have a ".ear", and I am instructed to perform clustering in Jboss 4.2 configure failover, the .ear comprises of ejb's but I feel that making changes in EJB's will break other code. So I want to configure failover, can you please let me know series of steps how to achieve.
As I said, I need to know the application structure, an EAR might contain everything.
What do you use EJB/Servlets/... and what is the customers view, mean what is accessable from outside?
I suppose the application is J2EE with EJB2.x, right?
Yes wolf, application is a J2EE with EJB3, Mule1.4.3, Seam2.2.1, , all that I am in needto do is to create an environment which support a fail over, that is while any user is using the application and for some reason if the application is down, then the environment I created must support that situation and end user should never know that node is down.
So how is the application accessed by client?
- EJB RMI calls
Where do you need failover?
If you use only SLSB's you don't need a replication of states.
This is a java application for the front we use Richfaces, for Businsess layer we use EJB and MULE, and back end MS SQL, and client use through 'http". I need failover at the server level.
For HTTP you need a loadbalancer, i.e. apache, between client and the JBoss cluster.
Stateless Beans need to mark clustered if it will be accessable from remote
Statefull Beans need to mark clustered to replicate the state for failover.
You might mark it with @Clustered by annotation or by deployment descriptor.
The replication is activated if you start a profile with the cluster componentes, you should use 'all' as example.
Hello Wolf, thanks for your reply. I have few thoughts, which i want to discuss on the same. Can you please share me your number I would be calling. My number is +91 8106 865 165. or kindly mail your number to email@example.com and please let me knw preferable timings to call you. your help is highly appreciated.
Hello Wolf, Let me give you a brief explanation of what all i am looking for.
1) I have an .EAR, which have EJB3.0, Mule, myfaces as front end in it.
2) I am using 4.2.3 Jboss server.
3) I have expected to create a cluster environment and deploy EAR. Now the task is my client uses http, to access my application deployed.
4) The end user should never know if any of the server is down in the cluster.
Please suggest me the series of steps in achieveing the above, I am speaking much about architectural perspective.
I am a new bee, kindly bare with me if i am troubling you.
As this is a voluntary based forum and such contact will be more a consulting.
If you need such professional support you may contact the Red Hat sales and ask for a consultant or buy a subscription.
As you use a old version of JBoss you will find documentations and community threads for your questions.
Also the cluster documentation will be a good starting point.
If I'm right you need to have a http loadbalancer like Apache or a hardware in front of JBoss. Your customer will only see one public address.
Depend on the environment (EJBs web pages ...) and the recommendations you might have also a JBoss cluster as backend.
You will find some high level documents how such application can be operated