1 of 1 people found this helpful
You should provide what application do you will run. Is it a EJB or a WEB application? How do you access it?
For web applications you might use external load balancer and/or mod_*** (I'm not very familiar with this may other have an answer for you)
For EJB there is no need for loadbalances, the JBoss-proxy will have this functionality
Benefit of running two instances on one single machine is that you have two different Java VM's, in case of 32bit VMs you might able to use more memory.
Also you spread some load accross two VMs, it depend to the application and load whether this will make sense, you should test it!
In case of JVM crash or planed restarts it will avoid a complete downtime.
I think you will not find a cook book in detail, it depend on your requirements.
If you have many single http requests a lot of independent (non clustered) server may best
If you need failover seesions a cluster will be the solution.
I've seen many different solutions that will work for exact one scenario.
If you describe a bit more the community may have a better idea and can give you some hints.
Thanks for shedding some light on the matter. It's will be running a web application. The user will access
it through a web browser pointing to a url / hostname. Basically it will running a web based application.
The user / client will be pointing to a url / hostname where the haproxy will load balance it between two
server running JBoss EAP. The HAProxy will also be providing a failover services if one of the JBoss
application server is down or doing maintenance.
The HAProxy loadbalance server will be deployed in two server to avoid a single point of failure at
the loadbalancer. This will running heartbeat (Linux-HA), in order to provide a single ip address / virtual
hostname for the JBoss apps server.
To the user / client, they see / connect to one server, but at the back there is two JBoss server , serving
the request. sharing session (I guess this will be http session replication), sticky session and application
1. If there is no external load balancer or an external proxy in front of a two server (node) JBoss cluster, how the user will access the
JBoss web application. Said that I have two server running JBoss for a web application access, to configure JBoss built in
cluster, I should use one of the mod_* (either proxy,jk,cluster) running on the web server on both of the node.
Surely there will be two server name and two ip addresses, but the end user will use only 1 hostname / ip address to
access the JBoss web application, do this cluster plugin / mod maintain some sort of server name / ip address take over
and keepalive (heartbeat) to make sure that the hostname / ipaddress that the user access will always be alive in case of
one of the server down ?
Thanks in advance for any information provided.
I'm able to run two different nodes in same system and its working fine.i'm using below commands to run different nodes
run.sh node1 -Djboss.service.binding.set=ports-01 -u 22.214.171.124 -c all -b 0.0.0.0 -Djboss.messaging.ServerPeerID=1
run.sh node2 -Djboss.service.binding.set=
ports-default-u 126.96.36.199 -c all -b 0.0.0.0 -Djboss.messaging.ServerPeerID=2
I have tested sending request its working fine if and only if change the port number in URL.
Manually I'm changing port numbers.Please let me know any another approach is there.
Thanks in advance.
first of all your post is different to the original post, in that case you should open a new thread for your questions.
To answer your questions, a common way is to use the -b option with different (virtual) IP addresses. in that case you can drop the 'jboss.service.binding.set' option all instances work with the same port numbers but a different IP.
It depend on the OS how to configure the IP's.
If you have additional questions here please open a new thread.