JBoss Clustering for High Availability

Hi Leighton

Please refer to the documentation to see how to implement the high availability, keep in mind you will need a shared folder for certain folders. When using http server you can configure sticky sessions.

Documentation:
docs.appian.com/.../Configuring_Apache_Web_Server_with_JBoss.html
docs.appian.com/.../High_Availability_and_Distributed_Installations.html
community.appian.com/.../setting-up-for-disaster-recovery

some KB Links
community.appian.com/.../kb-1072-requests-within-the-same-session-go-to-a-different-application-server

and some previous discussions:
community.appian.com/.../hi-i-am-trying-to-design-high-availability-for-appian-platform-currently-i-hav

The high availability of the engines in configured in the topology which is the core of Appian,

As I said before, you have to share folders, these shared folders includes the Engine data, so all the information from your process execution will be shared and replicated.
maybe you want to make sure this shared folder is well backed-up in a some RAID array or something that ensures you always have the last version of the file.

From the application server perspective replicating sessions is bandwidth and memory consuming and this will limit your concurrent users, even if you have 10 servers parallel all of them will be using mostly the same amount of memory and they will be talking to each other. but there are some scenarios when this is really needed.
www.mastertheboss.com/.../jboss-http-session-replication
serverfault.com/.../pros-cons-of-session-replication

After all, I want to ask you, is the application that critical so you really need to replicate the sessions, instead of just configuring with the engines shared folders?

I think session replication makes more sense when implementing JSF/JSP/Servlets or any framework and you don't have storage to the DB between pages for example. But Appian has In-Memory DB which keeps track of all the values every time you move in the process (affecting the process variables).
docs.appian.com/.../Process_History.html

With all this said I don't want to discourage you to configure or to test it, I am just sharing my opinion. :)

Please share your decision, your test results and/or your final solution, So, someone else can refer to this in the future.

Best regards
Jose Perez

When you setup clustering of jboss in appian...appian internally understands how many clusters it has and it will appropriately channel the load to each jboss server to provide smooth traffic flow..it is intelligent itself to do that

To clarify, do you want to persist sessions for use across multiple JBoss nodes? If that is the case, Appian does not support this configuration. Each JBoss app server runs in standalone mode and is not aware of any other app servers in the cluster. Sessions are sticky and if one of the nodes goes down, the session data will be lost and the user needs to re-login.

Hi Mike Cichy ,

I have raised a question regarding this, would you please advice me what to do with my case?
community.appian.com/.../configure-high-availability-in-linux-env

Thanks!