I am from the Oracle Fusion Applications Development team.
We are developing a new Application to be deployed and serviced on a WLS cluster.
Our application has a specific requirement: Only one request can be serviced on one managed server.
To satisfy this requirement, the application architecture is: If we need to service 'N' no. requests in parallel, then we need 'N' managed servers.
All these servers are made a part of a particular WLS cluster.
Generally other applications use either OHS or WLS cluster load balancing to redirect the requests to server within a cluster.
Since our architecture is slightly different, we have a custom logic to select which request will be redirected to which managed server within the cluster.
As with other Oracle Fusion Applications, OHS sits in between the client and the WLS cluster.
As part of our custom logic, we select the name of the managed server on which the request would be serviced.
OHS entries (in OHS mod conf file), contain the host and port of the each managed server identified by the managed server name.
The user flow:
The request is sent to another managed server even when WSLD URL clearly points to correct OHS entries to one particular managed server.
This does not happen when the servers are not a part of Weblogic cluster.
(When servers are not added to any cluster, the request is serviced on requested managed server.
But when servers are a part of WLS cluster, the request is redirected to some other server in the cluster).
So we suspect that some setting at WLS cluster level that is causing this re-direction of request.
The current cluster configuration has Default Load Algorithm as "Round Robin".
But I read that load balancing will not be performed when the requested URL contains the host:port of a particular managed server (which is our case).
I am reaching out to the Weblogic Clustering experts to figure out what we are missing in terms of the weblogic cluster set-up.