When running the IDM configuration wizard we are required to enter the load balancer URL but load balancer is not required for clustered installation. Thus URL for oim_server1 can be entered in the wizard and configuration be completed.
Now if in the clustered installation there are two oim severs (oim_server1, oim_server2 [oim_cluster]) and two soa servers (soa_server1, soa_server2 [soa_cluster]), there would be issues in approval flows if
oim_sever1 is shutdown (leaving oim_server2 and both the soa server running). The approvals would fail and in the BPEL fault we will see that the Approval Callback is trying to send the response back to oim_server1 rather then oim_server2. If we start back oim_server1 than the approvals along with call backs work fine.
Another scenarios which can be tried is to shut down oim_server2 instead of oim_server1 & than the approvals would works fine.
Now if in the clustered installation there are two oim severs (oim_server1, oim_server2 [oim_cluster]) and two soa servers (soa_server1, soa_server2 [soa_cluster]), there would be issues in approval flows if
oim_sever1 is shutdown (leaving oim_server2 and both the soa server running). The approvals would fail and in the BPEL fault we will see that the Approval Callback is trying to send the response back to oim_server1 rather then oim_server2. If we start back oim_server1 than the approvals along with call backs work fine.
Another scenarios which can be tried is to shut down oim_server2 instead of oim_server1 & than the approvals would works fine.
With the above test we can see that configuration is pointing the oim server to oim_server1 host rather than a cluster. So how to configure the domain post installation so that SOA would point to both the oim servers and would automatic failover? This can be done by changing the oimFrontEndUrl in the oim-config.xml. This value can be changed to point to any proxy server (say OHS, Apache) and use the mod so plugin at the proxy and redirect requests from proxy to OIM cluster.