R12.7.0.1 Access Gateway connecting to Policy Servers in the wrong Data Center.

Document ID : KB000124208
Last Modified Date : 08/01/2019
Show Technical Document Details
Introduction:
We have two VirtualHosts, each with a separate HCO, and are seeing the R12.7.0.1 Access Gateway make connections to the incorrect Data Center.

Following is Cluster #1 from the HCO; 

[31987/1708824320][Thu Dec 13 2018 01:56:52] cluster_1='4'. 
[31987/1708824320][Thu Dec 13 2018 01:56:52] cluster_1='138.83.24.79,44443'. 
[31987/1708824320][Thu Dec 13 2018 01:56:52] cluster_1='138.83.24.80,44443'. 
[31987/1708824320][Thu Dec 13 2018 01:56:52] cluster_1='138.83.24.81,44443'. 
[31987/1708824320][Thu Dec 13 2018 01:56:52] cluster_1='138.83.24.85,44443'. 
[31987/1708824320][Thu Dec 13 2018 01:56:52] cluster_1='50'. 

However with the AGENT_CON_MANAGER component selected, we can see the Agent is "Selecting" Policy Servers in the wrong Data Center, which are not even listed in the Agent's HCO; 

[12/13/2018][01:57:02.281][31325][140480931165952][][SmClient.cpp:3054][GetServer][][][][][][Selected server 138.83.89.84: Current total capacity: 99.0, current throughput: 1.00]
[12/13/2018][01:57:03.505][31325][140480931165952][][SmClient.cpp:3054][GetServer][][][][][][Selected server 138.83.89.81: Current total capacity: 96.0, current throughput: 4.00]
Question:
Why do we see each VirtualHost logging the "Selected server" messages for Policy Servers not listed in the VirtualHost's HCO?
Environment:
R12.7.0.1 Access Gateway on RedHat 7
Answer:

The Access GateWay creates two LLAWP processes for two virtual hosts and uses a single java process.

The Java process internally runs a thread with the java process ID, which periodically tries to 'ping' all policy servers to check connection details and creates a best match index. Only this health check related log message is logged in both VirtualHost logs.

Each VirtualHost will only connect to Policy Servers from it's HCO to process requests based on this "best match index".