Cache agents start first It is important that cache agents start before inference agents. Set the Cache Agent Quorum setting in the Object Management Configuration pane to ensure that enough cache agents start before other types of agents can begin processing. In general, at startup, you start cache agents then other types of agents. At system shutdown, you stop other kinds of agents, then stop cache agents.
If you are working in a copy of the examples that preserves the relationship between the
readme.html file and its resources (see
Dependency of Readme File on a Higher Level Directory, you can go ahead and open the copied readme and follow its instructions (adapting the path to the CDD and EAR files as needed). Otherwise, just use the provided example files as follows:
Navigate to BE_HOME/examples/standard/FraudDetectionCache and open
readme.html in your browser. Follow the instructions in the readme file to start two engines and send events to the inference agent.
Start a second cache agent. Experiment with stopping one cache agent engine then restarting it again, to see how one cache agent can operate successfully. Because one backup copy of the data is kept, and there are two cache agents, each cache agent has the complete cache contents. In a production system with more cache agents no single cache agent has all the cache data.