15.13. Understanding Drift and JBoss ON Agents and Servers

Both the JBoss ON agent and the JBoss ON server maintain their own inventories of the resources, directories, and files monitored for drift. When the agent starts up, it compares its inventory with the server inventory.

The drift information is stored, with the other agent data, in the agentRoot/rhq-agent/data/ directory. The information in this directory is deleted if the agent is started with new configuration (--cleanconfig) or it can be intentionally purged (--purgedata). If the drift information is lost, then the agent requests the last snapshot from the JBoss ON server.

The agent always sends the latest changeset to the server as a snapshot. If the server is offline for some period and misses updates, then the agent sends the most current snapshot, which effectively rolls all changes into one snapshot, even if the changes accumulated over several drift detection runs.

The server processes drift changes through a server-side plug-in. This plug-in must be enabled for the server to recognize and process drift data sent from the agent.

As with other server-side plug-ins, the drift plug-in can be disabled. However, this effectively and entirely disables drift monitoring on that server, and no drift information is processed or stored. That is slightly different than the behavior of other server subsystems. For example, an individual alert sender can be disabled, but alert detections are still run and alert information is still processed, stored, and displayed by the JBoss ON server.