External Patch Server Deployment
A patch server can run inside the Kubernetes cluster, or outside on an external server host. When deployed outside of Kubernetes, the patch server is delivered as a child process of ICE Agent. Install ICE Agent on Linux, Windows, macOS or Docker, and start the patch server using the agent external patch command (consult the product guide for details).
The agent connects to the Server Bridge component in ICE Server via a mechanism similar to the one used by ICE Mobile and ICE Desktop clients. Once the patch agent registers itself with the ICE Server it begins receiving patch configuration messages intended for this patch server instance. The agent is responsible for monitoring configuration changes made on the ICE Server and updating the configuration of the bridging component accordingly. The two processes communicate with each other through configuration and status files written to a common configuration directory:

The agent writes
conf/ice_engagebridge_bridge.json with a set of patches (and their constituent channel configurations) and
conf/ice_engagebridge_config.json with process-level configuration.

The bridge writes
status/ice_engagebridge_status.json with the patch server's operational report (viewable in human readable form in ICE Desktop 'Patch Servers' settings).
The bridging component is designed to listen for changes to ice_engagebridge_bridge.json and ice_engagebridge_config.json and react accordingly. Similarly, the agent component listens to status changes written to ice_engagebridge_status.json and publishes them to ICE Server where they can be viewed in ICE Desktop. Changes to configuration written by the agent do not produce an outage on existing patches. That is, an administrator is free to create, delete or update patches without concern for disrupting other patches active on the system.