This project simulates the flow of an assisted installer agent and allows easy troubleshooting of commands:

1. agent is an executable that starts a command-runner. It must run with sudo on the host.
2. command-runner is an executable packaged as a container that periodically polls a command-server and executes the command received from it via nsenter. The command can be another Podman command.
3. command-server is an HTTP server that loads commands from a JSON file passed via --commands-file, and loops through them returning one command at a time.

make build

1. Update commands.json with the commands you want to run.
2. Start all the three components with make run, or POLLING_INTERVAL=7s make run to change the command polling interval (in Go duration format).

Keep in mind that the agent runs in the foreground, so you will have to open another terminal window to interact with the command-runner and command-server.

When using make run, the server will run as a container.

You can edit commands.json and make the system pick up the changes:

• If you run the command-server manually as an executable on the host (not as a container).
• Using make edit target. In this case the server container will be restarted as soon as you exit the editor.

View the command execution log using sudo podman logs agent-simulator-command-runner.

Run make stop to stop and remove the containers. The agent will exit automatically as soon as the command-runner terminates.

Keep in mind that commands that were sent to the agent and are still running will not be terminated.

This project requires Podman. Make sure you do not have Docker installed on the same machine.