Lifecycle - Docker mode

The lifecycle of a Ponos agent in Docker mode is similar to that of the generic Ponos agent. Only the actions specific to task management differ.

Setup

In addition to the generic Ponos agent setup, the Ponos agent in Docker mode will:

Authenticate to Docker registries (using the docker.registries parameter of its configuration).
Log information about the host’s system (using the docker.fake_gpus parameter of its configuration).
List the tasks running on the host. The agent will list all Docker containers. For each Docker container, the agent will:
- Retrieve the task’s ID related to the Docker container using the container’s environments (which is of the form PONOS_TASK=<task_id>).
- Retrieve the task’s details (using the RetrieveTaskDefinition endpoint). If the task no longer exists or is assigned to another agent, the Docker container is ignored.
- Add the task to the list of running tasks.

Loop

For its loop, the Ponos agent in Docker mode uses the same lifecycle as the generic Ponos agent. But to know if the agent is ready, it will check the Docker socket. However, there are some specific points to note when the agent checks the running tasks:

To retrieve the task’s logs, the agent uses the logs related to the Docker container.
To find out the exit code, the agent retrieves the status code of the Docker container.

Start task

For a Ponos agent in Docker mode, starting a task means:

Download or pull the Docker image of the task (using the docker.check_local_cache parameter of its configuration).
Start the task in a Docker container (using the docker.shared_process_volume, docker.volumes, docker.network and docker.dns parameters of its configuration).

Stop task

For a Ponos agent in Docker mode, stopping a task means:

Delete the Docker container related to the task.