prefect.runtime.deployment

Access attributes of the current deployment run dynamically.

Note that if a deployment is not currently being run, all attributes will return empty values.

You can mock the runtime attributes for testing purposes by setting environment variables prefixed with PREFECT__RUNTIME__DEPLOYMENT.

Example usage

from prefect.runtime import deployment

def get_task_runner():
    task_runner_config = deployment.parameters.get("runner_config", "default config here")
    return DummyTaskRunner(task_runner_specs=task_runner_config)

Available attributes

• id: the deployment's unique ID
• name: the deployment's name
• version: the deployment's version
• flow_run_id: the current flow run ID for this deployment
• parameters: the parameters that were passed to this run; note that these do not necessarily include default values set on the flow function, only the parameter values set on the deployment object or those directly provided via API for this run

__getattr__(name)

Attribute accessor for this submodule; note that imports also work with this:

from prefect.runtime.flow_run import id

Source code in src/prefect/runtime/deployment.py

def __getattr__(name: str) -> Any:
    """
    Attribute accessor for this submodule; note that imports also work with this:

        from prefect.runtime.flow_run import id
    """

    func = FIELDS.get(name)

    # if `name` is an attribute but it is mocked through environment variable, the mocked type will be str,
    # which might be different from original one. For consistency, cast env var to the same type
    env_key = f"PREFECT__RUNTIME__DEPLOYMENT__{name.upper()}"

    if func is None:
        if env_key in os.environ:
            return os.environ[env_key]
        else:
            raise AttributeError(f"{__name__} has no attribute {name!r}")

    real_value = func()
    if env_key in os.environ:
        mocked_value = os.environ[env_key]
        # cast `mocked_value` to the same type as `real_value`
        try:
            cast_func = type_cast[type(real_value)]
            return cast_func(mocked_value)
        except KeyError:
            raise ValueError(
                "This runtime context attribute cannot be mocked using an"
                " environment variable. Please use monkeypatch instead."
            )
    else:
        return real_value