Skip to content

prefect_aws.ecs

DEPRECATION WARNING:

This module is deprecated as of March 2024 and will not be available after September 2024. It has been replaced by the ECS worker, which offers enhanced functionality and better performance.

For upgrade instructions, see https://docs.prefect.io/latest/guides/upgrade-guide-agents-to-workers/.

Integrations with the Amazon Elastic Container Service.

Examples:

Run a task using ECS Fargate

ECSTask(command=["echo", "hello world"]).run()

Run a task using ECS Fargate with a spot container instance

ECSTask(command=["echo", "hello world"], launch_type="FARGATE_SPOT").run()

Run a task using ECS with an EC2 container instance

ECSTask(command=["echo", "hello world"], launch_type="EC2").run()

Run a task on a specific VPC using ECS Fargate

ECSTask(command=["echo", "hello world"], vpc_id="vpc-01abcdf123456789a").run()

Run a task and stream the container's output to the local terminal. Note an execution role must be provided with permissions: logs:CreateLogStream, logs:CreateLogGroup, and logs:PutLogEvents.

ECSTask(
    command=["echo", "hello world"],
    stream_output=True,
    execution_role_arn="..."
)

Run a task using an existing task definition as a base

ECSTask(command=["echo", "hello world"], task_definition_arn="arn:aws:ecs:...")

Run a task with a specific image

ECSTask(command=["echo", "hello world"], image="alpine:latest")

Run a task with custom memory and CPU requirements

ECSTask(command=["echo", "hello world"], memory=4096, cpu=2048)

Run a task with custom environment variables

ECSTask(command=["echo", "hello $PLANET"], env={"PLANET": "earth"})

Run a task in a specific ECS cluster

ECSTask(command=["echo", "hello world"], cluster="my-cluster-name")

Run a task with custom VPC subnets

ECSTask(
    command=["echo", "hello world"],
    task_customizations=[
        {
            "op": "add",
            "path": "/networkConfiguration/awsvpcConfiguration/subnets",
            "value": ["subnet-80b6fbcd", "subnet-42a6fdgd"],
        },
    ]
)

Run a task without a public IP assigned

ECSTask(
    command=["echo", "hello world"],
    vpc_id="vpc-01abcdf123456789a",
    task_customizations=[
        {
            "op": "replace",
            "path": "/networkConfiguration/awsvpcConfiguration/assignPublicIp",
            "value": "DISABLED",
        },
    ]
)

Run a task with custom VPC security groups

ECSTask(
    command=["echo", "hello world"],
    vpc_id="vpc-01abcdf123456789a",
    task_customizations=[
        {
            "op": "add",
            "path": "/networkConfiguration/awsvpcConfiguration/securityGroups",
            "value": ["sg-d72e9599956a084f5"],
        },
    ],
)

Classes

ECSTask (Infrastructure) pydantic-model

Run a command as an ECS task.

Attributes:

Name Type Description
type Literal['ecs-task']

The slug for this task type with a default value of "ecs-task".

aws_credentials AwsCredentials

The AWS credentials to use to connect to ECS with a default factory of AwsCredentials.

task_definition_arn Optional[str]

An optional identifier for an existing task definition to use. If fields are set on the ECSTask that conflict with the task definition, a new copy will be registered with the required values. Cannot be used with task_definition. If not provided, Prefect will generate and register a minimal task definition.

task_definition Optional[dict]

An optional ECS task definition to use. Prefect may set defaults or override fields on this task definition to match other ECSTask fields. Cannot be used with task_definition_arn. If not provided, Prefect will generate and register a minimal task definition.

family Optional[str]

An optional family for the task definition. If not provided, it will be inferred from the task definition. If the task definition does not have a family, the name will be generated. When flow and deployment metadata is available, the generated name will include their names. Values for this field will be slugified to match AWS character requirements.

image Optional[str]

An optional image to use for the Prefect container in the task. If this value is not null, it will override the value in the task definition. This value defaults to a Prefect base image matching your local versions.

auto_deregister_task_definition bool

A boolean that controls if any task definitions that are created by this block will be deregistered or not. Existing task definitions linked by ARN will never be deregistered. Deregistering a task definition does not remove it from your AWS account, instead it will be marked as INACTIVE.

cpu int

The amount of CPU to provide to the ECS task. Valid amounts are specified in the AWS documentation. If not provided, a default value of ECS_DEFAULT_CPU will be used unless present on the task definition.

memory int

The amount of memory to provide to the ECS task. Valid amounts are specified in the AWS documentation. If not provided, a default value of ECS_DEFAULT_MEMORY will be used unless present on the task definition.

execution_role_arn str

An execution role to use for the task. This controls the permissions of the task when it is launching. If this value is not null, it will override the value in the task definition. An execution role must be provided to capture logs from the container.

configure_cloudwatch_logs bool

A boolean that controls if the Prefect container will be configured to send its output to the AWS CloudWatch logs service or not. This functionality requires an execution role with permissions to create log streams and groups.

cloudwatch_logs_options Dict[str, str]

A dictionary of options to pass to the CloudWatch logs configuration.

stream_output bool

A boolean indicating whether logs will be streamed from the Prefect container to the local console.

launch_type Optional[Literal['FARGATE', 'EC2', 'EXTERNAL', 'FARGATE_SPOT']]

An optional launch type for the ECS task run infrastructure.

vpc_id Optional[str]

An optional VPC ID to link the task run to. This is only applicable when using the 'awsvpc' network mode for your task.

cluster Optional[str]

An optional ECS cluster to run the task in. The ARN or name may be provided. If not provided, the default cluster will be used.

env Dict[str, Optional[str]]

A dictionary of environment variables to provide to the task run. These variables are set on the Prefect container at task runtime.

task_role_arn str

An optional role to attach to the task run. This controls the permissions of the task while it is running.

task_customizations JsonPatch

A list of JSON 6902 patches to apply to the task run request. If a string is given, it will parsed as a JSON expression.

task_start_timeout_seconds int

The amount of time to watch for the start of the ECS task before marking it as failed. The task must enter a RUNNING state to be considered started.

task_watch_poll_interval float

The amount of time to wait between AWS API calls while monitoring the state of an ECS task.

Source code in prefect_aws/ecs.py
@deprecated_class(
    start_date="Mar 2024",
    help=(
        "Use the ECS worker instead."
        " Refer to the upgrade guide for more information:"
        " https://docs.prefect.io/latest/guides/upgrade-guide-agents-to-workers/."
    ),
)
class ECSTask(Infrastructure):
    """
    Run a command as an ECS task.

    Attributes:
        type: The slug for this task type with a default value of "ecs-task".
        aws_credentials: The AWS credentials to use to connect to ECS with a
            default factory of AwsCredentials.
        task_definition_arn: An optional identifier for an existing task definition
            to use. If fields are set on the ECSTask that conflict with the task
            definition, a new copy will be registered with the required values.
            Cannot be used with task_definition. If not provided, Prefect will
            generate and register a minimal task definition.
        task_definition: An optional ECS task definition to use. Prefect may set
            defaults or override fields on this task definition to match other
            ECSTask fields. Cannot be used with task_definition_arn.
            If not provided, Prefect will generate and register
            a minimal task definition.
        family: An optional family for the task definition. If not provided,
            it will be inferred from the task definition. If the task definition
            does not have a family, the name will be generated. When flow and
            deployment metadata is available, the generated name will include
            their names. Values for this field will be slugified to match
            AWS character requirements.
        image: An optional image to use for the Prefect container in the task.
            If this value is not null, it will override the value in the task
            definition. This value defaults to a Prefect base image matching
            your local versions.
        auto_deregister_task_definition: A boolean that controls if any task
            definitions that are created by this block will be deregistered
            or not. Existing task definitions linked by ARN will never be
            deregistered. Deregistering a task definition does not remove
            it from your AWS account, instead it will be marked as INACTIVE.
        cpu: The amount of CPU to provide to the ECS task. Valid amounts are
            specified in the AWS documentation. If not provided, a default
            value of ECS_DEFAULT_CPU will be used unless present on
            the task definition.
        memory: The amount of memory to provide to the ECS task.
            Valid amounts are specified in the AWS documentation.
            If not provided, a default value of ECS_DEFAULT_MEMORY
            will be used unless present on the task definition.
        execution_role_arn: An execution role to use for the task.
            This controls the permissions of the task when it is launching.
            If this value is not null, it will override the value in the task
            definition. An execution role must be provided to capture logs
            from the container.
        configure_cloudwatch_logs: A boolean that controls if the Prefect
            container will be configured to send its output to the
            AWS CloudWatch logs service or not. This functionality requires
            an execution role with permissions to create log streams and groups.
        cloudwatch_logs_options: A dictionary of options to pass to
            the CloudWatch logs configuration.
        stream_output: A boolean indicating whether logs will be
            streamed from the Prefect container to the local console.
        launch_type: An optional launch type for the ECS task run infrastructure.
        vpc_id: An optional VPC ID to link the task run to.
            This is only applicable when using the 'awsvpc' network mode for your task.
        cluster: An optional ECS cluster to run the task in.
            The ARN or name may be provided. If not provided,
            the default cluster will be used.
        env: A dictionary of environment variables to provide to
            the task run. These variables are set on the
            Prefect container at task runtime.
        task_role_arn: An optional role to attach to the task run.
            This controls the permissions of the task while it is running.
        task_customizations: A list of JSON 6902 patches to apply to the task
            run request. If a string is given, it will parsed as a JSON expression.
        task_start_timeout_seconds: The amount of time to watch for the
            start of the ECS task before marking it as failed. The task must
            enter a RUNNING state to be considered started.
        task_watch_poll_interval: The amount of time to wait between AWS API
            calls while monitoring the state of an ECS task.
    """

    _block_type_slug = "ecs-task"
    _block_type_name = "ECS Task"
    _logo_url = "https://cdn.sanity.io/images/3ugk85nk/production/d74b16fe84ce626345adf235a47008fea2869a60-225x225.png"  # noqa
    _description = "Run a command as an ECS task."  # noqa
    _documentation_url = (
        "https://prefecthq.github.io/prefect-aws/ecs/#prefect_aws.ecs.ECSTask"  # noqa
    )

    type: Literal["ecs-task"] = Field(
        "ecs-task", description="The slug for this task type."
    )

    aws_credentials: AwsCredentials = Field(
        title="AWS Credentials",
        default_factory=AwsCredentials,
        description="The AWS credentials to use to connect to ECS.",
    )

    # Task definition settings
    task_definition_arn: Optional[str] = Field(
        default=None,
        description=(
            "An identifier for an existing task definition to use. If fields are set "
            "on the `ECSTask` that conflict with the task definition, a new copy "
            "will be registered with the required values. "
            "Cannot be used with `task_definition`. If not provided, Prefect will "
            "generate and register a minimal task definition."
        ),
    )
    task_definition: Optional[dict] = Field(
        default=None,
        description=(
            "An ECS task definition to use. Prefect may set defaults or override "
            "fields on this task definition to match other `ECSTask` fields. "
            "Cannot be used with `task_definition_arn`. If not provided, Prefect will "
            "generate and register a minimal task definition."
        ),
    )
    family: Optional[str] = Field(
        default=None,
        description=(
            "A family for the task definition. If not provided, it will be inferred "
            "from the task definition. If the task definition does not have a family, "
            "the name will be generated. When flow and deployment metadata is "
            "available, the generated name will include their names. Values for this "
            "field will be slugified to match AWS character requirements."
        ),
    )
    image: Optional[str] = Field(
        default=None,
        description=(
            "The image to use for the Prefect container in the task. If this value is "
            "not null, it will override the value in the task definition. This value "
            "defaults to a Prefect base image matching your local versions."
        ),
    )
    auto_deregister_task_definition: bool = Field(
        default=True,
        description=(
            "If set, any task definitions that are created by this block will be "
            "deregistered. Existing task definitions linked by ARN will never be "
            "deregistered. Deregistering a task definition does not remove it from "
            "your AWS account, instead it will be marked as INACTIVE."
        ),
    )

    # Mixed task definition / run settings
    cpu: int = Field(
        title="CPU",
        default=None,
        description=(
            "The amount of CPU to provide to the ECS task. Valid amounts are "
            "specified in the AWS documentation. If not provided, a default value of "
            f"{ECS_DEFAULT_CPU} will be used unless present on the task definition."
        ),
    )
    memory: int = Field(
        default=None,
        description=(
            "The amount of memory to provide to the ECS task. Valid amounts are "
            "specified in the AWS documentation. If not provided, a default value of "
            f"{ECS_DEFAULT_MEMORY} will be used unless present on the task definition."
        ),
    )
    execution_role_arn: str = Field(
        title="Execution Role ARN",
        default=None,
        description=(
            "An execution role to use for the task. This controls the permissions of "
            "the task when it is launching. If this value is not null, it will "
            "override the value in the task definition. An execution role must be "
            "provided to capture logs from the container."
        ),
    )
    configure_cloudwatch_logs: bool = Field(
        default=None,
        description=(
            "If `True`, the Prefect container will be configured to send its output "
            "to the AWS CloudWatch logs service. This functionality requires an "
            "execution role with logs:CreateLogStream, logs:CreateLogGroup, and "
            "logs:PutLogEvents permissions. The default for this field is `False` "
            "unless `stream_output` is set."
        ),
    )
    cloudwatch_logs_options: Dict[str, str] = Field(
        default_factory=dict,
        description=(
            "When `configure_cloudwatch_logs` is enabled, this setting may be used to "
            "pass additional options to the CloudWatch logs configuration or override "
            "the default options. See the AWS documentation for available options. "
            "https://docs.aws.amazon.com/AmazonECS/latest/developerguide/using_awslogs.html#create_awslogs_logdriver_options."  # noqa
        ),
    )
    stream_output: bool = Field(
        default=None,
        description=(
            "If `True`, logs will be streamed from the Prefect container to the local "
            "console. Unless you have configured AWS CloudWatch logs manually on your "
            "task definition, this requires the same prerequisites outlined in "
            "`configure_cloudwatch_logs`."
        ),
    )

    # Task run settings
    launch_type: Optional[Literal["FARGATE", "EC2", "EXTERNAL", "FARGATE_SPOT"]] = (
        Field(
            default="FARGATE",
            description=(
                "The type of ECS task run infrastructure that should be used. Note that"
                " 'FARGATE_SPOT' is not a formal ECS launch type, but we will configure"
                " the proper capacity provider strategy if set here."
            ),
        )
    )
    vpc_id: Optional[str] = Field(
        title="VPC ID",
        default=None,
        description=(
            "The AWS VPC to link the task run to. This is only applicable when using "
            "the 'awsvpc' network mode for your task. FARGATE tasks require this "
            "network  mode, but for EC2 tasks the default network mode is 'bridge'. "
            "If using the 'awsvpc' network mode and this field is null, your default "
            "VPC will be used. If no default VPC can be found, the task run will fail."
        ),
    )
    cluster: Optional[str] = Field(
        default=None,
        description=(
            "The ECS cluster to run the task in. The ARN or name may be provided. If "
            "not provided, the default cluster will be used."
        ),
    )
    env: Dict[str, Optional[str]] = Field(
        title="Environment Variables",
        default_factory=dict,
        description=(
            "Environment variables to provide to the task run. These variables are set "
            "on the Prefect container at task runtime. These will not be set on the "
            "task definition."
        ),
    )
    task_role_arn: str = Field(
        title="Task Role ARN",
        default=None,
        description=(
            "A role to attach to the task run. This controls the permissions of the "
            "task while it is running."
        ),
    )
    task_customizations: JsonPatch = Field(
        default_factory=lambda: JsonPatch([]),
        description=(
            "A list of JSON 6902 patches to apply to the task run request. "
            "If a string is given, it will parsed as a JSON expression."
        ),
    )

    # Execution settings
    task_start_timeout_seconds: int = Field(
        default=120,
        description=(
            "The amount of time to watch for the start of the ECS task "
            "before marking it as failed. The task must enter a RUNNING state to be "
            "considered started."
        ),
    )
    task_watch_poll_interval: float = Field(
        default=5.0,
        description=(
            "The amount of time to wait between AWS API calls while monitoring the "
            "state of an ECS task."
        ),
    )

    @root_validator(pre=True)
    def set_default_configure_cloudwatch_logs(cls, values: dict) -> dict:
        """
        Streaming output generally requires CloudWatch logs to be configured.

        To avoid entangled arguments in the simple case, `configure_cloudwatch_logs`
        defaults to matching the value of `stream_output`.
        """
        configure_cloudwatch_logs = values.get("configure_cloudwatch_logs")
        if configure_cloudwatch_logs is None:
            values["configure_cloudwatch_logs"] = values.get("stream_output")
        return values

    @root_validator
    def configure_cloudwatch_logs_requires_execution_role_arn(
        cls, values: dict
    ) -> dict:
        """
        Enforces that an execution role arn is provided (or could be provided by a
        runtime task definition) when configuring logging.
        """
        if (
            values.get("configure_cloudwatch_logs")
            and not values.get("execution_role_arn")
            # Do not raise if they've linked to another task definition or provided
            # it without using our shortcuts
            and not values.get("task_definition_arn")
            and not (values.get("task_definition") or {}).get("executionRoleArn")
        ):
            raise ValueError(
                "An `execution_role_arn` must be provided to use "
                "`configure_cloudwatch_logs` or `stream_logs`."
            )
        return values

    @root_validator
    def cloudwatch_logs_options_requires_configure_cloudwatch_logs(
        cls, values: dict
    ) -> dict:
        """
        Enforces that an execution role arn is provided (or could be provided by a
        runtime task definition) when configuring logging.
        """
        if values.get("cloudwatch_logs_options") and not values.get(
            "configure_cloudwatch_logs"
        ):
            raise ValueError(
                "`configure_cloudwatch_log` must be enabled to use "
                "`cloudwatch_logs_options`."
            )
        return values

    @root_validator(pre=True)
    def image_is_required(cls, values: dict) -> dict:
        """
        Enforces that an image is available if image is `None`.
        """
        has_image = bool(values.get("image"))
        has_task_definition_arn = bool(values.get("task_definition_arn"))

        # The image can only be null when the task_definition_arn is set
        if has_image or has_task_definition_arn:
            return values

        prefect_container = (
            get_prefect_container(
                (values.get("task_definition") or {}).get("containerDefinitions", [])
            )
            or {}
        )
        image_in_task_definition = prefect_container.get("image")

        # If a task_definition is given with a prefect container image, use that value
        if image_in_task_definition:
            values["image"] = image_in_task_definition
        # Otherwise, it should default to the Prefect base image
        else:
            values["image"] = get_prefect_image_name()
        return values

    @validator("task_customizations", pre=True)
    def cast_customizations_to_a_json_patch(
        cls, value: Union[List[Dict], JsonPatch, str]
    ) -> JsonPatch:
        """
        Casts lists to JsonPatch instances.
        """
        if isinstance(value, str):
            value = json.loads(value)
        if isinstance(value, list):
            return JsonPatch(value)
        return value  # type: ignore

    class Config:
        """Configuration of pydantic."""

        # Support serialization of the 'JsonPatch' type
        arbitrary_types_allowed = True
        json_encoders = {JsonPatch: lambda p: p.patch}

    def dict(self, *args, **kwargs) -> Dict:
        """
        Convert to a dictionary.
        """
        # Support serialization of the 'JsonPatch' type
        d = super().dict(*args, **kwargs)
        d["task_customizations"] = self.task_customizations.patch
        return d

    def prepare_for_flow_run(
        self: Self,
        flow_run: "FlowRun",
        deployment: Optional["Deployment"] = None,
        flow: Optional["Flow"] = None,
    ) -> Self:
        """
        Return an copy of the block that is prepared to execute a flow run.
        """
        new_family = None

        # Update the family if not specified elsewhere
        if (
            not self.family
            and not self.task_definition_arn
            and not (self.task_definition and self.task_definition.get("family"))
        ):
            if flow and deployment:
                new_family = f"{ECS_DEFAULT_FAMILY}__{flow.name}__{deployment.name}"
            elif flow and not deployment:
                new_family = f"{ECS_DEFAULT_FAMILY}__{flow.name}"
            elif deployment and not flow:
                # This is a weird case and should not be see in the wild
                new_family = f"{ECS_DEFAULT_FAMILY}__unknown-flow__{deployment.name}"

        new = super().prepare_for_flow_run(flow_run, deployment=deployment, flow=flow)

        if new_family:
            return new.copy(update={"family": new_family})
        else:
            # Avoid an extra copy if not needed
            return new

    @sync_compatible
    async def run(self, task_status: Optional[TaskStatus] = None) -> ECSTaskResult:
        """
        Run the configured task on ECS.
        """
        boto_session, ecs_client = await run_sync_in_worker_thread(
            self._get_session_and_client
        )

        (
            task_arn,
            cluster_arn,
            task_definition,
            is_new_task_definition,
        ) = await run_sync_in_worker_thread(
            self._create_task_and_wait_for_start, boto_session, ecs_client
        )

        # Display a nice message indicating the command and image
        command = self.command or get_prefect_container(
            task_definition["containerDefinitions"]
        ).get("command", [])
        self.logger.info(
            f"{self._log_prefix}: Running command {' '.join(command)!r} "
            f"in container {PREFECT_ECS_CONTAINER_NAME!r} ({self.image})..."
        )

        # The task identifier is "{cluster}::{task}" where we use the configured cluster
        # if set to preserve matching by name rather than arn
        # Note "::" is used despite the Prefect standard being ":" because ARNs contain
        # single colons.
        identifier = (self.cluster if self.cluster else cluster_arn) + "::" + task_arn

        if task_status:
            task_status.started(identifier)

        status_code = await run_sync_in_worker_thread(
            self._watch_task_and_get_exit_code,
            task_arn,
            cluster_arn,
            task_definition,
            is_new_task_definition and self.auto_deregister_task_definition,
            boto_session,
            ecs_client,
        )

        return ECSTaskResult(
            identifier=identifier,
            # If the container does not start the exit code can be null but we must
            # still report a status code. We use a -1 to indicate a special code.
            status_code=status_code if status_code is not None else -1,
        )

    @sync_compatible
    async def kill(self, identifier: str, grace_seconds: int = 30) -> None:
        """
        Kill a task running on ECS.

        Args:
            identifier: A cluster and task arn combination. This should match a value
                yielded by `ECSTask.run`.
        """
        if grace_seconds != 30:
            self.logger.warning(
                f"Kill grace period of {grace_seconds}s requested, but AWS does not "
                "support dynamic grace period configuration so 30s will be used. "
                "See https://docs.aws.amazon.com/AmazonECS/latest/developerguide/ecs-agent-config.html for configuration of grace periods."  # noqa
            )
        cluster, task = parse_task_identifier(identifier)
        await run_sync_in_worker_thread(self._stop_task, cluster, task)

    @staticmethod
    def get_corresponding_worker_type() -> str:
        """Return the corresponding worker type for this infrastructure block."""
        return ECSWorker.type

    async def generate_work_pool_base_job_template(self) -> dict:
        """
        Generate a base job template for a cloud-run work pool with the same
        configuration as this block.

        Returns:
            - dict: a base job template for a cloud-run work pool
        """
        base_job_template = copy.deepcopy(ECSWorker.get_default_base_job_template())
        for key, value in self.dict(exclude_unset=True, exclude_defaults=True).items():
            if key == "command":
                base_job_template["variables"]["properties"]["command"]["default"] = (
                    shlex.join(value)
                )
            elif key in [
                "type",
                "block_type_slug",
                "_block_document_id",
                "_block_document_name",
                "_is_anonymous",
                "task_customizations",
            ]:
                continue
            elif key == "aws_credentials":
                if not self.aws_credentials._block_document_id:
                    raise BlockNotSavedError(
                        "It looks like you are trying to use a block that"
                        " has not been saved. Please call `.save` on your block"
                        " before publishing it as a work pool."
                    )
                base_job_template["variables"]["properties"]["aws_credentials"][
                    "default"
                ] = {
                    "$ref": {
                        "block_document_id": str(
                            self.aws_credentials._block_document_id
                        )
                    }
                }
            elif key == "task_definition":
                base_job_template["job_configuration"]["task_definition"] = value
            elif key in base_job_template["variables"]["properties"]:
                base_job_template["variables"]["properties"][key]["default"] = value
            else:
                self.logger.warning(
                    f"Variable {key!r} is not supported by Cloud Run work pools."
                    " Skipping."
                )

        if self.task_customizations:
            network_config_patches = JsonPatch(
                [
                    patch
                    for patch in self.task_customizations
                    if "networkConfiguration" in patch["path"]
                ]
            )
            minimal_network_config = assemble_document_for_patches(
                network_config_patches
            )
            if minimal_network_config:
                minimal_network_config_with_patches = network_config_patches.apply(
                    minimal_network_config
                )
                base_job_template["variables"]["properties"]["network_configuration"][
                    "default"
                ] = minimal_network_config_with_patches["networkConfiguration"]
            try:
                base_job_template["job_configuration"]["task_run_request"] = (
                    self.task_customizations.apply(
                        base_job_template["job_configuration"]["task_run_request"]
                    )
                )
            except JsonPointerException:
                self.logger.warning(
                    "Unable to apply task customizations to the base job template."
                    "You may need to update the template manually."
                )

        return base_job_template

    def _stop_task(self, cluster: str, task: str) -> None:
        """
        Stop a running ECS task.
        """
        if self.cluster is not None and cluster != self.cluster:
            raise InfrastructureNotAvailable(
                "Cannot stop ECS task: this infrastructure block has access to "
                f"cluster {self.cluster!r} but the task is running in cluster "
                f"{cluster!r}."
            )

        _, ecs_client = self._get_session_and_client()
        try:
            ecs_client.stop_task(cluster=cluster, task=task)
        except Exception as exc:
            # Raise a special exception if the task does not exist
            if "ClusterNotFound" in str(exc):
                raise InfrastructureNotFound(
                    f"Cannot stop ECS task: the cluster {cluster!r} could not be found."
                ) from exc
            if "not find task" in str(exc) or "referenced task was not found" in str(
                exc
            ):
                raise InfrastructureNotFound(
                    f"Cannot stop ECS task: the task {task!r} could not be found in "
                    f"cluster {cluster!r}."
                ) from exc
            if "no registered tasks" in str(exc):
                raise InfrastructureNotFound(
                    f"Cannot stop ECS task: the cluster {cluster!r} has no tasks."
                ) from exc

            # Reraise unknown exceptions
            raise

    @property
    def _log_prefix(self) -> str:
        """
        Internal property for generating a prefix for logs where `name` may be null
        """
        if self.name is not None:
            return f"ECSTask {self.name!r}"
        else:
            return "ECSTask"

    def _get_session_and_client(self) -> Tuple[boto3.Session, _ECSClient]:
        """
        Retrieve a boto3 session and ECS client
        """
        boto_session = self.aws_credentials.get_boto3_session()
        ecs_client = boto_session.client("ecs")
        return boto_session, ecs_client

    def _create_task_and_wait_for_start(
        self, boto_session: boto3.Session, ecs_client: _ECSClient
    ) -> Tuple[str, str, dict, bool]:
        """
        Register the task definition, create the task run, and wait for it to start.

        Returns a tuple of
        - The task ARN
        - The task's cluster ARN
        - The task definition
        - A bool indicating if the task definition is newly registered
        """
        new_task_definition_registered = False
        requested_task_definition = (
            self._retrieve_task_definition(ecs_client, self.task_definition_arn)
            if self.task_definition_arn
            else self.task_definition
        ) or {}
        task_definition_arn = requested_task_definition.get("taskDefinitionArn", None)

        task_definition = self._prepare_task_definition(
            requested_task_definition, region=ecs_client.meta.region_name
        )

        # We must register the task definition if the arn is null or changes were made
        if task_definition != requested_task_definition or not task_definition_arn:
            # Before registering, check if the latest task definition in the family
            # can be used
            latest_task_definition = self._retrieve_latest_task_definition(
                ecs_client, task_definition["family"]
            )
            if self._task_definitions_equal(latest_task_definition, task_definition):
                self.logger.debug(
                    f"{self._log_prefix}: The latest task definition matches the "
                    "required task definition; using that instead of registering a new "
                    " one."
                )
                task_definition_arn = latest_task_definition["taskDefinitionArn"]
            else:
                if task_definition_arn:
                    self.logger.warning(
                        f"{self._log_prefix}: Settings require changes to the linked "
                        "task definition. A new task definition will be registered. "
                        + (
                            "Enable DEBUG level logs to see the difference."
                            if self.logger.level > logging.DEBUG
                            else ""
                        )
                    )
                    self.logger.debug(
                        f"{self._log_prefix}: Diff for requested task definition"
                        + _pretty_diff(requested_task_definition, task_definition)
                    )
                else:
                    self.logger.info(
                        f"{self._log_prefix}: Registering task definition..."
                    )
                    self.logger.debug(
                        "Task definition payload\n" + yaml.dump(task_definition)
                    )

                task_definition_arn = self._register_task_definition(
                    ecs_client, task_definition
                )
                new_task_definition_registered = True

        if task_definition.get("networkMode") == "awsvpc":
            network_config = self._load_vpc_network_config(self.vpc_id, boto_session)
        else:
            network_config = None

        task_run = self._prepare_task_run(
            network_config=network_config,
            task_definition_arn=task_definition_arn,
        )
        self.logger.info(f"{self._log_prefix}: Creating task run...")
        self.logger.debug("Task run payload\n" + yaml.dump(task_run))

        try:
            task = self._run_task(ecs_client, task_run)
            task_arn = task["taskArn"]
            cluster_arn = task["clusterArn"]
        except Exception as exc:
            self._report_task_run_creation_failure(task_run, exc)

        # Raises an exception if the task does not start
        self.logger.info(f"{self._log_prefix}: Waiting for task run to start...")
        self._wait_for_task_start(
            task_arn, cluster_arn, ecs_client, timeout=self.task_start_timeout_seconds
        )

        return task_arn, cluster_arn, task_definition, new_task_definition_registered

    def _watch_task_and_get_exit_code(
        self,
        task_arn: str,
        cluster_arn: str,
        task_definition: dict,
        deregister_task_definition: bool,
        boto_session: boto3.Session,
        ecs_client: _ECSClient,
    ) -> Optional[int]:
        """
        Wait for the task run to complete and retrieve the exit code of the Prefect
        container.
        """

        # Wait for completion and stream logs
        task = self._wait_for_task_finish(
            task_arn, cluster_arn, task_definition, ecs_client, boto_session
        )

        if deregister_task_definition:
            ecs_client.deregister_task_definition(
                taskDefinition=task["taskDefinitionArn"]
            )

        # Check the status code of the Prefect container
        prefect_container = get_prefect_container(task["containers"])
        assert (
            prefect_container is not None
        ), f"'prefect' container missing from task: {task}"
        status_code = prefect_container.get("exitCode")
        self._report_container_status_code(PREFECT_ECS_CONTAINER_NAME, status_code)

        return status_code

    def _task_definitions_equal(self, taskdef_1, taskdef_2) -> bool:
        """
        Compare two task definitions.

        Since one may come from the AWS API and have populated defaults, we do our best
        to homogenize the definitions without changing their meaning.
        """
        if taskdef_1 == taskdef_2:
            return True

        if taskdef_1 is None or taskdef_2 is None:
            return False

        taskdef_1 = copy.deepcopy(taskdef_1)
        taskdef_2 = copy.deepcopy(taskdef_2)

        def _set_aws_defaults(taskdef):
            """Set defaults that AWS would set after registration"""
            container_definitions = taskdef.get("containerDefinitions", [])
            essential = any(
                container.get("essential") for container in container_definitions
            )
            if not essential:
                container_definitions[0].setdefault("essential", True)

            taskdef.setdefault("networkMode", "bridge")

        _set_aws_defaults(taskdef_1)
        _set_aws_defaults(taskdef_2)

        def _drop_empty_keys(dict_):
            """Recursively drop keys with 'empty' values"""
            for key, value in tuple(dict_.items()):
                if not value:
                    dict_.pop(key)
                if isinstance(value, dict):
                    _drop_empty_keys(value)
                if isinstance(value, list):
                    for v in value:
                        if isinstance(v, dict):
                            _drop_empty_keys(v)

        _drop_empty_keys(taskdef_1)
        _drop_empty_keys(taskdef_2)

        # Clear fields that change on registration for comparison
        for field in POST_REGISTRATION_FIELDS:
            taskdef_1.pop(field, None)
            taskdef_2.pop(field, None)

        return taskdef_1 == taskdef_2

    def preview(self) -> str:
        """
        Generate a preview of the task definition and task run that will be sent to AWS.
        """
        preview = ""

        task_definition_arn = self.task_definition_arn or "<registered at runtime>"

        if self.task_definition or not self.task_definition_arn:
            task_definition = self._prepare_task_definition(
                self.task_definition or {},
                region=self.aws_credentials.region_name
                or "<loaded from client at runtime>",
            )
            preview += "---\n# Task definition\n"
            preview += yaml.dump(task_definition)
            preview += "\n"
        else:
            task_definition = None

        if task_definition and task_definition.get("networkMode") == "awsvpc":
            vpc = "the default VPC" if not self.vpc_id else self.vpc_id
            network_config = {
                "awsvpcConfiguration": {
                    "subnets": f"<loaded from {vpc} at runtime>",
                    "assignPublicIp": "ENABLED",
                }
            }
        else:
            network_config = None

        task_run = self._prepare_task_run(network_config, task_definition_arn)
        preview += "---\n# Task run request\n"
        preview += yaml.dump(task_run)

        return preview

    def _report_container_status_code(
        self, name: str, status_code: Optional[int]
    ) -> None:
        """
        Display a log for the given container status code.
        """
        if status_code is None:
            self.logger.error(
                f"{self._log_prefix}: Task exited without reporting an exit status "
                f"for container {name!r}."
            )
        elif status_code == 0:
            self.logger.info(
                f"{self._log_prefix}: Container {name!r} exited successfully."
            )
        else:
            self.logger.warning(
                f"{self._log_prefix}: Container {name!r} exited with non-zero exit "
                f"code {status_code}."
            )

    def _report_task_run_creation_failure(self, task_run: dict, exc: Exception) -> None:
        """
        Wrap common AWS task run creation failures with nicer user-facing messages.
        """
        # AWS generates exception types at runtime so they must be captured a bit
        # differently than normal.
        if "ClusterNotFoundException" in str(exc):
            cluster = task_run.get("cluster", "default")
            raise RuntimeError(
                f"Failed to run ECS task, cluster {cluster!r} not found. "
                "Confirm that the cluster is configured in your region."
            ) from exc
        elif "No Container Instances" in str(exc) and self.launch_type == "EC2":
            cluster = task_run.get("cluster", "default")
            raise RuntimeError(
                f"Failed to run ECS task, cluster {cluster!r} does not appear to "
                "have any container instances associated with it. Confirm that you "
                "have EC2 container instances available."
            ) from exc
        elif (
            "failed to validate logger args" in str(exc)
            and "AccessDeniedException" in str(exc)
            and self.configure_cloudwatch_logs
        ):
            raise RuntimeError(
                "Failed to run ECS task, the attached execution role does not appear "
                "to have sufficient permissions. Ensure that the execution role "
                f"{self.execution_role!r} has permissions logs:CreateLogStream, "
                "logs:CreateLogGroup, and logs:PutLogEvents."
            )
        else:
            raise

    def _watch_task_run(
        self,
        task_arn: str,
        cluster_arn: str,
        ecs_client: _ECSClient,
        current_status: str = "UNKNOWN",
        until_status: str = None,
        timeout: int = None,
    ) -> Generator[None, None, dict]:
        """
        Watches an ECS task run by querying every `poll_interval` seconds. After each
        query, the retrieved task is yielded. This function returns when the task run
        reaches a STOPPED status or the provided `until_status`.

        Emits a log each time the status changes.
        """
        last_status = status = current_status
        t0 = time.time()
        while status != until_status:
            tasks = ecs_client.describe_tasks(
                tasks=[task_arn], cluster=cluster_arn, include=["TAGS"]
            )["tasks"]

            if tasks:
                task = tasks[0]

                status = task["lastStatus"]
                if status != last_status:
                    self.logger.info(f"{self._log_prefix}: Status is {status}.")

                yield task

                # No point in continuing if the status is final
                if status == "STOPPED":
                    break

                last_status = status

            else:
                # Intermittently, the task will not be described. We wat to respect the
                # watch timeout though.
                self.logger.debug(f"{self._log_prefix}: Task not found.")

            elapsed_time = time.time() - t0
            if timeout is not None and elapsed_time > timeout:
                raise RuntimeError(
                    f"Timed out after {elapsed_time}s while watching task for status "
                    f"{until_status or 'STOPPED'}"
                )
            time.sleep(self.task_watch_poll_interval)

    def _wait_for_task_start(
        self, task_arn: str, cluster_arn: str, ecs_client: _ECSClient, timeout: int
    ) -> dict:
        """
        Waits for an ECS task run to reach a RUNNING status.

        If a STOPPED status is reached instead, an exception is raised indicating the
        reason that the task run did not start.
        """
        for task in self._watch_task_run(
            task_arn, cluster_arn, ecs_client, until_status="RUNNING", timeout=timeout
        ):
            # TODO: It is possible that the task has passed _through_ a RUNNING
            #       status during the polling interval. In this case, there is not an
            #       exception to raise.
            if task["lastStatus"] == "STOPPED":
                code = task.get("stopCode")
                reason = task.get("stoppedReason")
                # Generate a dynamic exception type from the AWS name
                raise type(code, (RuntimeError,), {})(reason)

        return task

    def _wait_for_task_finish(
        self,
        task_arn: str,
        cluster_arn: str,
        task_definition: dict,
        ecs_client: _ECSClient,
        boto_session: boto3.Session,
    ):
        """
        Watch an ECS task until it reaches a STOPPED status.

        If configured, logs from the Prefect container are streamed to stderr.

        Returns a description of the task on completion.
        """
        can_stream_output = False

        if self.stream_output:
            container_def = get_prefect_container(
                task_definition["containerDefinitions"]
            )
            if not container_def:
                self.logger.warning(
                    f"{self._log_prefix}: Prefect container definition not found in "
                    "task definition. Output cannot be streamed."
                )
            elif not container_def.get("logConfiguration"):
                self.logger.warning(
                    f"{self._log_prefix}: Logging configuration not found on task. "
                    "Output cannot be streamed."
                )
            elif not container_def["logConfiguration"].get("logDriver") == "awslogs":
                self.logger.warning(
                    f"{self._log_prefix}: Logging configuration uses unsupported "
                    " driver {container_def['logConfiguration'].get('logDriver')!r}. "
                    "Output cannot be streamed."
                )
            else:
                # Prepare to stream the output
                log_config = container_def["logConfiguration"]["options"]
                logs_client = boto_session.client("logs")
                can_stream_output = True
                # Track the last log timestamp to prevent double display
                last_log_timestamp: Optional[int] = None
                # Determine the name of the stream as "prefix/container/run-id"
                stream_name = "/".join(
                    [
                        log_config["awslogs-stream-prefix"],
                        PREFECT_ECS_CONTAINER_NAME,
                        task_arn.rsplit("/")[-1],
                    ]
                )
                self.logger.info(
                    f"{self._log_prefix}: Streaming output from container "
                    f"{PREFECT_ECS_CONTAINER_NAME!r}..."
                )

        for task in self._watch_task_run(
            task_arn, cluster_arn, ecs_client, current_status="RUNNING"
        ):
            if self.stream_output and can_stream_output:
                # On each poll for task run status, also retrieve available logs
                last_log_timestamp = self._stream_available_logs(
                    logs_client,
                    log_group=log_config["awslogs-group"],
                    log_stream=stream_name,
                    last_log_timestamp=last_log_timestamp,
                )

        return task

    def _stream_available_logs(
        self,
        logs_client: Any,
        log_group: str,
        log_stream: str,
        last_log_timestamp: Optional[int] = None,
    ) -> Optional[int]:
        """
        Stream logs from the given log group and stream since the last log timestamp.

        Will continue on paginated responses until all logs are returned.

        Returns the last log timestamp which can be used to call this method in the
        future.
        """
        last_log_stream_token = "NO-TOKEN"
        next_log_stream_token = None

        # AWS will return the same token that we send once the end of the paginated
        # response is reached
        while last_log_stream_token != next_log_stream_token:
            last_log_stream_token = next_log_stream_token

            request = {
                "logGroupName": log_group,
                "logStreamName": log_stream,
            }

            if last_log_stream_token is not None:
                request["nextToken"] = last_log_stream_token

            if last_log_timestamp is not None:
                # Bump the timestamp by one ms to avoid retrieving the last log again
                request["startTime"] = last_log_timestamp + 1

            try:
                response = logs_client.get_log_events(**request)
            except Exception:
                self.logger.error(
                    (
                        f"{self._log_prefix}: Failed to read log events with request "
                        f"{request}"
                    ),
                    exc_info=True,
                )
                return last_log_timestamp

            log_events = response["events"]
            for log_event in log_events:
                # TODO: This doesn't forward to the local logger, which can be
                #       bad for customizing handling and understanding where the
                #       log is coming from, but it avoid nesting logger information
                #       when the content is output from a Prefect logger on the
                #       running infrastructure
                print(log_event["message"], file=sys.stderr)

                if (
                    last_log_timestamp is None
                    or log_event["timestamp"] > last_log_timestamp
                ):
                    last_log_timestamp = log_event["timestamp"]

            next_log_stream_token = response.get("nextForwardToken")
            if not log_events:
                # Stop reading pages if there was no data
                break

        return last_log_timestamp

    def _retrieve_latest_task_definition(
        self, ecs_client: _ECSClient, task_definition_family: str
    ) -> Optional[dict]:
        try:
            latest_task_definition = self._retrieve_task_definition(
                ecs_client, task_definition_family
            )
        except Exception:
            # The family does not exist...
            return None

        return latest_task_definition

    def _retrieve_task_definition(
        self, ecs_client: _ECSClient, task_definition_arn: str
    ):
        """
        Retrieve an existing task definition from AWS.
        """
        self.logger.info(
            f"{self._log_prefix}: Retrieving task definition {task_definition_arn!r}..."
        )
        response = ecs_client.describe_task_definition(
            taskDefinition=task_definition_arn
        )
        return response["taskDefinition"]

    def _register_task_definition(
        self, ecs_client: _ECSClient, task_definition: dict
    ) -> str:
        """
        Register a new task definition with AWS.
        """
        # TODO: Consider including a global cache for this task definition since
        #       registration of task definitions is frequently rate limited
        task_definition_request = copy.deepcopy(task_definition)

        # We need to remove some fields here if copying an existing task definition
        for field in POST_REGISTRATION_FIELDS:
            task_definition_request.pop(field, None)

        response = ecs_client.register_task_definition(**task_definition_request)
        return response["taskDefinition"]["taskDefinitionArn"]

    def _prepare_task_definition(self, task_definition: dict, region: str) -> dict:
        """
        Prepare a task definition by inferring any defaults and merging overrides.
        """
        task_definition = copy.deepcopy(task_definition)

        # Configure the Prefect runtime container
        task_definition.setdefault("containerDefinitions", [])
        container = get_prefect_container(task_definition["containerDefinitions"])
        if container is None:
            container = {"name": PREFECT_ECS_CONTAINER_NAME}
            task_definition["containerDefinitions"].append(container)

        if self.image:
            container["image"] = self.image

        # Remove any keys that have been explicitly "unset"
        unset_keys = {key for key, value in self.env.items() if value is None}
        for item in tuple(container.get("environment", [])):
            if item["name"] in unset_keys:
                container["environment"].remove(item)

        if self.configure_cloudwatch_logs:
            container["logConfiguration"] = {
                "logDriver": "awslogs",
                "options": {
                    "awslogs-create-group": "true",
                    "awslogs-group": "prefect",
                    "awslogs-region": region,
                    "awslogs-stream-prefix": self.name or "prefect",
                    **self.cloudwatch_logs_options,
                },
            }

        family = self.family or task_definition.get("family") or ECS_DEFAULT_FAMILY
        task_definition["family"] = slugify(
            family,
            max_length=255,
            regex_pattern=r"[^a-zA-Z0-9-_]+",
        )

        # CPU and memory are required in some cases, retrieve the value to use
        cpu = self.cpu or task_definition.get("cpu") or ECS_DEFAULT_CPU
        memory = self.memory or task_definition.get("memory") or ECS_DEFAULT_MEMORY

        if self.launch_type == "FARGATE" or self.launch_type == "FARGATE_SPOT":
            # Task level memory and cpu are required when using fargate
            task_definition["cpu"] = str(cpu)
            task_definition["memory"] = str(memory)

            # The FARGATE compatibility is required if it will be used as as launch type
            requires_compatibilities = task_definition.setdefault(
                "requiresCompatibilities", []
            )
            if "FARGATE" not in requires_compatibilities:
                task_definition["requiresCompatibilities"].append("FARGATE")

            # Only the 'awsvpc' network mode is supported when using FARGATE
            # However, we will not enforce that here if the user has set it
            network_mode = task_definition.setdefault("networkMode", "awsvpc")

            if network_mode != "awsvpc":
                warnings.warn(
                    f"Found network mode {network_mode!r} which is not compatible with "
                    f"launch type {self.launch_type!r}. Use either the 'EC2' launch "
                    "type or the 'awsvpc' network mode."
                )

        elif self.launch_type == "EC2":
            # Container level memory and cpu are required when using ec2
            container.setdefault("cpu", int(cpu))
            container.setdefault("memory", int(memory))

        if self.execution_role_arn and not self.task_definition_arn:
            task_definition["executionRoleArn"] = self.execution_role_arn

        if self.configure_cloudwatch_logs and not task_definition.get(
            "executionRoleArn"
        ):
            raise ValueError(
                "An execution role arn must be set on the task definition to use "
                "`configure_cloudwatch_logs` or `stream_logs` but no execution role "
                "was found on the task definition."
            )

        return task_definition

    def _prepare_task_run_overrides(self) -> dict:
        """
        Prepare the 'overrides' payload for a task run request.
        """
        overrides = {
            "containerOverrides": [
                {
                    "name": PREFECT_ECS_CONTAINER_NAME,
                    "environment": [
                        {"name": key, "value": value}
                        for key, value in {
                            **self._base_environment(),
                            **self.env,
                        }.items()
                        if value is not None
                    ],
                }
            ],
        }

        prefect_container_overrides = overrides["containerOverrides"][0]

        if self.command:
            prefect_container_overrides["command"] = self.command

        if self.execution_role_arn:
            overrides["executionRoleArn"] = self.execution_role_arn

        if self.task_role_arn:
            overrides["taskRoleArn"] = self.task_role_arn

        if self.memory:
            overrides["memory"] = str(self.memory)
            prefect_container_overrides.setdefault("memory", self.memory)

        if self.cpu:
            overrides["cpu"] = str(self.cpu)
            prefect_container_overrides.setdefault("cpu", self.cpu)

        return overrides

    def _load_vpc_network_config(
        self, vpc_id: Optional[str], boto_session: boto3.Session
    ) -> dict:
        """
        Load settings from a specific VPC or the default VPC and generate a task
        run request's network configuration.
        """
        ec2_client = boto_session.client("ec2")
        vpc_message = "the default VPC" if not vpc_id else f"VPC with ID {vpc_id}"

        if not vpc_id:
            # Retrieve the default VPC
            describe = {"Filters": [{"Name": "isDefault", "Values": ["true"]}]}
        else:
            describe = {"VpcIds": [vpc_id]}

        vpcs = ec2_client.describe_vpcs(**describe)["Vpcs"]
        if not vpcs:
            help_message = (
                "Pass an explicit `vpc_id` or configure a default VPC."
                if not vpc_id
                else "Check that the VPC exists in the current region."
            )
            raise ValueError(
                f"Failed to find {vpc_message}. "
                "Network configuration cannot be inferred. "
                + help_message
            )

        vpc_id = vpcs[0]["VpcId"]
        subnets = ec2_client.describe_subnets(
            Filters=[{"Name": "vpc-id", "Values": [vpc_id]}]
        )["Subnets"]
        if not subnets:
            raise ValueError(
                f"Failed to find subnets for {vpc_message}. "
                "Network configuration cannot be inferred."
            )

        return {
            "awsvpcConfiguration": {
                "subnets": [s["SubnetId"] for s in subnets],
                "assignPublicIp": "ENABLED",
                "securityGroups": [],
            }
        }

    def _prepare_task_run(
        self,
        network_config: Optional[dict],
        task_definition_arn: str,
    ) -> dict:
        """
        Prepare a task run request payload.
        """
        task_run = {
            "overrides": self._prepare_task_run_overrides(),
            "tags": [
                {
                    "key": slugify(
                        key,
                        regex_pattern=_TAG_REGEX,
                        allow_unicode=True,
                        lowercase=False,
                    ),
                    "value": slugify(
                        value,
                        regex_pattern=_TAG_REGEX,
                        allow_unicode=True,
                        lowercase=False,
                    ),
                }
                for key, value in self.labels.items()
            ],
            "taskDefinition": task_definition_arn,
        }

        if self.cluster:
            task_run["cluster"] = self.cluster

        if self.launch_type:
            if self.launch_type == "FARGATE_SPOT":
                task_run["capacityProviderStrategy"] = [
                    {"capacityProvider": "FARGATE_SPOT", "weight": 1}
                ]
            else:
                task_run["launchType"] = self.launch_type

        if network_config:
            task_run["networkConfiguration"] = network_config

        task_run = self.task_customizations.apply(task_run)
        return task_run

    def _run_task(self, ecs_client: _ECSClient, task_run: dict):
        """
        Run the task using the ECS client.

        This is isolated as a separate method for testing purposes.
        """
        return ecs_client.run_task(**task_run)["tasks"][0]

Attributes

auto_deregister_task_definition: bool pydantic-field

If set, any task definitions that are created by this block will be deregistered. Existing task definitions linked by ARN will never be deregistered. Deregistering a task definition does not remove it from your AWS account, instead it will be marked as INACTIVE.

aws_credentials: AwsCredentials pydantic-field

The AWS credentials to use to connect to ECS.

cloudwatch_logs_options: Dict[str, str] pydantic-field

When configure_cloudwatch_logs is enabled, this setting may be used to pass additional options to the CloudWatch logs configuration or override the default options. See the AWS documentation for available options. https://docs.aws.amazon.com/AmazonECS/latest/developerguide/using_awslogs.html#create_awslogs_logdriver_options.

cluster: str pydantic-field

The ECS cluster to run the task in. The ARN or name may be provided. If not provided, the default cluster will be used.

configure_cloudwatch_logs: bool pydantic-field

If True, the Prefect container will be configured to send its output to the AWS CloudWatch logs service. This functionality requires an execution role with logs:CreateLogStream, logs:CreateLogGroup, and logs:PutLogEvents permissions. The default for this field is False unless stream_output is set.

cpu: int pydantic-field

The amount of CPU to provide to the ECS task. Valid amounts are specified in the AWS documentation. If not provided, a default value of 1024 will be used unless present on the task definition.

execution_role_arn: str pydantic-field

An execution role to use for the task. This controls the permissions of the task when it is launching. If this value is not null, it will override the value in the task definition. An execution role must be provided to capture logs from the container.

family: str pydantic-field

A family for the task definition. If not provided, it will be inferred from the task definition. If the task definition does not have a family, the name will be generated. When flow and deployment metadata is available, the generated name will include their names. Values for this field will be slugified to match AWS character requirements.

image: str pydantic-field

The image to use for the Prefect container in the task. If this value is not null, it will override the value in the task definition. This value defaults to a Prefect base image matching your local versions.

launch_type: Literal['FARGATE', 'EC2', 'EXTERNAL', 'FARGATE_SPOT'] pydantic-field

The type of ECS task run infrastructure that should be used. Note that 'FARGATE_SPOT' is not a formal ECS launch type, but we will configure the proper capacity provider strategy if set here.

memory: int pydantic-field

The amount of memory to provide to the ECS task. Valid amounts are specified in the AWS documentation. If not provided, a default value of 2048 will be used unless present on the task definition.

stream_output: bool pydantic-field

If True, logs will be streamed from the Prefect container to the local console. Unless you have configured AWS CloudWatch logs manually on your task definition, this requires the same prerequisites outlined in configure_cloudwatch_logs.

task_customizations: JsonPatch pydantic-field

A list of JSON 6902 patches to apply to the task run request. If a string is given, it will parsed as a JSON expression.

task_definition: dict pydantic-field

An ECS task definition to use. Prefect may set defaults or override fields on this task definition to match other ECSTask fields. Cannot be used with task_definition_arn. If not provided, Prefect will generate and register a minimal task definition.

task_definition_arn: str pydantic-field

An identifier for an existing task definition to use. If fields are set on the ECSTask that conflict with the task definition, a new copy will be registered with the required values. Cannot be used with task_definition. If not provided, Prefect will generate and register a minimal task definition.

task_role_arn: str pydantic-field

A role to attach to the task run. This controls the permissions of the task while it is running.

task_start_timeout_seconds: int pydantic-field

The amount of time to watch for the start of the ECS task before marking it as failed. The task must enter a RUNNING state to be considered started.

task_watch_poll_interval: float pydantic-field

The amount of time to wait between AWS API calls while monitoring the state of an ECS task.

vpc_id: str pydantic-field

The AWS VPC to link the task run to. This is only applicable when using the 'awsvpc' network mode for your task. FARGATE tasks require this network mode, but for EC2 tasks the default network mode is 'bridge'. If using the 'awsvpc' network mode and this field is null, your default VPC will be used. If no default VPC can be found, the task run will fail.

Classes

Config

Configuration of pydantic.

Source code in prefect_aws/ecs.py
class Config:
    """Configuration of pydantic."""

    # Support serialization of the 'JsonPatch' type
    arbitrary_types_allowed = True
    json_encoders = {JsonPatch: lambda p: p.patch}

Methods

__init__ special

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

Source code in prefect_aws/ecs.py
def __init__(self, *args, **kwargs):
    super().__init__(*args, **kwargs)
    self.block_initialization()
__json_encoder__ special staticmethod

partial(func, args, *keywords) - new function with partial application of the given arguments and keywords.

cast_customizations_to_a_json_patch classmethod

Casts lists to JsonPatch instances.

Source code in prefect_aws/ecs.py
@validator("task_customizations", pre=True)
def cast_customizations_to_a_json_patch(
    cls, value: Union[List[Dict], JsonPatch, str]
) -> JsonPatch:
    """
    Casts lists to JsonPatch instances.
    """
    if isinstance(value, str):
        value = json.loads(value)
    if isinstance(value, list):
        return JsonPatch(value)
    return value  # type: ignore
cloudwatch_logs_options_requires_configure_cloudwatch_logs classmethod

Enforces that an execution role arn is provided (or could be provided by a runtime task definition) when configuring logging.

Source code in prefect_aws/ecs.py
@root_validator
def cloudwatch_logs_options_requires_configure_cloudwatch_logs(
    cls, values: dict
) -> dict:
    """
    Enforces that an execution role arn is provided (or could be provided by a
    runtime task definition) when configuring logging.
    """
    if values.get("cloudwatch_logs_options") and not values.get(
        "configure_cloudwatch_logs"
    ):
        raise ValueError(
            "`configure_cloudwatch_log` must be enabled to use "
            "`cloudwatch_logs_options`."
        )
    return values
configure_cloudwatch_logs_requires_execution_role_arn classmethod

Enforces that an execution role arn is provided (or could be provided by a runtime task definition) when configuring logging.

Source code in prefect_aws/ecs.py
@root_validator
def configure_cloudwatch_logs_requires_execution_role_arn(
    cls, values: dict
) -> dict:
    """
    Enforces that an execution role arn is provided (or could be provided by a
    runtime task definition) when configuring logging.
    """
    if (
        values.get("configure_cloudwatch_logs")
        and not values.get("execution_role_arn")
        # Do not raise if they've linked to another task definition or provided
        # it without using our shortcuts
        and not values.get("task_definition_arn")
        and not (values.get("task_definition") or {}).get("executionRoleArn")
    ):
        raise ValueError(
            "An `execution_role_arn` must be provided to use "
            "`configure_cloudwatch_logs` or `stream_logs`."
        )
    return values
dict

Convert to a dictionary.

Source code in prefect_aws/ecs.py
def dict(self, *args, **kwargs) -> Dict:
    """
    Convert to a dictionary.
    """
    # Support serialization of the 'JsonPatch' type
    d = super().dict(*args, **kwargs)
    d["task_customizations"] = self.task_customizations.patch
    return d
generate_work_pool_base_job_template async

Generate a base job template for a cloud-run work pool with the same configuration as this block.

Returns:

Type Description
- dict

a base job template for a cloud-run work pool

Source code in prefect_aws/ecs.py
async def generate_work_pool_base_job_template(self) -> dict:
    """
    Generate a base job template for a cloud-run work pool with the same
    configuration as this block.

    Returns:
        - dict: a base job template for a cloud-run work pool
    """
    base_job_template = copy.deepcopy(ECSWorker.get_default_base_job_template())
    for key, value in self.dict(exclude_unset=True, exclude_defaults=True).items():
        if key == "command":
            base_job_template["variables"]["properties"]["command"]["default"] = (
                shlex.join(value)
            )
        elif key in [
            "type",
            "block_type_slug",
            "_block_document_id",
            "_block_document_name",
            "_is_anonymous",
            "task_customizations",
        ]:
            continue
        elif key == "aws_credentials":
            if not self.aws_credentials._block_document_id:
                raise BlockNotSavedError(
                    "It looks like you are trying to use a block that"
                    " has not been saved. Please call `.save` on your block"
                    " before publishing it as a work pool."
                )
            base_job_template["variables"]["properties"]["aws_credentials"][
                "default"
            ] = {
                "$ref": {
                    "block_document_id": str(
                        self.aws_credentials._block_document_id
                    )
                }
            }
        elif key == "task_definition":
            base_job_template["job_configuration"]["task_definition"] = value
        elif key in base_job_template["variables"]["properties"]:
            base_job_template["variables"]["properties"][key]["default"] = value
        else:
            self.logger.warning(
                f"Variable {key!r} is not supported by Cloud Run work pools."
                " Skipping."
            )

    if self.task_customizations:
        network_config_patches = JsonPatch(
            [
                patch
                for patch in self.task_customizations
                if "networkConfiguration" in patch["path"]
            ]
        )
        minimal_network_config = assemble_document_for_patches(
            network_config_patches
        )
        if minimal_network_config:
            minimal_network_config_with_patches = network_config_patches.apply(
                minimal_network_config
            )
            base_job_template["variables"]["properties"]["network_configuration"][
                "default"
            ] = minimal_network_config_with_patches["networkConfiguration"]
        try:
            base_job_template["job_configuration"]["task_run_request"] = (
                self.task_customizations.apply(
                    base_job_template["job_configuration"]["task_run_request"]
                )
            )
        except JsonPointerException:
            self.logger.warning(
                "Unable to apply task customizations to the base job template."
                "You may need to update the template manually."
            )

    return base_job_template
get_corresponding_worker_type staticmethod

Return the corresponding worker type for this infrastructure block.

Source code in prefect_aws/ecs.py
@staticmethod
def get_corresponding_worker_type() -> str:
    """Return the corresponding worker type for this infrastructure block."""
    return ECSWorker.type
image_is_required classmethod

Enforces that an image is available if image is None.

Source code in prefect_aws/ecs.py
@root_validator(pre=True)
def image_is_required(cls, values: dict) -> dict:
    """
    Enforces that an image is available if image is `None`.
    """
    has_image = bool(values.get("image"))
    has_task_definition_arn = bool(values.get("task_definition_arn"))

    # The image can only be null when the task_definition_arn is set
    if has_image or has_task_definition_arn:
        return values

    prefect_container = (
        get_prefect_container(
            (values.get("task_definition") or {}).get("containerDefinitions", [])
        )
        or {}
    )
    image_in_task_definition = prefect_container.get("image")

    # If a task_definition is given with a prefect container image, use that value
    if image_in_task_definition:
        values["image"] = image_in_task_definition
    # Otherwise, it should default to the Prefect base image
    else:
        values["image"] = get_prefect_image_name()
    return values
kill async

Kill a task running on ECS.

Parameters:

Name Type Description Default
identifier str

A cluster and task arn combination. This should match a value yielded by ECSTask.run.

required
Source code in prefect_aws/ecs.py
@sync_compatible
async def kill(self, identifier: str, grace_seconds: int = 30) -> None:
    """
    Kill a task running on ECS.

    Args:
        identifier: A cluster and task arn combination. This should match a value
            yielded by `ECSTask.run`.
    """
    if grace_seconds != 30:
        self.logger.warning(
            f"Kill grace period of {grace_seconds}s requested, but AWS does not "
            "support dynamic grace period configuration so 30s will be used. "
            "See https://docs.aws.amazon.com/AmazonECS/latest/developerguide/ecs-agent-config.html for configuration of grace periods."  # noqa
        )
    cluster, task = parse_task_identifier(identifier)
    await run_sync_in_worker_thread(self._stop_task, cluster, task)
prepare_for_flow_run

Return an copy of the block that is prepared to execute a flow run.

Source code in prefect_aws/ecs.py
def prepare_for_flow_run(
    self: Self,
    flow_run: "FlowRun",
    deployment: Optional["Deployment"] = None,
    flow: Optional["Flow"] = None,
) -> Self:
    """
    Return an copy of the block that is prepared to execute a flow run.
    """
    new_family = None

    # Update the family if not specified elsewhere
    if (
        not self.family
        and not self.task_definition_arn
        and not (self.task_definition and self.task_definition.get("family"))
    ):
        if flow and deployment:
            new_family = f"{ECS_DEFAULT_FAMILY}__{flow.name}__{deployment.name}"
        elif flow and not deployment:
            new_family = f"{ECS_DEFAULT_FAMILY}__{flow.name}"
        elif deployment and not flow:
            # This is a weird case and should not be see in the wild
            new_family = f"{ECS_DEFAULT_FAMILY}__unknown-flow__{deployment.name}"

    new = super().prepare_for_flow_run(flow_run, deployment=deployment, flow=flow)

    if new_family:
        return new.copy(update={"family": new_family})
    else:
        # Avoid an extra copy if not needed
        return new
preview

Generate a preview of the task definition and task run that will be sent to AWS.

Source code in prefect_aws/ecs.py
def preview(self) -> str:
    """
    Generate a preview of the task definition and task run that will be sent to AWS.
    """
    preview = ""

    task_definition_arn = self.task_definition_arn or "<registered at runtime>"

    if self.task_definition or not self.task_definition_arn:
        task_definition = self._prepare_task_definition(
            self.task_definition or {},
            region=self.aws_credentials.region_name
            or "<loaded from client at runtime>",
        )
        preview += "---\n# Task definition\n"
        preview += yaml.dump(task_definition)
        preview += "\n"
    else:
        task_definition = None

    if task_definition and task_definition.get("networkMode") == "awsvpc":
        vpc = "the default VPC" if not self.vpc_id else self.vpc_id
        network_config = {
            "awsvpcConfiguration": {
                "subnets": f"<loaded from {vpc} at runtime>",
                "assignPublicIp": "ENABLED",
            }
        }
    else:
        network_config = None

    task_run = self._prepare_task_run(network_config, task_definition_arn)
    preview += "---\n# Task run request\n"
    preview += yaml.dump(task_run)

    return preview
run async

Run the configured task on ECS.

Source code in prefect_aws/ecs.py
@sync_compatible
async def run(self, task_status: Optional[TaskStatus] = None) -> ECSTaskResult:
    """
    Run the configured task on ECS.
    """
    boto_session, ecs_client = await run_sync_in_worker_thread(
        self._get_session_and_client
    )

    (
        task_arn,
        cluster_arn,
        task_definition,
        is_new_task_definition,
    ) = await run_sync_in_worker_thread(
        self._create_task_and_wait_for_start, boto_session, ecs_client
    )

    # Display a nice message indicating the command and image
    command = self.command or get_prefect_container(
        task_definition["containerDefinitions"]
    ).get("command", [])
    self.logger.info(
        f"{self._log_prefix}: Running command {' '.join(command)!r} "
        f"in container {PREFECT_ECS_CONTAINER_NAME!r} ({self.image})..."
    )

    # The task identifier is "{cluster}::{task}" where we use the configured cluster
    # if set to preserve matching by name rather than arn
    # Note "::" is used despite the Prefect standard being ":" because ARNs contain
    # single colons.
    identifier = (self.cluster if self.cluster else cluster_arn) + "::" + task_arn

    if task_status:
        task_status.started(identifier)

    status_code = await run_sync_in_worker_thread(
        self._watch_task_and_get_exit_code,
        task_arn,
        cluster_arn,
        task_definition,
        is_new_task_definition and self.auto_deregister_task_definition,
        boto_session,
        ecs_client,
    )

    return ECSTaskResult(
        identifier=identifier,
        # If the container does not start the exit code can be null but we must
        # still report a status code. We use a -1 to indicate a special code.
        status_code=status_code if status_code is not None else -1,
    )
set_default_configure_cloudwatch_logs classmethod

Streaming output generally requires CloudWatch logs to be configured.

To avoid entangled arguments in the simple case, configure_cloudwatch_logs defaults to matching the value of stream_output.

Source code in prefect_aws/ecs.py
@root_validator(pre=True)
def set_default_configure_cloudwatch_logs(cls, values: dict) -> dict:
    """
    Streaming output generally requires CloudWatch logs to be configured.

    To avoid entangled arguments in the simple case, `configure_cloudwatch_logs`
    defaults to matching the value of `stream_output`.
    """
    configure_cloudwatch_logs = values.get("configure_cloudwatch_logs")
    if configure_cloudwatch_logs is None:
        values["configure_cloudwatch_logs"] = values.get("stream_output")
    return values

ECSTaskResult (InfrastructureResult) pydantic-model

The result of a run of an ECS task

Source code in prefect_aws/ecs.py
class ECSTaskResult(InfrastructureResult):
    """The result of a run of an ECS task"""

Functions

get_container

Extract a container from a list of containers or container definitions. If not found, None is returned.

Source code in prefect_aws/ecs.py
def get_container(containers: List[dict], name: str) -> Optional[dict]:
    """
    Extract a container from a list of containers or container definitions.
    If not found, `None` is returned.
    """
    for container in containers:
        if container.get("name") == name:
            return container
    return None

get_prefect_container

Extract the Prefect container from a list of containers or container definitions. If not found, None is returned.

Source code in prefect_aws/ecs.py
def get_prefect_container(containers: List[dict]) -> Optional[dict]:
    """
    Extract the Prefect container from a list of containers or container definitions.
    If not found, `None` is returned.
    """
    return get_container(containers, PREFECT_ECS_CONTAINER_NAME)

parse_task_identifier

Splits identifier into its cluster and task components, e.g. input "cluster_name::task_arn" outputs ("cluster_name", "task_arn").

Source code in prefect_aws/ecs.py
def parse_task_identifier(identifier: str) -> Tuple[str, str]:
    """
    Splits identifier into its cluster and task components, e.g.
    input "cluster_name::task_arn" outputs ("cluster_name", "task_arn").
    """
    cluster, task = identifier.split("::", maxsplit=1)
    return cluster, task