SDK for Python (Boto3) を使用する Amazon EMR の例

Spark ステップを追加します。このステップは、追加されるとすぐにクラスターによって実行されます。

def add_step(cluster_id, name, script_uri, script_args, emr_client):
    """
    Adds a job step to the specified cluster. This example adds a Spark
    step, which is run by the cluster as soon as it is added.

    :param cluster_id: The ID of the cluster.
    :param name: The name of the step.
    :param script_uri: The URI where the Python script is stored.
    :param script_args: Arguments to pass to the Python script.
    :param emr_client: The Boto3 EMR client object.
    :return: The ID of the newly added step.
    """
    try:
        response = emr_client.add_job_flow_steps(
            JobFlowId=cluster_id,
            Steps=[
                {
                    "Name": name,
                    "ActionOnFailure": "CONTINUE",
                    "HadoopJarStep": {
                        "Jar": "command-runner.jar",
                        "Args": [
                            "spark-submit",
                            "--deploy-mode",
                            "cluster",
                            script_uri,
                            *script_args,
                        ],
                    },
                }
            ],
        )
        step_id = response["StepIds"][0]
        logger.info("Started step with ID %s", step_id)
    except ClientError:
        logger.exception("Couldn't start step %s with URI %s.", name, script_uri)
        raise
    else:
        return step_id

Amazon EMR ファイルシステム (EMRFS) コマンドをクラスターのジョブステップとして実行します。これを使用すると、SSH 接続を介して手動でコマンドを実行する代わりに、クラスターで EMRFS コマンドを自動化できます。

import boto3
from botocore.exceptions import ClientError


def add_emrfs_step(command, bucket_url, cluster_id, emr_client):
    """
    Add an EMRFS command as a job flow step to an existing cluster.

    :param command: The EMRFS command to run.
    :param bucket_url: The URL of a bucket that contains tracking metadata.
    :param cluster_id: The ID of the cluster to update.
    :param emr_client: The Boto3 Amazon EMR client object.
    :return: The ID of the added job flow step. Status can be tracked by calling
             the emr_client.describe_step() function.
    """
    job_flow_step = {
        "Name": "Example EMRFS Command Step",
        "ActionOnFailure": "CONTINUE",
        "HadoopJarStep": {
            "Jar": "command-runner.jar",
            "Args": ["/usr/bin/emrfs", command, bucket_url],
        },
    }

    try:
        response = emr_client.add_job_flow_steps(
            JobFlowId=cluster_id, Steps=[job_flow_step]
        )
        step_id = response["StepIds"][0]
        print(f"Added step {step_id} to cluster {cluster_id}.")
    except ClientError:
        print(f"Couldn't add a step to cluster {cluster_id}.")
        raise
    else:
        return step_id


def usage_demo():
    emr_client = boto3.client("emr")
    # Assumes the first waiting cluster has EMRFS enabled and has created metadata
    # with the default name of 'EmrFSMetadata'.
    cluster = emr_client.list_clusters(ClusterStates=["WAITING"])["Clusters"][0]
    add_emrfs_step(
        "sync", "s3://elasticmapreduce/samples/cloudfront", cluster["Id"], emr_client
    )


if __name__ == "__main__":
    usage_demo()

def describe_cluster(cluster_id, emr_client):
    """
    Gets detailed information about a cluster.

    :param cluster_id: The ID of the cluster to describe.
    :param emr_client: The Boto3 EMR client object.
    :return: The retrieved cluster information.
    """
    try:
        response = emr_client.describe_cluster(ClusterId=cluster_id)
        cluster = response["Cluster"]
        logger.info("Got data for cluster %s.", cluster["Name"])
    except ClientError:
        logger.exception("Couldn't get data for cluster %s.", cluster_id)
        raise
    else:
        return cluster

def describe_step(cluster_id, step_id, emr_client):
    """
    Gets detailed information about the specified step, including the current state of
    the step.

    :param cluster_id: The ID of the cluster.
    :param step_id: The ID of the step.
    :param emr_client: The Boto3 EMR client object.
    :return: The retrieved information about the specified step.
    """
    try:
        response = emr_client.describe_step(ClusterId=cluster_id, StepId=step_id)
        step = response["Step"]
        logger.info("Got data for step %s.", step_id)
    except ClientError:
        logger.exception("Couldn't get data for step %s.", step_id)
        raise
    else:
        return step

def list_steps(cluster_id, emr_client):
    """
    Gets a list of steps for the specified cluster. In this example, all steps are
    returned, including completed and failed steps.

    :param cluster_id: The ID of the cluster.
    :param emr_client: The Boto3 EMR client object.
    :return: The list of steps for the specified cluster.
    """
    try:
        response = emr_client.list_steps(ClusterId=cluster_id)
        steps = response["Steps"]
        logger.info("Got %s steps for cluster %s.", len(steps), cluster_id)
    except ClientError:
        logger.exception("Couldn't get steps for cluster %s.", cluster_id)
        raise
    else:
        return steps

def run_job_flow(
    name,
    log_uri,
    keep_alive,
    applications,
    job_flow_role,
    service_role,
    security_groups,
    steps,
    emr_client,
):
    """
    Runs a job flow with the specified steps. A job flow creates a cluster of
    instances and adds steps to be run on the cluster. Steps added to the cluster
    are run as soon as the cluster is ready.

    This example uses the 'emr-5.30.1' release. A list of recent releases can be
    found here:
        https://docs.aws.amazon.com/emr/latest/ReleaseGuide/emr-release-components.html.

    :param name: The name of the cluster.
    :param log_uri: The URI where logs are stored. This can be an Amazon S3 bucket URL,
                    such as 's3://my-log-bucket'.
    :param keep_alive: When True, the cluster is put into a Waiting state after all
                       steps are run. When False, the cluster terminates itself when
                       the step queue is empty.
    :param applications: The applications to install on each instance in the cluster,
                         such as Hive or Spark.
    :param job_flow_role: The IAM role assumed by the cluster.
    :param service_role: The IAM role assumed by the service.
    :param security_groups: The security groups to assign to the cluster instances.
                            Amazon EMR adds all needed rules to these groups, so
                            they can be empty if you require only the default rules.
    :param steps: The job flow steps to add to the cluster. These are run in order
                  when the cluster is ready.
    :param emr_client: The Boto3 EMR client object.
    :return: The ID of the newly created cluster.
    """
    try:
        response = emr_client.run_job_flow(
            Name=name,
            LogUri=log_uri,
            ReleaseLabel="emr-5.30.1",
            Instances={
                "MasterInstanceType": "m5.xlarge",
                "SlaveInstanceType": "m5.xlarge",
                "InstanceCount": 3,
                "KeepJobFlowAliveWhenNoSteps": keep_alive,
                "EmrManagedMasterSecurityGroup": security_groups["manager"].id,
                "EmrManagedSlaveSecurityGroup": security_groups["worker"].id,
            },
            Steps=[
                {
                    "Name": step["name"],
                    "ActionOnFailure": "CONTINUE",
                    "HadoopJarStep": {
                        "Jar": "command-runner.jar",
                        "Args": [
                            "spark-submit",
                            "--deploy-mode",
                            "cluster",
                            step["script_uri"],
                            *step["script_args"],
                        ],
                    },
                }
                for step in steps
            ],
            Applications=[{"Name": app} for app in applications],
            JobFlowRole=job_flow_role.name,
            ServiceRole=service_role.name,
            EbsRootVolumeSize=10,
            VisibleToAllUsers=True,
        )
        cluster_id = response["JobFlowId"]
        logger.info("Created cluster %s.", cluster_id)
    except ClientError:
        logger.exception("Couldn't create cluster.")
        raise
    else:
        return cluster_id

def terminate_cluster(cluster_id, emr_client):
    """
    Terminates a cluster. This terminates all instances in the cluster and cannot
    be undone. Any data not saved elsewhere, such as in an Amazon S3 bucket, is lost.

    :param cluster_id: The ID of the cluster to terminate.
    :param emr_client: The Boto3 EMR client object.
    """
    try:
        emr_client.terminate_job_flows(JobFlowIds=[cluster_id])
        logger.info("Terminated cluster %s.", cluster_id)
    except ClientError:
        logger.exception("Couldn't terminate cluster %s.", cluster_id)
        raise

Apache Spark を使用して多数の計算を並列化し pi の値を推定する、短期運用する Amazon EMR クラスターを作成します。このジョブの出力は、Amazon EMR ログと Amazon Simple Storage Service (Amazon S3) バケットに書き込まれます。このクラスターは、このジョブの完了後に自動的に終了します。

この例は GitHub で最もよく確認できます。完全なソースコードとセットアップおよび実行の手順については、GitHub で完全な例を参照してください。

import argparse
import time
import boto3


def install_libraries_on_core_nodes(cluster_id, script_path, emr_client, ssm_client):
    """
    Copies and runs a shell script on the core nodes in the cluster.

    :param cluster_id: The ID of the cluster.
    :param script_path: The path to the script, typically an Amazon S3 object URL.
    :param emr_client: The Boto3 Amazon EMR client.
    :param ssm_client: The Boto3 AWS Systems Manager client.
    """
    core_nodes = emr_client.list_instances(
        ClusterId=cluster_id, InstanceGroupTypes=["CORE"]
    )["Instances"]
    core_instance_ids = [node["Ec2InstanceId"] for node in core_nodes]
    print(f"Found core instances: {core_instance_ids}.")

    commands = [
        # Copy the shell script from Amazon S3 to each node instance.
        f"aws s3 cp {script_path} /home/hadoop",
        # Run the shell script to install libraries on each node instance.
        "bash /home/hadoop/install_libraries.sh",
    ]
    for command in commands:
        print(f"Sending '{command}' to core instances...")
        command_id = ssm_client.send_command(
            InstanceIds=core_instance_ids,
            DocumentName="AWS-RunShellScript",
            Parameters={"commands": [command]},
            TimeoutSeconds=3600,
        )["Command"]["CommandId"]
        while True:
            # Verify the previous step succeeded before running the next step.
            cmd_result = ssm_client.list_commands(CommandId=command_id)["Commands"][0]
            if cmd_result["StatusDetails"] == "Success":
                print(f"Command succeeded.")
                break
            elif cmd_result["StatusDetails"] in ["Pending", "InProgress"]:
                print(f"Command status is {cmd_result['StatusDetails']}, waiting...")
                time.sleep(10)
            else:
                print(f"Command status is {cmd_result['StatusDetails']}, quitting.")
                raise RuntimeError(
                    f"Command {command} failed to run. "
                    f"Details: {cmd_result['StatusDetails']}"
                )


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("cluster_id", help="The ID of the cluster.")
    parser.add_argument("script_path", help="The path to the script in Amazon S3.")
    args = parser.parse_args()

    emr_client = boto3.client("emr")
    ssm_client = boto3.client("ssm")

    install_libraries_on_core_nodes(
        args.cluster_id, args.script_path, emr_client, ssm_client
    )


if __name__ == "__main__":
    main()