Kill yarn apps on EMR clusters that take longer than a specified amount of time

emr_kill_hanging_job.py

# Copyright 2019 Parsely
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# Author: Conrad Lee <[email protected]>

# Consider running this via cron

import json
import os
import subprocess
import urllib
from xml.etree import ElementTree
import yarn_api_client


MAX_JOB_DURATION_SECONDS = 60 * 60 * 3 # How long a YARN app can live before its killed


def kill_yarn_app(app_id):
    command_args = ["/usr/bin/yarn", "application", "-kill", app_id]
    subprocess.run(command_args, check=True)


def get_resource_manager_addr(yarn_conf_path="/etc/hadoop/conf/yarn-site.xml"):
    # First wait for the yarn conf file
    if not os.path.isfile(yarn_conf_path):
        return None

    dom = ElementTree.parse(yarn_conf_path)
    address = None
    settings = dom.findall("property")
    for setting in settings:
        if setting.find("name").text == "yarn.resourcemanager.address":
            address = setting.find("value").text

    if address is None:
        return None
    else:
        host, port = address.split(":")
        return host


def get_longest_running_yarn_app(host, port=8088):
    rm = yarn_api_client.resource_manager.ResourceManager(address=host, port=port)
    apps_data = rm.cluster_applications(state="RUNNING").data  # ['apps']['app']
    if apps_data["apps"] is None:
        return None
    running_apps = apps_data["apps"]["app"]
    running_apps.sort(key=lambda x: x["elapsedTime"], reverse=True)
    return running_apps[0]


def get_instance_id(aws_endpoint="http://169.254.169.254/latest/meta-data/instance-id"):
    return urllib.request.urlopen(aws_endpoint).read().decode()


def current_node_master():
    """Returns boolean indicating whether the node on which this
    function is executed the EMR master node."""
    current_instance_id = get_instance_id()
    master_instance_id = get_cluster_config()["masterInstanceId"]
    return current_instance_id == master_instance_id


def get_cluster_config():
    with open("/mnt/var/lib/info/job-flow.json") as configfile:
        config = json.loads(configfile.read().strip())
    return config


if __name__ == "__main__":
    if current_node_master():
        host = get_resource_manager_addr()
        app = get_longest_running_yarn_app(host)
        if app is not None:
            elapsed_time_seconds = app["elapsedTime"] / 1000
            if elapsed_time_seconds > MAX_JOB_DURATION_SECONDS:
                kill_yarn_app(app["id"])