Hadoop application automatic restart script

Recently, the cluster is not stable, the application always hangs, and the manual restart is too annoying. I wrote an automatic restart script and share it.

#!--encoding=utf-8
import os
import shlex
import subprocess

from apscheduler.schedulers.blocking import BlockingScheduler

# app_name : restart path[shell_path, jar_path]
restart_dict = {
    "app_mr": "/app/mr/start.sh /app/mr/jars/app_mr.jar",
    "app_streaming": "/app/streaming/start.sh /app/streaming/jars/app_streaming.jar",
                }


def task():
    app_list = os.popen(" yarn application -list | awk '{print $2}'| tail -n +3 ")
    app_set = set()
    for app in app_list:
        app_set.add(app.replace("\n", ""))
    for k, v in restart_dict.items():
        if k not in app_set:
            p = subprocess.Popen(shlex.split(v))
            # (stdout, stderr) = p.communicate()
            # p.send_signal(CTRL_C_EVENT)


if __name__ == "__main__":
    scheduler = BlockingScheduler()
    # 每30s一次
    scheduler.add_job(task, 'cron', second='*/30')
    try:
        scheduler.start()
    except Exception, ex:
        print ex
        scheduler.shutdown()