task

#!/usr/bin/python

from requests import get
from config import config
from sys import argv
from datetime import datetime
from getopt import getopt, GetoptError
import json, operator

# Limit - only display the latest N tasks
limit = 0
human_time = False

try:
    opts, argvs = getopt(argv[1:], 'h', ['human-time'])
    for key, value in opts:
        if key in ('-h', '--human-time'):
            human_time = True
    limit = int(argvs[0])
except IndexError:
    limit = 0
except (GetoptError, KeyError) as e:
    print(e.message)
    exit(0)

namenode_host = config['master']
resp = get('http://%s:8088/ws/v1/cluster/apps?states=FINISHED' % namenode_host)

try:
    task_list = json.loads(resp.content)['apps']['app']
except KeyError:
    print('No tasks in record.')
    exit(0)

task_list = sorted(task_list, cmp=lambda x, y: operator.ge(x['startedTime'], y['startedTime']))

if limit > 0:
    task_list = task_list[-limit:]

# print 'Task Name             Start Time\tEnd Time'
for task in task_list:
    task['name'] = task['name'].replace(' ', '_')
    #task['name'] = task['name'].ljust(21)
    if human_time:
        start_dt = datetime.fromtimestamp(int(task['startedTime']) / 1000)
        end_dt = datetime.fromtimestamp(int(task['finishedTime']) / 1000)
        task_duration = end_dt - start_dt
        print('%s %s\t%s -> %s' % (task['name'], str(start_dt), str(end_dt), str(task_duration)))
    else:
        print('%s %s\t%s' % (task['name'], task['startedTime'], task['finishedTime']))