get_active_namenode.py

#Determine the active namenode of HDFS, this is required as the webHDFS implementation doesn't redirect to active namenode

import os
import json
import xml.etree.ElementTree as ET
try:
    from urllib import urlopen
except ImportError:
    from urllib.request import urlopen

def find_element(conf, property):
    tree = ET.parse(conf)
    root = tree.getroot()
    for elem in root.iter('property'):
        if elem[0].text == property:
            return elem[1].text

conf = os.environ.get('HADOOP_CONF_DIR','/etc/hadoop/conf') + '/hdfs-site.xml'
nameservice = find_element(conf, 'dfs.internal.nameservices')
namenodes = find_element(conf, 'dfs.ha.namenodes' + '.' + nameservice)
for namenode in namenodes.split(','):
    if json.loads(urlopen('http://' + namenode
                      + ':50070/jmx?get=Hadoop:service=NameNode,name=NameNodeStatus::State'
                      ).read())['beans'][0]['State'] == 'active':
        break
print(namenode)