A Zabbix script that will monitor Elasticsearch on a node. Updated and originally from here: http://blog.sepa.spb.ru/2014/12/elasticsearch-internals-monitoring-by.html

elasticsearch.py

#!/usr/bin/env python

import requests
import json
import sys
import os
import time

# Run without parameters to debug zabbix_sender
# Run with param jvm.uptime_in_millis to trigger trap sending and 
# return jvm.uptime_in_millis value
# (active check used as trigger)
sender = '/usr/bin/zabbix_sender'      # path zabbix_sender
cfg = '/etc/zabbix/zabbix_agentd.conf' # path to zabbix-agent config
tmp = '/tmp/es_stats.tmp'              # temp file to use


# list of traps to send

# keys for health page
traps1 = [                             
  "status: None",
  "active_primary_shards: None",
  "active_shards: None",
  "initializing_shards: None",
  "relocating_shards: None",
  "unassigned_shards: None",
  "number_of_nodes: None",
  "number_of_data_nodes: None"
]

# keys for cluster stats page
traps2 = [                             
  "indices.docs.count: None",
  "indices.docs.deleted: None",
  "indices.flush.total: None",
  "indices.flush.total_time_in_millis: None",
  "indices.get.exists_time_in_millis: None",
  "indices.get.exists_total: None",
  "indices.get.missing_time_in_millis: None",
  "indices.get.missing_total: None",
  "indices.indexing.delete_time_in_millis: None",
  "indices.indexing.delete_total: None",
  "indices.indexing.index_time_in_millis: None",
  "indices.indexing.index_total: None",
  "indices.merges.total: None",
  "indices.merges.total_time_in_millis: None",
  "indices.refresh.total: None",
  "indices.refresh.total_time_in_millis: None",
  "indices.search.fetch_time_in_millis: None",
  "indices.search.fetch_total: None",
  "indices.search.query_time_in_millis: None",
  "indices.search.query_total: None",
  "indices.store.throttle_time_in_millis: None",
  "indices.warmer.total: None",
  "indices.warmer.total_time_in_millis: None",
  "jvm.mem.heap_committed_in_bytes: None",
  "jvm.mem.heap_used_in_bytes: None",
  "os.mem.actual_free_in_bytes: None",
  "os.mem.actual_used_in_bytes: None"
]



# read specified keys from json data
def getKeys(stats,traps):
  out=''
  for t in traps:
    c=t.split('.')
    s=stats
    while len(c): s=s.get(c.pop(0),{})
    if s=={}: continue
    out += "- es.{0} {1}\n".format(t,s)
  return out

def main():
  # load json data
  try:
    f = requests.get("http://localhost:9200/_cluster/health")
    health = f.json()
    f = requests.get("http://localhost:9200/_nodes/_local/stats?all=true")
    all = f.json()
    
    # only for current node
    for node_id in all['nodes']:
      if all['nodes'][node_id]['host'].startswith(os.uname()[1]):
        node = all['nodes'][node_id]
        if len(sys.argv) == 1: 
          print "node found"
  except:
    print "Unable to load JSON data!"
    sys.exit(1)

  out = getKeys(health,traps1)  #getting health values
  out += getKeys(node,traps2)    #getting stats  values

  # write data for zabbix sender
  if len(sys.argv) == 1: 
    print out

  try:
    with open(tmp,'w') as f: f.write(out)
  except:
    print "Unable to save data to send!"
    sys.exit(1)

  # return active check
  if len(sys.argv) > 1 and sys.argv[1] == 'jvm.uptime_in_millis':
    os.system("{0} -c {1} -i {2} >/dev/null 2>&1".format(sender,cfg,tmp))
    print node['jvm']['uptime_in_millis']
  # send data with debug
  else:
    os.system("{0} -c {1} -i {2} -vv".format(sender,cfg,tmp))

  os.remove(tmp)

if __name__ == "__main__":
    main()