1 # -*- coding: utf-8 -*-
2 # Description: elastic search node stats netdata python.d module
5 from base import UrlService
6 from requests import get
7 from socket import gethostbyname
9 from queue import Queue
11 from Queue import Queue
12 from threading import Thread
14 # default module values (can be overridden per job in `config`)
20 # charts order (can be overridden if you want less charts, or different order)
21 ORDER = ['search_perf_total', 'search_perf_time', 'search_latency', 'index_perf_total', 'index_perf_time',
22 'index_latency', 'jvm_mem_heap', 'jvm_gc_count', 'jvm_gc_time', 'thread_pool_qr', 'fdata_cache',
23 'fdata_ev_tr', 'cluster_health_nodes', 'cluster_health_shards', 'cluster_stats_nodes',
24 'cluster_stats_query_cache', 'cluster_stats_docs', 'cluster_stats_store', 'cluster_stats_indices_shards']
27 'search_perf_total': {
28 'options': [None, 'Number of queries, fetches', 'queries', 'Search performance', 'es.search_query', 'stacked'],
30 ['query_total', 'search_total', 'incremental'],
31 ["fetch_total", 'fetch_total', 'incremental'],
32 ["query_current", 'search_current', 'absolute'],
33 ["fetch_current", 'fetch_current', 'absolute']
36 'options': [None, 'Time spent on queries, fetches', 'seconds', 'Search performance', 'es.search_time', 'stacked'],
38 ["query_time_in_millis", 'query', 'incremental', 1, 1000],
39 ["fetch_time_in_millis", 'fetch', 'incremental', 1, 1000]
42 'options': [None, 'Query and fetch latency', 'ms', 'Search performance', 'es.search_latency', 'stacked'],
44 ["query_latency", 'query', 'absolute', 1, 1000],
45 ["fetch_latency", 'fetch', 'absolute', 1, 1000]
48 'options': [None, 'Number of documents indexed, index refreshes, flushes', 'documents/indexes',
49 'Indexing performance', 'es.index_doc', 'stacked'],
51 ['indexing_index_total', 'indexed', 'incremental'],
52 ["refresh_total", 'refreshes', 'incremental'],
53 ["flush_total", 'flushes', 'incremental'],
54 ["indexing_index_current", 'indexed_current', 'absolute'],
57 'options': [None, 'Time spent on indexing, refreshing, flushing', 'seconds', 'Indexing performance',
58 'es.search_time', 'stacked'],
60 ['indexing_index_time_in_millis', 'indexing', 'incremental', 1, 1000],
61 ['refresh_total_time_in_millis', 'refreshing', 'incremental', 1, 1000],
62 ['flush_total_time_in_millis', 'flushing', 'incremental', 1, 1000]
65 'options': [None, 'Indexing and flushing latency', 'ms', 'Indexing performance',
66 'es.index_latency', 'stacked'],
68 ['indexing_latency', 'indexing', 'absolute', 1, 1000],
69 ['flushing_latency', 'flushing', 'absolute', 1, 1000]
72 'options': [None, 'JVM heap currently in use/committed', 'percent/MB', 'Memory usage and gc',
73 'es.jvm_heap', 'area'],
75 ["jvm_heap_percent", 'inuse', 'absolute'],
76 ["jvm_heap_commit", 'commit', 'absolute', -1, 1048576]
79 'options': [None, 'Count of garbage collections', 'counts', 'Memory usage and gc', 'es.gc_count', 'stacked'],
81 ["young_collection_count", 'young', 'incremental'],
82 ["old_collection_count", 'old', 'incremental']
85 'options': [None, 'Time spent on garbage collections', 'ms', 'Memory usage and gc', 'es.gc_time', 'stacked'],
87 ["young_collection_time_in_millis", 'young', 'incremental'],
88 ["old_collection_time_in_millis", 'old', 'incremental']
91 'options': [None, 'Number of queued/rejected threads in thread pool', 'threads', 'Queues and rejections',
94 ["bulk_queue", 'bulk_queue', 'absolute'],
95 ["index_queue", 'index_queue', 'absolute'],
96 ["search_queue", 'search_queue', 'absolute'],
97 ["merge_queue", 'merge_queue', 'absolute'],
98 ["bulk_rejected", 'bulk_rej', 'absolute'],
99 ["index_rejected", 'index_rej', 'absolute'],
100 ["search_rejected", 'search_rej', 'absolute'],
101 ["merge_rejected", 'merge_rej', 'absolute']
104 'options': [None, 'Fielddata cache size', 'MB', 'Fielddata cache', 'es.fdata_cache', 'line'],
106 ["index_fdata_mem", 'mem_size', 'absolute', 1, 1048576]
109 'options': [None, 'Fielddata evictions and circuit breaker tripped count', 'number of events',
110 'Fielddata cache', 'es.fdata_ev_tr', 'line'],
112 ["index_fdata_evic", 'evictions', 'incremental'],
113 ["breakers_fdata_trip", 'tripped', 'incremental']
115 'cluster_health_nodes': {
116 'options': [None, 'Nodes and tasks statistics', 'units', 'Cluster health API',
117 'es.cluster_health', 'stacked'],
119 ["health_number_of_nodes", 'nodes', 'absolute'],
120 ["health_number_of_data_nodes", 'data_nodes', 'absolute'],
121 ["health_number_of_pending_tasks", 'pending_tasks', 'absolute'],
122 ["health_number_of_in_flight_fetch", 'inflight_fetch', 'absolute']
124 'cluster_health_shards': {
125 'options': [None, 'Shards statistics', 'shards', 'Cluster health API',
126 'es.cluster_health_sharts', 'stacked'],
128 ["health_active_shards", 'active_shards', 'absolute'],
129 ["health_relocating_shards", 'relocating_shards', 'absolute'],
130 ["health_unassigned_shards", 'unassigned', 'absolute'],
131 ["health_delayed_unassigned_shards", 'delayed_unassigned', 'absolute'],
132 ["health_initializing_shards", 'initializing', 'absolute'],
133 ["health_active_shards_percent_as_number", 'active_percent', 'absolute']
135 'cluster_stats_nodes': {
136 'options': [None, 'Nodes statistics', 'nodes', 'Cluster stats API',
137 'es.cluster_stats_nodes', 'stacked'],
139 ['count_data_only', 'data_only', 'absolute'],
140 ['count_master_data', 'master_data', 'absolute'],
141 ['count_total', 'total', 'absolute'],
142 ['count_master_only', 'master_only', 'absolute'],
143 ['count_client', 'client', 'absolute']
145 'cluster_stats_query_cache': {
146 'options': [None, 'Query cache statistics', 'queries', 'Cluster stats API',
147 'es.cluster_stats_query_cache', 'stacked'],
149 ['query_cache_hit_count', 'hit', 'incremental'],
150 ['query_cache_miss_count', 'miss', 'incremental']
152 'cluster_stats_docs': {
153 'options': [None, 'Docs statistics', 'count', 'Cluster stats API',
154 'es.cluster_stats_docs', 'line'],
156 ['docs_count', 'docs', 'absolute']
158 'cluster_stats_store': {
159 'options': [None, 'Store statistics', 'MB', 'Cluster stats API',
160 'es.cluster_stats_store', 'line'],
162 ['store_size_in_bytes', 'size', 'absolute', 1, 1048567]
164 'cluster_stats_indices_shards': {
165 'options': [None, 'Indices and shards statistics', 'count', 'Cluster stats API',
166 'es.cluster_stats_ind_sha', 'stacked'],
168 ['indices_count', 'indices', 'absolute'],
169 ['shards_total', 'shards', 'absolute']
174 class Service(UrlService):
175 def __init__(self, configuration=None, name=None):
176 UrlService.__init__(self, configuration=configuration, name=name)
178 self.definitions = CHARTS
179 self.host = self.configuration.get('host')
180 self.port = self.configuration.get('port')
181 self.user = self.configuration.get('user')
182 self.password = self.configuration.get('pass')
183 self.latency = dict()
186 # We can't start if <host> AND <port> not specified
187 if not all([self.host, self.port]):
190 # It as a bad idea to use hostname.
191 # Hostname -> ipaddress
193 self.host = gethostbyname(self.host)
194 except Exception as e:
198 # HTTP Auth? NOT TESTED
199 self.auth = self.user and self.password
201 # Create URL for every Elasticsearch API
202 url_node_stats = 'http://%s:%s/_nodes/_local/stats' % (self.host, self.port)
203 url_cluster_health = 'http://%s:%s/_cluster/health' % (self.host, self.port)
204 url_cluster_stats = 'http://%s:%s/_cluster/stats' % (self.host, self.port)
206 # Create list of enabled API calls
207 user_choice = [bool(self.configuration.get('node_stats', True)),
208 bool(self.configuration.get('cluster_health', True)),
209 bool(self.configuration.get('cluster_stats', True))]
211 avail_methods = [(self._get_node_stats, url_node_stats),
212 (self._get_cluster_health, url_cluster_health),
213 (self._get_cluster_stats, url_cluster_stats)]
215 # Remove disabled API calls from 'avail methods'
216 self.methods = [choice[0] for choice in list(zip(avail_methods, user_choice)) if choice[1]]
218 # Run _get_data for ALL active API calls.
220 for method in self.methods:
221 api_result[method[1]] = (bool(self._get_raw_data(method[1])))
223 # We can start ONLY if all active API calls returned NOT None
224 if not all(api_result.values()):
225 self.error('Plugin could not get data from all APIs')
226 self.error('%s' % api_result)
229 self.info('%s' % api_result)
230 self.info('Plugin was started successfully')
234 def _get_raw_data(self, url):
239 raw_data = get(url, auth=(self.user, self.password))
250 for method in self.methods:
251 th = Thread(target=method[0], args=(queue, method[1]))
255 for thread in threads:
257 result.update(queue.get())
259 return result or None
261 def _get_cluster_health(self, queue, url):
263 Format data received from http request
267 data = self._get_raw_data(url)
275 to_netdata.update(update_key('health', data))
277 queue.put(to_netdata)
279 def _get_cluster_stats(self, queue, url):
281 Format data received from http request
285 data = self._get_raw_data(url)
293 to_netdata.update(update_key('count', data['nodes']['count']))
294 to_netdata.update(update_key('query_cache', data['indices']['query_cache']))
295 to_netdata.update(update_key('docs', data['indices']['docs']))
296 to_netdata.update(update_key('store', data['indices']['store']))
297 to_netdata['indices_count'] = data['indices']['count']
298 to_netdata['shards_total'] = data['indices']['shards']['total']
300 queue.put(to_netdata)
302 def _get_node_stats(self, queue, url):
304 Format data received from http request
308 data = self._get_raw_data(url)
314 node = list(data['nodes'].keys())[0]
316 # Search performance metrics
317 to_netdata.update(data['nodes'][node]['indices']['search'])
318 to_netdata['query_latency'] = self.find_avg(to_netdata['query_total'],
319 to_netdata['query_time_in_millis'], 'query_latency')
320 to_netdata['fetch_latency'] = self.find_avg(to_netdata['fetch_total'],
321 to_netdata['fetch_time_in_millis'], 'fetch_latency')
323 # Indexing performance metrics
324 for key in ['indexing', 'refresh', 'flush']:
325 to_netdata.update(update_key(key, data['nodes'][node]['indices'].get(key, {})))
326 to_netdata['indexing_latency'] = self.find_avg(to_netdata['indexing_index_total'],
327 to_netdata['indexing_index_time_in_millis'], 'index_latency')
328 to_netdata['flushing_latency'] = self.find_avg(to_netdata['flush_total'],
329 to_netdata['flush_total_time_in_millis'], 'flush_latency')
330 # Memory usage and garbage collection
331 to_netdata.update(update_key('young', data['nodes'][node]['jvm']['gc']['collectors']['young']))
332 to_netdata.update(update_key('old', data['nodes'][node]['jvm']['gc']['collectors']['old']))
333 to_netdata['jvm_heap_percent'] = data['nodes'][node]['jvm']['mem']['heap_used_percent']
334 to_netdata['jvm_heap_commit'] = data['nodes'][node]['jvm']['mem']['heap_committed_in_bytes']
336 # Thread pool queues and rejections
337 for key in ['bulk', 'index', 'search', 'merge']:
338 to_netdata.update(update_key(key, data['nodes'][node]['thread_pool'].get(key, {})))
341 to_netdata['index_fdata_mem'] = data['nodes'][node]['indices']['fielddata']['memory_size_in_bytes']
342 to_netdata['index_fdata_evic'] = data['nodes'][node]['indices']['fielddata']['evictions']
343 to_netdata['breakers_fdata_trip'] = data['nodes'][node]['breakers']['fielddata']['tripped']
345 queue.put(to_netdata)
347 def find_avg(self, value1, value2, key):
348 if key not in self.latency.keys():
349 self.latency.update({key: [value1, value2]})
352 if not self.latency[key][0] == value1:
353 latency = round(float(value2 - self.latency[key][1]) / float(value1 - self.latency[key][0]) * 1000)
354 self.latency.update({key: [value1, value2]})
357 self.latency.update({key: [value1, value2]})
361 def update_key(string, dictionary):
362 return {'_'.join([string, k]): v for k, v in dictionary.items()}