-
Notifications
You must be signed in to change notification settings - Fork 814
/
agent.py
executable file
·638 lines (531 loc) · 23.8 KB
/
agent.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
#!/opt/datadog-agent/embedded/bin/python
"""
Datadog
www.datadoghq.com
----
Cloud-Scale Monitoring. Monitoring that tracks your dynamic infrastructure.
Licensed under Simplified BSD License (see LICENSE)
(C) Boxed Ice 2010 all rights reserved
(C) Datadog, Inc. 2010-2016 all rights reserved
"""
# set up logging before importing any other components
from config import get_version, initialize_logging # noqa
initialize_logging('collector')
# stdlib
from copy import copy
import logging
import os
import signal
import sys
import time
import xmlrpclib
# For pickle & PID files, see issue 293
os.umask(022)
# 3p
try:
import supervisor.xmlrpc
except ImportError:
# Not used/shipped on Windows
pass
# project
from checks.check_status import CollectorStatus
from checks.collector import Collector
from config import (
get_config,
get_jmx_pipe_path,
get_parsed_args,
get_system_stats,
load_check_directory,
load_check,
generate_jmx_configs,
_is_affirmative,
SD_PIPE_NAME
)
from daemon import AgentSupervisor, Daemon
from emitter import http_emitter
from jmxfetch import get_jmx_checks, JMXFetch
# utils
from utils.cloud_metadata import EC2
from utils.configcheck import configcheck, sd_configcheck
from utils.flare import Flare
from utils.hostname import get_hostname
from utils.jmx import jmx_command
from utils.pidfile import PidFile
from utils.platform import Platform
from utils.profile import AgentProfiler
from utils.service_discovery.config_stores import get_config_store
from utils.service_discovery.sd_backend import get_sd_backend
from utils.watchdog import Watchdog
from utils.windows_configuration import get_sdk_integration_paths
# Constants
PID_NAME = "dd-agent"
PID_DIR = None
WATCHDOG_MULTIPLIER = 10
RESTART_INTERVAL = 4 * 24 * 60 * 60 # Defaults to 4 days
JMX_SUPERVISOR_ENTRY = 'datadog-agent:jmxfetch'
JMX_GRACE_SECS = 2
SERVICE_DISCOVERY_PREFIX = 'SD-'
SD_CONFIG_SEP = "#### SERVICE-DISCOVERY ####\n"
SD_CONFIG_TERM = "#### SERVICE-DISCOVERY TERM ####\n"
DEFAULT_SUPERVISOR_SOCKET = '/opt/datadog-agent/run/datadog-supervisor.sock'
DEFAULT_COLLECTOR_PROFILE_INTERVAL = 20
# Globals
log = logging.getLogger('collector')
class Agent(Daemon):
"""
The agent class is a daemon that runs the collector in a background process.
"""
def __init__(self, pidfile, autorestart, start_event=True, in_developer_mode=False):
Daemon.__init__(self, pidfile, autorestart=autorestart)
self.run_forever = True
self.collector = None
self.start_event = start_event
self.in_developer_mode = in_developer_mode
self._agentConfig = {}
self._checksd = []
self.collector_profile_interval = DEFAULT_COLLECTOR_PROFILE_INTERVAL
self.check_frequency = None
# this flag can be set to True, False, or a set of checks (for partial reload)
self.reload_configs_flag = False
self.sd_backend = None
self.supervisor_proxy = None
self.sd_pipe = None
self.last_jmx_piped = None
def _handle_sigterm(self, signum, frame):
"""Handles SIGTERM and SIGINT, which gracefully stops the agent."""
log.debug("Caught sigterm. Stopping run loop.")
self.run_forever = False
if self.collector:
self.collector.stop()
log.debug("Collector is stopped.")
def _handle_sigusr1(self, signum, frame):
"""Handles SIGUSR1, which signals an exit with an autorestart."""
self._handle_sigterm(signum, frame)
self._do_restart()
def _handle_sighup(self, signum, frame):
"""Handles SIGHUP, which signals a configuration reload."""
log.info("SIGHUP caught! Scheduling configuration reload before next collection run.")
self.reload_configs_flag = True
def reload_configs(self, checks_to_reload=set()):
"""Reload the agent configuration and checksd configurations.
Can also reload only an explicit set of checks."""
log.info("Attempting a configuration reload...")
hostname = get_hostname(self._agentConfig)
jmx_sd_configs = None
# if no check was given, reload them all
if not checks_to_reload:
log.debug("No check list was passed, reloading every check")
# stop checks
for check in self._checksd.get('initialized_checks', []):
check.stop()
self._checksd = load_check_directory(self._agentConfig, hostname)
if self._jmx_service_discovery_enabled:
jmx_sd_configs = generate_jmx_configs(self._agentConfig, hostname)
else:
new_checksd = copy(self._checksd)
all_jmx_checks = get_jmx_checks(auto_conf=True)
jmx_checks = [check for check in checks_to_reload if check in all_jmx_checks]
py_checks = set(checks_to_reload) - set(jmx_checks)
self.refresh_specific_checks(hostname, new_checksd, py_checks)
if self._jmx_service_discovery_enabled:
jmx_sd_configs = generate_jmx_configs(self._agentConfig, hostname, jmx_checks)
# once the reload is done, replace existing checks with the new ones
self._checksd = new_checksd
if jmx_sd_configs:
self._submit_jmx_service_discovery(jmx_sd_configs)
# Logging
num_checks = len(self._checksd['initialized_checks'])
if num_checks > 0:
opt_msg = " (refreshed %s checks)" % len(checks_to_reload) if checks_to_reload else ''
msg = "Check reload was successful. Running {num_checks} checks{opt_msg}.".format(
num_checks=num_checks, opt_msg=opt_msg)
log.info(msg)
else:
log.info("No checksd configs found")
def refresh_specific_checks(self, hostname, checksd, checks):
"""take a set of checks and for each of them:
- remove it from the init_failed_checks if it was there
- load a fresh config for it
- replace its old config with the new one in initialized_checks if there was one
- disable the check if no new config was found
- otherwise, append it to initialized_checks
"""
for check_name in checks:
idx = None
for num, check in enumerate(checksd['initialized_checks']):
if check.name == check_name:
idx = num
# stop the existing check before reloading it
check.stop()
if not idx and check_name in checksd['init_failed_checks']:
# if the check previously failed to load, pop it from init_failed_checks
checksd['init_failed_checks'].pop(check_name)
fresh_check = load_check(self._agentConfig, hostname, check_name)
# this is an error dict
# checks that failed to load are added to init_failed_checks
# and poped from initialized_checks
if isinstance(fresh_check, dict) and 'error' in fresh_check.keys():
checksd['init_failed_checks'][fresh_check.keys()[0]] = fresh_check.values()[0]
if idx:
checksd['initialized_checks'].pop(idx)
elif not fresh_check:
# no instance left of it to monitor so the check was not loaded
if idx:
checksd['initialized_checks'].pop(idx)
# the check was not previously running so we were trying to instantiate it and it failed
else:
log.error("Configuration for check %s was not found, it won't be reloaded." % check_name)
# successfully reloaded check are added to initialized_checks
# (appended or replacing a previous version)
else:
if idx is not None:
checksd['initialized_checks'][idx] = fresh_check
# it didn't exist before and doesn't need to be replaced so we append it
else:
checksd['initialized_checks'].append(fresh_check)
@classmethod
def info(cls, verbose=None):
logging.getLogger().setLevel(logging.ERROR)
return CollectorStatus.print_latest_status(verbose=verbose)
def sd_pipe_jmx_configs(self, hostname):
jmx_sd_configs = generate_jmx_configs(self._agentConfig, hostname)
if jmx_sd_configs:
self._submit_jmx_service_discovery(jmx_sd_configs)
def run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
if not Platform.is_windows():
# A SIGUSR1 signals an exit with an autorestart
signal.signal(signal.SIGUSR1, self._handle_sigusr1)
# Handle Keyboard Interrupt
signal.signal(signal.SIGINT, self._handle_sigterm)
# A SIGHUP signals a configuration reload
signal.signal(signal.SIGHUP, self._handle_sighup)
else:
sdk_integrations = get_sdk_integration_paths()
for name, path in sdk_integrations.iteritems():
lib_path = os.path.join(path, 'lib')
if os.path.exists(lib_path):
sys.path.append(lib_path)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
try:
config = get_config(parse_args=True)
except:
log.warning("Failed to load configuration")
sys.exit(2)
self._agentConfig = self._set_agent_config_hostname(config)
hostname = get_hostname(self._agentConfig)
systemStats = get_system_stats(
proc_path=self._agentConfig.get('procfs_path', '/proc').rstrip('/')
)
emitters = self._get_emitters()
# Initialize service discovery
if self._agentConfig.get('service_discovery'):
self.sd_backend = get_sd_backend(self._agentConfig)
if self.sd_backend and _is_affirmative(self._agentConfig.get('sd_jmx_enable', False)):
pipe_path = get_jmx_pipe_path()
if Platform.is_windows():
pipe_name = pipe_path.format(pipename=SD_PIPE_NAME)
else:
pipe_name = os.path.join(pipe_path, SD_PIPE_NAME)
if os.access(pipe_path, os.W_OK):
if not os.path.exists(pipe_name):
os.mkfifo(pipe_name)
self.sd_pipe = os.open(pipe_name, os.O_RDWR) # RW to avoid blocking (will only W)
# Initialize Supervisor proxy
self.supervisor_proxy = self._get_supervisor_socket(self._agentConfig)
else:
log.debug('Unable to create pipe in temporary directory. JMX service discovery disabled.')
# Load the checks.d checks
self._checksd = load_check_directory(self._agentConfig, hostname)
# Load JMX configs if available
if self._jmx_service_discovery_enabled:
self.sd_pipe_jmx_configs(hostname)
# Initialize the Collector
self.collector = Collector(self._agentConfig, emitters, systemStats, hostname)
# In developer mode, the number of runs to be included in a single collector profile
try:
self.collector_profile_interval = int(
self._agentConfig.get('collector_profile_interval', DEFAULT_COLLECTOR_PROFILE_INTERVAL))
except ValueError:
log.warn('collector_profile_interval is invalid. '
'Using default value instead (%s).' % DEFAULT_COLLECTOR_PROFILE_INTERVAL)
self.collector_profile_interval = DEFAULT_COLLECTOR_PROFILE_INTERVAL
# Configure the watchdog.
self.check_frequency = int(self._agentConfig['check_freq'])
watchdog = self._get_watchdog(self.check_frequency)
# Initialize the auto-restarter
self.restart_interval = int(self._agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
self.allow_profiling = _is_affirmative(self._agentConfig.get('allow_profiling', True))
profiled = False
collector_profiled_runs = 0
# Run the main loop.
while self.run_forever:
# Setup profiling if necessary
if self.allow_profiling and self.in_developer_mode and not profiled:
try:
profiler = AgentProfiler()
profiler.enable_profiling()
profiled = True
except Exception as e:
log.warn("Cannot enable profiler: %s" % str(e))
if self.reload_configs_flag:
if isinstance(self.reload_configs_flag, set):
self.reload_configs(checks_to_reload=self.reload_configs_flag)
else:
self.reload_configs()
# JMXFetch restarts should prompt re-piping *all* JMX configs
if self._jmx_service_discovery_enabled and \
(not self.reload_configs_flag or isinstance(self.reload_configs_flag, set)):
try:
jmx_launch = JMXFetch._get_jmx_launchtime()
if self.last_jmx_piped and self.last_jmx_piped < jmx_launch:
self.sd_pipe_jmx_configs(hostname)
except Exception as e:
log.debug("could not stat JMX lunch file: %s", e)
# Do the work. Pass `configs_reloaded` to let the collector know if it needs to
# look for the AgentMetrics check and pop it out.
self.collector.run(checksd=self._checksd,
start_event=self.start_event,
configs_reloaded=True if self.reload_configs_flag else False)
self.reload_configs_flag = False
# Look for change in the config template store.
# The self.sd_backend.reload_check_configs flag is set
# to True if a config reload is needed.
if self._agentConfig.get('service_discovery') and self.sd_backend and \
not self.sd_backend.reload_check_configs:
try:
self.sd_backend.reload_check_configs = get_config_store(
self._agentConfig).crawl_config_template()
except Exception as e:
log.warn('Something went wrong while looking for config template changes: %s' % str(e))
# Check if we should run service discovery
# The `reload_check_configs` flag can be set through the docker_daemon check or
# using ConfigStore.crawl_config_template
if self._agentConfig.get('service_discovery') and self.sd_backend and \
self.sd_backend.reload_check_configs:
self.reload_configs_flag = self.sd_backend.reload_check_configs
self.sd_backend.reload_check_configs = False
if profiled:
if collector_profiled_runs >= self.collector_profile_interval:
try:
profiler.disable_profiling()
profiled = False
collector_profiled_runs = 0
except Exception as e:
log.warn("Cannot disable profiler: %s" % str(e))
# Check if we should restart.
if self.autorestart and self._should_restart():
self._do_restart()
# Only plan for next loop if we will continue, otherwise exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
if profiled:
collector_profiled_runs += 1
log.debug("Sleeping for {0} seconds".format(self.check_frequency))
time.sleep(self.check_frequency)
# Now clean-up.
try:
CollectorStatus.remove_latest_status()
except Exception:
pass
# Explicitly kill the process, because it might be running as a daemon.
log.info("Exiting. Bye bye.")
sys.exit(0)
def _get_emitters(self):
return [http_emitter]
def _get_watchdog(self, check_freq):
watchdog = None
if self._agentConfig.get("watchdog", True):
watchdog = Watchdog.create(check_freq * WATCHDOG_MULTIPLIER)
watchdog.reset()
return watchdog
def _set_agent_config_hostname(self, agentConfig):
# Try to fetch instance Id from EC2 if not hostname has been set
# in the config file.
# DEPRECATED
if agentConfig.get('hostname') is None and agentConfig.get('use_ec2_instance_id'):
instanceId = EC2.get_instance_id(agentConfig)
if instanceId is not None:
log.info("Running on EC2, instanceId: %s" % instanceId)
agentConfig['hostname'] = instanceId
else:
log.info('Not running on EC2, using hostname to identify this server')
return agentConfig
def _get_supervisor_socket(self, agentConfig):
if Platform.is_windows():
return None
sockfile = agentConfig.get('supervisor_socket', DEFAULT_SUPERVISOR_SOCKET)
supervisor_proxy = xmlrpclib.ServerProxy(
'http://127.0.0.1',
transport=supervisor.xmlrpc.SupervisorTransport(
None, None, serverurl="unix://{socket}".format(socket=sockfile))
)
return supervisor_proxy
@property
def _jmx_service_discovery_enabled(self):
return self.sd_pipe is not None
def _submit_jmx_service_discovery(self, jmx_sd_configs):
if not jmx_sd_configs or not self.sd_pipe:
return
if self.supervisor_proxy is not None:
try:
jmx_state = self.supervisor_proxy.supervisor.getProcessInfo(JMX_SUPERVISOR_ENTRY)
log.debug("Current JMX check state: %s", jmx_state['statename'])
except Exception as e:
log.exception("Cannot submit JMX autodiscovery configurations. Unable to get JMXFetch process state from supervisor: %s", e)
return
# restart jmx if stopped
if jmx_state['statename'] in ['STOPPED', 'EXITED', 'FATAL'] and self._agentConfig.get('sd_jmx_enable'):
self.supervisor_proxy.supervisor.startProcess(JMX_SUPERVISOR_ENTRY)
time.sleep(JMX_GRACE_SECS)
else:
log.debug("Unable to automatically start jmxfetch on Windows via supervisor.")
buffer = ""
names = []
try:
for name, yaml in jmx_sd_configs.iteritems():
buffer += SD_CONFIG_SEP
buffer += "# {}\n".format(name)
buffer += yaml
names.append(name)
if buffer:
# will block if len(buffer) > OS pipe buffer.
# JMX will unblock when it reads on the other end.
os.write(self.sd_pipe, buffer)
os.write(self.sd_pipe, SD_CONFIG_TERM)
self.last_jmx_piped = time.time()
except Exception as e:
log.exception("unable to submit YAML via pipe: %s", e)
else:
log.info("JMX SD Configs submitted via named pipe successfully: %s", names)
def _should_restart(self):
if time.time() - self.agent_start > self.restart_interval:
return True
return False
def _do_restart(self):
log.info("Running an auto-restart.")
if self.collector:
self.collector.stop()
sys.exit(AgentSupervisor.RESTART_EXIT_STATUS)
def main():
options, args = get_parsed_args()
agentConfig = get_config(options=options)
autorestart = agentConfig.get('autorestart', False)
hostname = get_hostname(agentConfig)
in_developer_mode = agentConfig.get('developer_mode')
COMMANDS_AGENT = [
'start',
'stop',
'restart',
'status',
'foreground',
]
COMMANDS_NO_AGENT = [
'info',
'check',
'configcheck',
'jmx',
'flare',
]
COMMANDS = COMMANDS_AGENT + COMMANDS_NO_AGENT
if len(args) < 1:
sys.stderr.write("Usage: %s %s\n" % (sys.argv[0], "|".join(COMMANDS)))
return 2
command = args[0]
if command not in COMMANDS:
sys.stderr.write("Unknown command: %s\n" % command)
return 3
# TODO: actually kill the start/stop/restart/status command for 5.11
if command in ['start', 'stop', 'restart', 'status'] and not in_developer_mode:
logging.error('Please use supervisor to manage the agent')
return 1
if command in COMMANDS_AGENT:
agent = Agent(PidFile(PID_NAME, PID_DIR).get_path(), autorestart, in_developer_mode=in_developer_mode)
if 'start' == command:
log.info('Start daemon')
agent.start()
elif 'stop' == command:
log.info('Stop daemon')
agent.stop()
elif 'restart' == command:
log.info('Restart daemon')
agent.restart()
elif 'status' == command:
agent.status()
elif 'info' == command:
return Agent.info(verbose=options.verbose)
elif 'foreground' == command:
log.info('Agent version %s' % get_version())
if autorestart:
# Set-up the supervisor callbacks and fork it.
logging.info('Running Agent with auto-restart ON')
def child_func():
agent.start(foreground=True)
def parent_func():
agent.start_event = False
AgentSupervisor.start(parent_func, child_func)
else:
# Run in the standard foreground.
agent.start(foreground=True)
elif 'check' == command:
if len(args) < 2:
sys.stderr.write(
"Usage: %s check <check_name> [check_rate]\n"
"Add check_rate as last argument to compute rates\n"
% sys.argv[0]
)
return 1
check_name = args[1]
try:
import checks.collector
# Try the old-style check first
print getattr(checks.collector, check_name)(log).check(agentConfig)
except Exception:
# If not an old-style check, try checks.d
checks = load_check_directory(agentConfig, hostname)
for check in checks['initialized_checks']:
if check.name == check_name:
if in_developer_mode:
check.run = AgentProfiler.wrap_profiling(check.run)
cs = Collector.run_single_check(check, verbose=True)
print CollectorStatus.render_check_status(cs)
if len(args) == 3 and args[2] == 'check_rate':
print "Running 2nd iteration to capture rate metrics"
time.sleep(1)
cs = Collector.run_single_check(check, verbose=True)
print CollectorStatus.render_check_status(cs)
else:
print "Check has run only once, if some metrics are missing you can run the command again with the 'check_rate' argument appended at the end to see any other metrics if available."
check.stop()
elif 'configcheck' == command or 'configtest' == command:
sd_configcheck(agentConfig)
return configcheck()
elif 'jmx' == command:
jmx_command(args[1:], agentConfig)
elif 'flare' == command:
Flare.check_user_rights()
case_id = int(args[1]) if len(args) > 1 else None
f = Flare(True, case_id)
f.collect()
try:
f.upload()
except Exception as e:
print 'The upload failed:\n{0}'.format(str(e))
return 0
if __name__ == '__main__':
try:
sys.exit(main())
except StandardError:
# Try our best to log the error.
try:
log.exception("Uncaught error running the Agent")
except Exception:
pass
raise