def testLoad(self):
global logger
load = Load(logger)
res = load.check({'system_stats': get_system_stats()})
cores = int(get_system_stats().get('cpuCores'))
assert 'system.load.1' in res
assert 'system.load.norm.1' in res
assert abs(res['system.load.1'] - cores * res['system.load.norm.1']) <= 0.1, (res['system.load.1'], cores * res['system.load.norm.1'])
# same test but without cpu count, no normalized load sent.
res = load.check({})
assert 'system.load.1' in res
assert 'system.load.norm.1' not in res
def testLoad(self):
global logger
load = Load(logger)
res = load.check({'system_stats': get_system_stats()})
assert 'system.load.1' in res
if Platform.is_linux():
cores = int(get_system_stats().get('cpuCores'))
assert 'system.load.norm.1' in res
assert abs(res['system.load.1'] - cores * res['system.load.norm.1']) <= 0.1, (res['system.load.1'], cores * res['system.load.norm.1'])
# same test but without cpu count, no normalized load sent.
res = load.check({})
assert 'system.load.1' in res
assert 'system.load.norm.1' not in res
def testLoad(self):
global logger
load = Load(logger)
res = load.check({"system_stats": get_system_stats()})
assert "system.load.1" in res
if Platform.is_linux():
cores = int(get_system_stats().get("cpuCores"))
assert "system.load.norm.1" in res
assert abs(res["system.load.1"] - cores * res["system.load.norm.1"]) <= 0.1, (
res["system.load.1"],
cores * res["system.load.norm.1"],
)
# same test but without cpu count, no normalized load sent.
res = load.check({})
assert "system.load.1" in res
assert "system.load.norm.1" not in res
def _build_payload(self, start_event=True):
"""
Return an dictionary that contains all of the generic payload data.
"""
now = time.time()
payload = {
'collection_timestamp': now,
'os' : self.os,
'python': sys.version,
'agentVersion' : self.agentConfig['version'],
'apiKey': self.agentConfig['api_key'],
'events': {},
'metrics': [],
'service_checks': [],
'resources': {},
'internalHostname' : get_hostname(self.agentConfig),
'uuid' : get_uuid(),
'host-tags': {},
}
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
'timestamp': now,
'event_type':'Agent Startup',
'msg_text': 'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._is_first_run() or self._should_send_metadata():
payload['systemStats'] = get_system_stats()
payload['meta'] = self._get_metadata()
self.metadata_cache = payload['meta']
# Add static tags from the configuration file
host_tags = []
if self.agentConfig['tags'] is not None:
host_tags.extend([unicode(tag.strip()) for tag in self.agentConfig['tags'].split(",")])
if self.agentConfig['collect_ec2_tags']:
host_tags.extend(EC2.get_tags())
if host_tags:
payload['host-tags']['system'] = host_tags
GCE_tags = GCE.get_tags()
if GCE_tags is not None:
payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
# Log the metadata on the first run
if self._is_first_run():
log.info("Hostnames: %s, tags: %s" % (repr(self.metadata_cache), payload['host-tags']))
return payload
def _build_payload(self, start_event=True):
"""
Return an dictionary that contains all of the generic payload data.
"""
now = time.time()
payload = {
'collection_timestamp': now,
'os': self.os,
'python': sys.version,
'agentVersion': self.agentConfig['version'],
'apiKey': self.agentConfig['api_key'],
'events': {},
'metrics': [],
'resources': {},
'internalHostname': get_hostname(self.agentConfig),
'uuid': get_uuid(),
}
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{
'api_key':
self.agentConfig['api_key'],
'host':
payload['internalHostname'],
'timestamp':
now,
'event_type':
'Agent Startup',
'msg_text':
'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._is_first_run() or self._should_send_metadata():
payload['systemStats'] = get_system_stats()
payload['meta'] = self._get_metadata()
self.metadata_cache = payload['meta']
# Add static tags from the configuration file
if self.agentConfig['tags'] is not None:
payload['tags'] = self.agentConfig['tags']
# Log the metadata on the first run
if self._is_first_run():
if self.agentConfig['tags'] is not None:
log.info(u"Hostnames: %s, tags: %s" \
% (repr(self.metadata_cache),
self.agentConfig['tags']))
else:
log.info(u"Hostnames: %s" % repr(self.metadata_cache))
return payload
def _build_payload(self, start_event=True):
"""
Return an dictionary that contains all of the generic payload data.
"""
now = time.time()
payload = {
"collection_timestamp": now,
"os": self.os,
"python": sys.version,
"agentVersion": self.agentConfig["version"],
"apiKey": self.agentConfig["api_key"],
"events": {},
"metrics": [],
"resources": {},
"internalHostname": get_hostname(self.agentConfig),
"uuid": get_uuid(),
"host-tags": {},
}
# Include system stats on first postback
if start_event and self._is_first_run():
payload["systemStats"] = self.agentConfig.get("system_stats", {})
# Also post an event in the newsfeed
payload["events"]["System"] = [
{
"api_key": self.agentConfig["api_key"],
"host": payload["internalHostname"],
"timestamp": now,
"event_type": "Agent Startup",
"msg_text": "Version %s" % get_version(),
}
]
# Periodically send the host metadata.
if self._is_first_run() or self._should_send_metadata():
payload["systemStats"] = get_system_stats()
payload["meta"] = self._get_metadata()
self.metadata_cache = payload["meta"]
# Add static tags from the configuration file
host_tags = []
if self.agentConfig["tags"] is not None:
host_tags.extend([unicode(tag.strip()) for tag in self.agentConfig["tags"].split(",")])
if self.agentConfig["collect_ec2_tags"]:
host_tags.extend(EC2.get_tags())
if host_tags:
payload["host-tags"]["system"] = host_tags
# Log the metadata on the first run
if self._is_first_run():
log.info(u"Hostnames: %s, tags: %s" % (repr(self.metadata_cache), payload["host-tags"]))
return payload
def run(self):
emitters = self.get_emitters()
systemStats = get_system_stats()
collector = Collector(self.config, emitters, systemStats)
# Load the checks.d checks
checksd = load_check_directory(self.config)
# Main agent loop will run until interrupted
while self.running:
collector.run(checksd=checksd)
time.sleep(self.config['check_freq'])
def run(self):
emitters = self.get_emitters()
systemStats = get_system_stats()
collector = Collector(self.config, emitters, systemStats)
# Load the checks.d checks
checksd = load_check_directory(self.config)
# Main agent loop will run until interrupted
while self.running:
collector.run(checksd=checksd)
time.sleep(self.config['check_freq'])
def run(self):
log.debug("Windows Service - Starting collector")
emitters = self.get_emitters()
systemStats = get_system_stats()
self.collector = Collector(self.config, emitters, systemStats)
# Load the checks.d checks
checksd = load_check_directory(self.config)
# Main agent loop will run until interrupted
while self.running:
self.collector.run(checksd=checksd, start_event=self.start_event)
time.sleep(self.config['check_freq'])
def run(self):
log.debug("Windows Service - Starting collector")
emitters = self.get_emitters()
systemStats = get_system_stats()
self.collector = Collector(self.config, emitters, systemStats)
# Load the checks.d checks
checksd = load_check_directory(self.config)
# Main agent loop will run until interrupted
while self.running:
self.collector.run(checksd=checksd, start_event=self.start_event)
time.sleep(self.config['check_freq'])
def run(self):
from config import initialize_logging; initialize_logging('windows_collector')
log.debug("Windows Service - Starting collector")
emitters = self.get_emitters()
systemStats = get_system_stats()
self.collector = Collector(self.config, emitters, systemStats, self.hostname)
# Load the checks.d checks
checksd = load_check_directory(self.config, self.hostname)
# Main agent loop will run until interrupted
while self.running:
self.collector.run(checksd=checksd, start_event=self.start_event)
time.sleep(self.config['check_freq'])
def run(self, agentConfig=None, run_forever=True):
"""Main loop of the collector"""
agentLogger = logging.getLogger('agent')
systemStats = get_system_stats()
if agentConfig is None:
agentConfig = get_config()
# Load the checks.d checks
checksd = load_check_directory(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()
if instanceId is not None:
agentLogger.info("Running on EC2, instanceId: %s" % instanceId)
agentConfig['hostname'] = instanceId
else:
agentLogger.info(
'Not running on EC2, using hostname to identify this server'
)
emitters = [http_emitter]
for emitter_spec in [
s.strip()
for s in agentConfig.get('custom_emitters', '').split(',')
]:
if len(emitter_spec) == 0: continue
emitters.append(modules.load(emitter_spec, 'emitter'))
check_freq = int(agentConfig['check_freq'])
# Checks instance
collector = Collector(agentConfig, emitters, systemStats)
# Watchdog
watchdog = None
if agentConfig.get("watchdog", True):
watchdog = Watchdog(check_freq * WATCHDOG_MULTIPLIER)
watchdog.reset()
# Main loop
while run_forever:
collector.run(checksd=checksd)
if watchdog is not None:
watchdog.reset()
time.sleep(check_freq)
def run(self):
emitters = self.get_emitters();
chk = checks(self.config, emitters)
# Load the checks.d checks
checksd = load_check_directory(self.config)
# Main agent loop will run until interrupted
systemStats = get_system_stats()
chk.doChecks(True, systemStats, checksd)
while self.running:
chk.doChecks(checksd=checksd)
time.sleep(self.config['check_freq'])
def run(self, agentConfig=None, run_forever=True):
"""Main loop of the collector"""
agentLogger = logging.getLogger('agent')
systemStats = get_system_stats()
agentLogger.debug('System Properties: ' + str(systemStats))
if agentConfig is None:
agentConfig = get_config()
# Load the checks.d checks
checksd = load_check_directory(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()
if instanceId is not None:
agentLogger.info("Running on EC2, instanceId: %s" % instanceId)
agentConfig['hostname'] = instanceId
else:
agentLogger.info('Not running on EC2, using hostname to identify this server')
emitters = [http_emitter]
for emitter_spec in [s.strip() for s in agentConfig.get('custom_emitters', '').split(',')]:
if len(emitter_spec) == 0: continue
emitters.append(modules.load(emitter_spec, 'emitter'))
check_freq = int(agentConfig['check_freq'])
# Checks instance
c = checks(agentConfig, emitters)
# Watchdog
watchdog = None
if agentConfig.get("watchdog", True):
watchdog = Watchdog(check_freq * WATCHDOG_MULTIPLIER)
watchdog.reset()
# Run checks once, to get once-in-a-run data
c.doChecks(True, systemStats, checksd)
# Main loop
while run_forever:
if watchdog is not None:
watchdog.reset()
time.sleep(check_freq)
c.doChecks(checksd=checksd)
def _build_payload(self, start_event=True):
"""
Return an dictionary that contains all of the generic payload data.
"""
now = time.time()
payload = {
'collection_timestamp': now,
'os' : self.os,
'python': sys.version,
'agentVersion' : self.agentConfig['version'],
'apiKey': self.agentConfig['api_key'],
'events': {},
'metrics': [],
'resources': {},
'internalHostname' : get_hostname(self.agentConfig),
'uuid' : get_uuid(),
}
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
'timestamp': now,
'event_type':'Agent Startup',
'msg_text': 'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._is_first_run() or self._should_send_metadata():
payload['systemStats'] = get_system_stats()
payload['meta'] = self._get_metadata()
self.metadata_cache = payload['meta']
# Add static tags from the configuration file
if self.agentConfig['tags'] is not None:
payload['tags'] = self.agentConfig['tags']
# Log the metadata on the first run
if self._is_first_run():
if self.agentConfig['tags'] is not None:
log.info(u"Hostnames: %s, tags: %s" \
% (repr(self.metadata_cache),
self.agentConfig['tags']))
else:
log.info(u"Hostnames: %s" % repr(self.metadata_cache))
return payload
def run(self):
from config import initialize_logging
initialize_logging('windows_collector')
log.debug("Windows Service - Starting collector")
set_win32_requests_ca_bundle_path()
emitters = self.get_emitters()
systemStats = get_system_stats()
self.collector = Collector(self.config, emitters, systemStats,
self.hostname)
in_developer_mode = self.config.get('developer_mode')
# In developer mode, the number of runs to be included in a single collector profile
collector_profile_interval = self.config.get(
'collector_profile_interval', DEFAULT_COLLECTOR_PROFILE_INTERVAL)
profiled = False
collector_profiled_runs = 0
# Load the checks.d checks
checksd = load_check_directory(self.config, self.hostname)
# Main agent loop will run until interrupted
while self.running:
if self._heartbeat:
self._heartbeat.send(0)
if 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))
self.collector.run(checksd=checksd)
if profiled:
if collector_profiled_runs >= 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))
else:
collector_profiled_runs += 1
time.sleep(self.config['check_freq'])
def run(self):
from config import initialize_logging; initialize_logging('windows_collector')
log.debug("Windows Service - Starting collector")
emitters = self.get_emitters()
systemStats = get_system_stats()
self.collector = Collector(self.config, emitters, systemStats, self.hostname)
# Load the checks.d checks
checksd = load_check_directory(self.config, self.hostname)
# Main agent loop will run until interrupted
while self.running:
if self._heartbeat:
self._heartbeat.send(0)
self.collector.run(checksd=checksd)
time.sleep(self.config['check_freq'])
def run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
agentConfig = self._set_agent_config_hostname(config)
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
self.collector = Collector(agentConfig, emitters, systemStats)
# Load the checks.d checks
checksd = load_check_directory(agentConfig)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Run the main loop.
while self.run_forever:
# Do the work.
self.collector.run(checksd=checksd)
# Only plan for the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
time.sleep(check_frequency)
# Now clean-up.
try:
CollectorStatus.remove_latest_status()
except:
pass
# Explicitly kill the process, because it might be running
# as a daemon.
log.info("Exiting. Bye bye.")
sys.exit(0)
def run(self):
from config import initialize_logging
initialize_logging('windows_collector')
log.debug("Windows Service - Starting collector")
set_win32_requests_ca_bundle_path()
emitters = self.get_emitters()
systemStats = get_system_stats()
self.collector = Collector(self.config, emitters, systemStats, self.hostname)
in_developer_mode = self.config.get('developer_mode')
# In developer mode, the number of runs to be included in a single collector profile
collector_profile_interval = self.config.get('collector_profile_interval',
DEFAULT_COLLECTOR_PROFILE_INTERVAL)
profiled = False
collector_profiled_runs = 0
# Load the checks.d checks
checksd = load_check_directory(self.config, self.hostname)
# Main agent loop will run until interrupted
while self.running:
if self._heartbeat:
self._heartbeat.send(0)
if 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))
self.collector.run(checksd=checksd)
if profiled:
if collector_profiled_runs >= 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))
else:
collector_profiled_runs += 1
time.sleep(self.config['check_freq'])
def run(self):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
agentConfig = self._set_agent_config_hostname(get_config())
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
self.collector = Collector(agentConfig, emitters, systemStats)
# Load the checks.d checks
checksd = load_check_directory(agentConfig)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Run the main loop.
while self.run_forever:
# Do the work.
self.collector.run(checksd=checksd)
# Only plan for the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
time.sleep(check_frequency)
# Now clean-up.
try:
CollectorStatus.remove_latest_status()
except:
pass
# Explicitly kill the process, because it might be running
# as a daemon.
agent_logger.info("Exiting. Bye bye.")
sys.exit(0)
def run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
agentConfig = self._set_agent_config_hostname(config)
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
# Load the checks.d checks
checksd = load_check_directory(agentConfig)
self.collector = Collector(agentConfig, emitters, systemStats)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Initialize the auto-restarter
self.restart_interval = int(agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
# Run the main loop.
while self.run_forever:
# Do the work.
self.collector.run(checksd=checksd, start_event=self.start_event)
# Check if we should restart.
if self.autorestart and self._should_restart():
self._do_restart()
# Only plan for the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
time.sleep(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 _populate_payload_metadata(self, payload, check_statuses, start_event=True):
"""
Periodically populate the payload with metadata related to the system, host, and/or checks.
"""
now = time.time()
# Include system stats on first postback
if start_event and self._is_first_run():
payload["systemStats"] = self.agentConfig.get("system_stats", {})
# Also post an event in the newsfeed
payload["events"]["System"] = [
{
"api_key": self.agentConfig["api_key"],
"host": payload["internalHostname"],
"timestamp": now,
"event_type": "Agent Startup",
"msg_text": "Version %s" % get_version(),
}
]
# Periodically send the host metadata.
if self._should_send_additional_data("host_metadata"):
# gather metadata with gohai
try:
if not Platform.is_windows():
command = "gohai"
else:
command = "gohai\gohai.exe"
gohai_metadata, gohai_err, _ = get_subprocess_output([command], log)
payload["gohai"] = gohai_metadata
if gohai_err:
log.warning("GOHAI LOG | {0}".format(gohai_err))
except OSError as e:
if e.errno == 2: # file not found, expected when install from source
log.info("gohai file not found")
else:
raise e
except Exception as e:
log.warning("gohai command failed with error %s" % str(e))
payload["systemStats"] = get_system_stats()
payload["meta"] = self._get_hostname_metadata()
self.hostname_metadata_cache = payload["meta"]
# Add static tags from the configuration file
host_tags = []
if self.agentConfig["tags"] is not None:
host_tags.extend([unicode(tag.strip()) for tag in self.agentConfig["tags"].split(",")])
if self.agentConfig["collect_ec2_tags"]:
host_tags.extend(EC2.get_tags(self.agentConfig))
if host_tags:
payload["host-tags"]["system"] = host_tags
# If required by the user, let's create the dd_check:xxx host tags
if self.agentConfig["create_dd_check_tags"]:
app_tags_list = [DD_CHECK_TAG.format(c.name) for c in self.initialized_checks_d]
app_tags_list.extend([DD_CHECK_TAG.format(cname) for cname in JMXFiles.get_jmx_appnames()])
if "system" not in payload["host-tags"]:
payload["host-tags"]["system"] = []
payload["host-tags"]["system"].extend(app_tags_list)
GCE_tags = GCE.get_tags(self.agentConfig)
if GCE_tags is not None:
payload["host-tags"][GCE.SOURCE_TYPE_NAME] = GCE_tags
# Log the metadata on the first run
if self._is_first_run():
log.info("Hostnames: %s, tags: %s" % (repr(self.hostname_metadata_cache), payload["host-tags"]))
# Periodically send extra hosts metadata (vsphere)
# Metadata of hosts that are not the host where the agent runs, not all the checks use
# that
external_host_tags = []
if self._should_send_additional_data("external_host_tags"):
for check in self.initialized_checks_d:
try:
getter = getattr(check, "get_external_host_tags")
check_tags = getter()
external_host_tags.extend(check_tags)
except AttributeError:
pass
if external_host_tags:
payload["external_host_tags"] = external_host_tags
# Periodically send agent_checks metadata
if self._should_send_additional_data("agent_checks"):
# Add agent checks statuses and error/warning messages
agent_checks = []
for check in check_statuses:
if check.instance_statuses is not None:
for i, instance_status in enumerate(check.instance_statuses):
agent_checks.append(
(
check.name,
check.source_type_name,
instance_status.instance_id,
instance_status.status,
# put error message or list of warning messages in the same field
# it will be handled by the UI
instance_status.error or instance_status.warnings or "",
check.service_metadata[i],
)
)
else:
agent_checks.append(
(
check.name,
check.source_type_name,
"initialization",
check.status,
repr(check.init_failed_error),
)
)
payload["agent_checks"] = agent_checks
payload["meta"] = self.hostname_metadata_cache # add hostname metadata
def run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
agentConfig = self._set_agent_config_hostname(config)
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
# Load the checks.d checks
checksd = load_check_directory(agentConfig)
self.collector = Collector(agentConfig, emitters, systemStats)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Initialize the auto-restarter
self.restart_interval = int(
agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
# Run the main loop.
while self.run_forever:
# enable profiler if needed
profiled = False
if agentConfig.get(
'profile',
False) and agentConfig.get('profile').lower() == 'yes':
try:
import cProfile
profiler = cProfile.Profile()
profiled = True
profiler.enable()
log.debug("Agent profiling is enabled")
except Exception:
log.warn("Cannot enable profiler")
# Do the work.
self.collector.run(checksd=checksd, start_event=self.start_event)
# disable profiler and printout stats to stdout
if agentConfig.get('profile', False) and agentConfig.get(
'profile').lower() == 'yes' and profiled:
try:
profiler.disable()
import pstats
from cStringIO import StringIO
s = StringIO()
ps = pstats.Stats(profiler,
stream=s).sort_stats("cumulative")
ps.print_stats()
log.debug(s.getvalue())
except Exception:
log.warn("Cannot disable profiler")
# Check if we should restart.
if self.autorestart and self._should_restart():
self._do_restart()
# Only plan for the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
time.sleep(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 run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
agentConfig = self._set_agent_config_hostname(config)
hostname = get_hostname(agentConfig)
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
# Load the checks.d checks
checksd = load_check_directory(agentConfig, hostname)
self.collector = Collector(agentConfig, emitters, systemStats, hostname)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Initialize the auto-restarter
self.restart_interval = int(agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
# Run the main loop.
while self.run_forever:
# enable profiler if needed
profiled = False
if agentConfig.get('profile', False) and agentConfig.get('profile').lower() == 'yes':
try:
import cProfile
profiler = cProfile.Profile()
profiled = True
profiler.enable()
log.debug("Agent profiling is enabled")
except Exception:
log.warn("Cannot enable profiler")
# Do the work.
self.collector.run(checksd=checksd, start_event=self.start_event)
# disable profiler and printout stats to stdout
if agentConfig.get('profile', False) and agentConfig.get('profile').lower() == 'yes' and profiled:
try:
profiler.disable()
import pstats
from cStringIO import StringIO
s = StringIO()
ps = pstats.Stats(profiler, stream=s).sort_stats("cumulative")
ps.print_stats()
log.debug(s.getvalue())
except Exception:
log.warn("Cannot disable profiler")
# Check if we should restart.
if self.autorestart and self._should_restart():
self._do_restart()
# Only plan for the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
time.sleep(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 run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
self._agentConfig = self._set_agent_config_hostname(config)
hostname = get_hostname(self._agentConfig)
systemStats = get_system_stats()
emitters = self._get_emitters()
# Load the checks.d checks
self._checksd = load_check_directory(self._agentConfig, 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
self.collector_profile_interval = self._agentConfig.get('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()
profiled = False
collector_profiled_runs = 0
# Run the main loop.
while self.run_forever:
log.debug("Found {num_checks} checks".format(num_checks=len(self._checksd['initialized_checks'])))
# Setup profiling if necessary
if 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))
# Do the work.
self.collector.run(checksd=self._checksd,
start_event=self.start_event,
configs_reloaded=self.configs_reloaded)
if self.configs_reloaded:
self.configs_reloaded = 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.info("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 run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
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 _is_affirmative(self._agentConfig.get('sd_jmx_enable')):
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:
jmx_sd_configs = generate_jmx_configs(self._agentConfig, hostname)
if jmx_sd_configs:
self._submit_jmx_service_discovery(jmx_sd_configs)
# 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 = 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()
# 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 run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
self._agentConfig = self._set_agent_config_hostname(config)
hostname = get_hostname(self._agentConfig)
systemStats = get_system_stats()
emitters = self._get_emitters()
# Initialize service discovery
if self._agentConfig.get("service_discovery"):
self.sd_backend = get_sd_backend(self._agentConfig)
# Load the checks.d checks
self._checksd = load_check_directory(self._agentConfig, 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
self.collector_profile_interval = self._agentConfig.get(
"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()
profiled = False
collector_profiled_runs = 0
# Run the main loop.
while self.run_forever:
log.debug("Found {num_checks} checks".format(num_checks=len(self._checksd["initialized_checks"])))
# Setup profiling if necessary
if 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))
# Do the work.
self.collector.run(
checksd=self._checksd, start_event=self.start_event, configs_reloaded=self.configs_reloaded
)
# This flag is used to know if the check configs have been reloaded at the current
# run of the agent yet or not. It's used by the collector to know if it needs to
# look for the AgentMetrics check and pop it out.
# See: https://github.com/DataDog/dd-agent/blob/5.6.x/checks/collector.py#L265-L272
self.configs_reloaded = 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()
self.configs_reloaded = True
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 _build_payload(self, start_event=True):
"""
Return an dictionary that contains all of the generic payload data.
"""
now = time.time()
payload = {
'collection_timestamp': now,
'os': self.os,
'python': sys.version,
'agentVersion': self.agentConfig['version'],
'apiKey': self.agentConfig['api_key'],
'events': {},
'metrics': [],
'service_checks': [],
'resources': {},
'internalHostname': self.hostname,
'uuid': get_uuid(),
'host-tags': {},
'external_host_tags': {}
}
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{
'api_key':
self.agentConfig['api_key'],
'host':
payload['internalHostname'],
'timestamp':
now,
'event_type':
'Agent Startup',
'msg_text':
'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._should_send_additional_data('metadata'):
# gather metadata with gohai
try:
if get_os() != 'windows':
command = "gohai"
else:
command = "gohai\gohai.exe"
gohai_metadata = subprocess.Popen(
[command], stdout=subprocess.PIPE).communicate()[0]
payload['gohai'] = gohai_metadata
except OSError as e:
if e.errno == 2: # file not found, expected when install from source
log.info("gohai file not found")
else:
raise e
except Exception as e:
log.warning("gohai command failed with error %s" % str(e))
payload['systemStats'] = get_system_stats()
payload['meta'] = self._get_metadata()
self.metadata_cache = payload['meta']
# Add static tags from the configuration file
host_tags = []
if self.agentConfig['tags'] is not None:
host_tags.extend([
unicode(tag.strip())
for tag in self.agentConfig['tags'].split(",")
])
if self.agentConfig['collect_ec2_tags']:
host_tags.extend(EC2.get_tags(self.agentConfig))
if host_tags:
payload['host-tags']['system'] = host_tags
GCE_tags = GCE.get_tags(self.agentConfig)
if GCE_tags is not None:
payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
# Log the metadata on the first run
if self._is_first_run():
log.info("Hostnames: %s, tags: %s" %
(repr(self.metadata_cache), payload['host-tags']))
# Periodically send extra hosts metadata (vsphere)
# Metadata of hosts that are not the host where the agent runs, not all the checks use
# that
external_host_tags = []
if self._should_send_additional_data('external_host_tags'):
for check in self.initialized_checks_d:
try:
getter = getattr(check, 'get_external_host_tags')
check_tags = getter()
external_host_tags.extend(check_tags)
except AttributeError:
pass
if external_host_tags:
payload['external_host_tags'] = external_host_tags
return payload
def _build_payload(self, start_event=True):
"""
Return an dictionary that contains all of the generic payload data.
"""
now = time.time()
payload = {
'collection_timestamp': now,
'os' : self.os,
'python': sys.version,
'agentVersion' : self.agentConfig['version'],
'apiKey': self.agentConfig['api_key'],
'events': {},
'metrics': [],
'service_checks': [],
'resources': {},
'internalHostname' : self.hostname,
'uuid' : get_uuid(),
'host-tags': {},
'external_host_tags': {}
}
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
'timestamp': now,
'event_type':'Agent Startup',
'msg_text': 'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._should_send_additional_data('metadata'):
# gather metadata with gohai
try:
if get_os() != 'windows':
command = "gohai"
else:
command = "gohai\gohai.exe"
gohai_metadata = subprocess.Popen(
[command], stdout=subprocess.PIPE
).communicate()[0]
payload['gohai'] = gohai_metadata
except OSError as e:
if e.errno == 2: # file not found, expected when install from source
log.info("gohai file not found")
else:
raise e
except Exception as e:
log.warning("gohai command failed with error %s" % str(e))
payload['systemStats'] = get_system_stats()
payload['meta'] = self._get_metadata()
self.metadata_cache = payload['meta']
# Add static tags from the configuration file
host_tags = []
if self.agentConfig['tags'] is not None:
host_tags.extend([unicode(tag.strip()) for tag in self.agentConfig['tags'].split(",")])
if self.agentConfig['collect_ec2_tags']:
host_tags.extend(EC2.get_tags(self.agentConfig))
if host_tags:
payload['host-tags']['system'] = host_tags
GCE_tags = GCE.get_tags(self.agentConfig)
if GCE_tags is not None:
payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
# Log the metadata on the first run
if self._is_first_run():
log.info("Hostnames: %s, tags: %s" % (repr(self.metadata_cache), payload['host-tags']))
# Periodically send extra hosts metadata (vsphere)
# Metadata of hosts that are not the host where the agent runs, not all the checks use
# that
external_host_tags = []
if self._should_send_additional_data('external_host_tags'):
for check in self.initialized_checks_d:
try:
getter = getattr(check, 'get_external_host_tags')
check_tags = getter()
external_host_tags.extend(check_tags)
except AttributeError:
pass
if external_host_tags:
payload['external_host_tags'] = external_host_tags
return payload
def run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
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)
# Load the checks.d checks
self._checksd = load_check_directory(self._agentConfig, 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
self.collector_profile_interval = self._agentConfig.get(
'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()
profiled = False
collector_profiled_runs = 0
# Run the main loop.
while self.run_forever:
log.debug("Found {num_checks} checks".format(
num_checks=len(self._checksd['initialized_checks'])))
# Setup profiling if necessary
if 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))
# Do the work.
self.collector.run(checksd=self._checksd,
start_event=self.start_event,
configs_reloaded=self.configs_reloaded)
# This flag is used to know if the check configs have been reloaded at the current
# run of the agent yet or not. It's used by the collector to know if it needs to
# look for the AgentMetrics check and pop it out.
# See: https://github.com/DataDog/dd-agent/blob/5.6.x/checks/collector.py#L265-L272
self.configs_reloaded = 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()
self.configs_reloaded = True
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 _populate_payload_metadata(self, payload, check_statuses, start_event=True):
"""
Periodically populate the payload with metadata related to the system, host, and/or checks.
"""
now = time.time()
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{
'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
'timestamp': now,
'event_type':'Agent Startup',
'msg_text': 'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._should_send_additional_data('host_metadata'):
# gather metadata with gohai
gohai_metadata = self._run_gohai_metadata()
if gohai_metadata:
payload['gohai'] = gohai_metadata
payload['systemStats'] = get_system_stats(
proc_path=self.agentConfig.get('procfs_path', '/proc').rstrip('/')
)
payload['meta'] = self._get_hostname_metadata()
self.hostname_metadata_cache = payload['meta']
# Add static tags from the configuration file
host_tags = []
if self.agentConfig['tags'] is not None:
host_tags.extend([unicode(tag.strip())
for tag in self.agentConfig['tags'].split(",")])
if self.agentConfig['collect_ec2_tags']:
host_tags.extend(EC2.get_tags(self.agentConfig))
if host_tags:
payload['host-tags']['system'] = host_tags
# If required by the user, let's create the dd_check:xxx host tags
if self.agentConfig['create_dd_check_tags']:
app_tags_list = [DD_CHECK_TAG.format(c.name) for c in self.initialized_checks_d]
app_tags_list.extend([DD_CHECK_TAG.format(cname) for cname
in JMXFiles.get_jmx_appnames()])
if 'system' not in payload['host-tags']:
payload['host-tags']['system'] = []
payload['host-tags']['system'].extend(app_tags_list)
GCE_tags = GCE.get_tags(self.agentConfig)
if GCE_tags is not None:
payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
# Log the metadata on the first run
if self._is_first_run():
log.info("Hostnames: %s, tags: %s" %
(repr(self.hostname_metadata_cache), payload['host-tags']))
# Periodically send extra hosts metadata (vsphere)
# Metadata of hosts that are not the host where the agent runs, not all the checks use
# that
external_host_tags = []
if self._should_send_additional_data('external_host_tags'):
for check in self.initialized_checks_d:
try:
getter = getattr(check, 'get_external_host_tags')
check_tags = getter()
external_host_tags.extend(check_tags)
except AttributeError:
pass
if external_host_tags:
payload['external_host_tags'] = external_host_tags
# Periodically send agent_checks metadata
if self._should_send_additional_data('agent_checks'):
# Add agent checks statuses and error/warning messages
agent_checks = []
for check in check_statuses:
if check.instance_statuses is not None:
for i, instance_status in enumerate(check.instance_statuses):
agent_checks.append(
(
check.name, check.source_type_name,
instance_status.instance_id,
instance_status.status,
# put error message or list of warning messages in the same field
# it will be handled by the UI
instance_status.error or instance_status.warnings or "",
check.service_metadata[i]
)
)
else:
agent_checks.append(
(
check.name, check.source_type_name,
"initialization",
check.status, repr(check.init_failed_error)
)
)
payload['agent_checks'] = agent_checks
payload['meta'] = self.hostname_metadata_cache # add hostname metadata
def run(self, config=None):
signal.signal(signal.SIGTERM, self._handle_sigterm)
signal.signal(signal.SIGUSR1, self._handle_sigusr1)
signal.signal(signal.SIGINT, self._handle_sigterm)
signal.signal(signal.SIGHUP, self._handle_sighup)
CollectorStatus().persist()
if not config:
config = get_config(parse_args=True)
self._agentConfig = self._set_agent_config_hostname(config)
hostname = get_hostname(self._agentConfig)
systemStats = get_system_stats()
emitters = self._get_emitters()
self._checksd = load_check_directory(self._agentConfig, hostname)
self.collector = Collector(self._agentConfig, emitters, systemStats, hostname)
self.collector_profile_interval = self._agentConfig.get('collector_profile_interval',
DEFAULT_COLLECTOR_PROFILE_INTERVAL)
self.check_frequency = int(self._agentConfig['check_freq'])
watchmonitor = self._get_watchmonitor(self.check_frequency)
self.restart_interval = int(self._agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
profiled = False
collector_profiled_runs = 0
while self.run_forever:
log.debug("Found {num_checks} checks".format(num_checks=len(self._checksd['initialized_checks'])))
if 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))
self.collector.run(checksd=self._checksd,
start_event=self.start_event,
configs_reloaded=self.configs_reloaded)
if self.configs_reloaded:
self.configs_reloaded = 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))
if self.autorestart and self._should_restart():
self._do_restart()
if self.run_forever:
if watchmonitor:
watchmonitor.reset()
if profiled:
collector_profiled_runs += 1
log.debug("Sleeping for {0} seconds".format(self.check_frequency))
time.sleep(self.check_frequency)
try:
CollectorStatus.remove_latest_status()
except Exception:
pass
log.info("Exiting. Bye bye.")
sys.exit(0)
def run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
agentConfig = self._set_agent_config_hostname(config)
hostname = get_hostname(agentConfig)
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
# Load the checks.d checks
checksd = load_check_directory(agentConfig, hostname)
self.collector = Collector(agentConfig, emitters, systemStats,
hostname)
# In developer mode, the number of runs to be included in a single collector profile
collector_profile_interval = agentConfig.get(
'collector_profile_interval', DEFAULT_COLLECTOR_PROFILE_INTERVAL)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Initialize the auto-restarter
self.restart_interval = int(
agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
profiled = False
collector_profiled_runs = 0
# Run the main loop.
while self.run_forever:
# Setup profiling if necessary
if 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))
# Do the work.
self.collector.run(checksd=checksd, start_event=self.start_event)
if profiled:
if collector_profiled_runs >= 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 the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
if profiled:
collector_profiled_runs += 1
time.sleep(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 run(self, config=None):
"""Main loop of the collector"""
# Gracefully exit on sigterm.
signal.signal(signal.SIGTERM, self._handle_sigterm)
# 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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
agentConfig = self._set_agent_config_hostname(config)
systemStats = get_system_stats()
emitters = self._get_emitters(agentConfig)
# Load the checks.d checks
checksd = load_check_directory(agentConfig)
self.collector = Collector(agentConfig, emitters, systemStats)
# Configure the watchdog.
check_frequency = int(agentConfig['check_freq'])
watchdog = self._get_watchdog(check_frequency, agentConfig)
# Initialize the auto-restarter
self.restart_interval = int(
agentConfig.get('restart_interval', RESTART_INTERVAL))
self.agent_start = time.time()
# Run the main loop.
while self.run_forever:
# Do the work.
self.collector.run(checksd=checksd, start_event=self.start_event)
# Check if we should restart.
if self.autorestart and self._should_restart():
self._do_restart()
# Only plan for the next loop if we will continue,
# otherwise just exit quickly.
if self.run_forever:
if watchdog:
watchdog.reset()
time.sleep(check_frequency)
# Now clean-up.
try:
CollectorStatus.remove_latest_status()
except:
pass
# Explicitly kill the process, because it might be running
# as a daemon.
log.info("Exiting. Bye bye.")
sys.exit(0)
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)
# Save the agent start-up stats.
CollectorStatus().persist()
# Intialize the collector.
if not config:
config = get_config(parse_args=True)
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 _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:
jmx_sd_configs = generate_jmx_configs(self._agentConfig, hostname)
if jmx_sd_configs:
self._submit_jmx_service_discovery(jmx_sd_configs)
# 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 = 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()
# 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 _populate_payload_metadata(self, payload, check_statuses, start_event=True):
now = time.time()
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
payload['events']['System'] = [{
'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
'timestamp': now,
'event_type': 'Agent Startup',
'msg_text': 'Version %s' % get_version()
}]
if self._should_send_additional_data('host_metadata'):
payload['gohai'] = get_gohai_data()
payload['systemStats'] = get_system_stats()
payload['meta'] = self._get_hostname_metadata()
log.info('GOHAI data: {0}'.format(payload['gohai']))
self.hostname_metadata_cache = payload['meta']
host_tags = []
if self.agentConfig['tags'] is not None:
host_tags.extend([tag.strip()
for tag in self.agentConfig['tags'].split(",")])
if self.agentConfig['collect_ec2_tags']:
host_tags.extend(EC2.get_tags(self.agentConfig))
if host_tags:
payload['host-tags']['system'] = host_tags
if self.agentConfig['create_dd_check_tags']:
app_tags_list = [DD_CHECK_TAG.format(c.name) for c in self.initialized_checks_d]
app_tags_list.extend([DD_CHECK_TAG.format(cname) for cname
in JMXFiles.get_jmx_appnames()])
if 'system' not in payload['host-tags']:
payload['host-tags']['system'] = []
payload['host-tags']['system'].extend(app_tags_list)
GCE_tags = GCE.get_tags(self.agentConfig)
if GCE_tags is not None:
payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
if self._is_first_run():
log.info("Hostnames: %s, tags: %s" %
(repr(self.hostname_metadata_cache), payload['host-tags']))
external_host_tags = []
if self._should_send_additional_data('external_host_tags'):
for check in self.initialized_checks_d:
try:
getter = getattr(check, 'get_external_host_tags')
check_tags = getter()
external_host_tags.extend(check_tags)
except AttributeError:
pass
if external_host_tags:
payload['external_host_tags'] = external_host_tags
if self._should_send_additional_data('agent_checks'):
agent_checks = []
for check in check_statuses:
if check.instance_statuses is not None:
for i, instance_status in enumerate(check.instance_statuses):
agent_checks.append(
(
check.name, check.source_type_name,
instance_status.instance_id,
instance_status.status,
instance_status.error or instance_status.warnings or "",
check.service_metadata[i]
)
)
else:
agent_checks.append(
(
check.name, check.source_type_name,
"initialization",
check.status, repr(check.init_failed_error)
)
)
payload['agent_checks'] = agent_checks
payload['meta'] = self.hostname_metadata_cache