def testSpeed(self):
# Pretend to be gmetad and serve a large piece of content
original_file = Fixtures.file('ganglia.txt')
server = subprocess.Popen("nc -l 8651 < %s" % original_file,
shell=True)
# Wait for 1 second
time.sleep(1)
pfile = tempfile.NamedTemporaryFile()
g = Ganglia(logging.getLogger(__file__))
# Running the profiler
# profile.runctx("g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651})", {}, {"g": g}, pfile.name)
# p = pstats.Stats(pfile.name)
# p.sort_stats('time').print_stats()
parsed = StringIO(
g.check({
'ganglia_host': 'localhost',
'ganglia_port': 8651
}))
original = Fixtures.file('ganglia.txt')
x1 = tree.parse(parsed)
x2 = tree.parse(original)
# Cursory test
self.assertEquals([c.tag for c in x1.getroot()],
[c.tag for c in x2.getroot()])
def testSpeed(self):
# Pretend to be gmetad and serve a large piece of content
server = subprocess.Popen("nc -l 8651 < %s" % TEST_FN, shell=True)
# Wait for 1 second
time.sleep(1)
pfile = tempfile.NamedTemporaryFile()
g = Ganglia(logging.getLogger(__file__))
# Running the profiler
# profile.runctx("g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651})", {}, {"g": g}, pfile.name)
# p = pstats.Stats(pfile.name)
# p.sort_stats('time').print_stats()
self.assertEquals(md5(g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651})).hexdigest(), md5(open(TEST_FN).read()).hexdigest())
def testSpeed(self):
# Pretend to be gmetad and serve a large piece of content
original_file = Fixtures.file('ganglia.txt')
subprocess.Popen("nc -l 8651 < %s" % original_file, shell=True)
# Wait for 1 second
time.sleep(1)
g = Ganglia(logging.getLogger(__file__))
parsed = StringIO(g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651}))
original = Fixtures.file('ganglia.txt')
x1 = tree.parse(parsed)
x2 = tree.parse(original)
# Cursory test
self.assertEquals([c.tag for c in x1.getroot()], [c.tag for c in x2.getroot()])
def testSpeed(self):
# Pretend to be gmetad and serve a large piece of content
server = subprocess.Popen("nc -l 8651 < %s" % TEST_FN, shell=True)
# Wait for 1 second
time.sleep(1)
pfile = tempfile.NamedTemporaryFile()
g = Ganglia(logging.getLogger(__file__))
# Running the profiler
# profile.runctx("g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651})", {}, {"g": g}, pfile.name)
# p = pstats.Stats(pfile.name)
# p.sort_stats('time').print_stats()
parsed = StringIO(g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651}))
original = open(TEST_FN)
x1 = tree.parse(parsed)
x2 = tree.parse(original)
# Cursory test
self.assertEquals([c.tag for c in x1.getroot()], [c.tag for c in x2.getroot()])
def testSpeed(self):
# Pretend to be gmetad and serve a large piece of content
server = subprocess.Popen("nc -l 8651 < %s" % TEST_FN, shell=True)
# Wait for 1 second
time.sleep(1)
pfile = tempfile.NamedTemporaryFile()
g = Ganglia(logging.getLogger(__file__))
# Running the profiler
# profile.runctx("g.check({'ganglia_host': 'localhost', 'ganglia_port': 8651})", {}, {"g": g}, pfile.name)
# p = pstats.Stats(pfile.name)
# p.sort_stats('time').print_stats()
self.assertEquals(
md5(g.check({
'ganglia_host': 'localhost',
'ganglia_port': 8651
})).hexdigest(),
md5(open(TEST_FN).read()).hexdigest())
def __init__(self, agentConfig, emitters, systemStats):
self.emit_duration = None
self.agentConfig = agentConfig
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.metadata_interval = int(agentConfig.get('metadata_interval', 10 * 60))
self.metadata_start = time.time()
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = []
# Unix System Checks
self._unix_system_checks = {
'disk': u.Disk(log),
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log)
}
# Win32 System `Checks
self._win32_system_checks = {
'disk': w32.Disk(log),
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log)
}
# Old-style metric checks
self._ganglia = Ganglia(log)
self._dogstream = Dogstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
# Agent Metrics
self._agent_metrics = CollectorMetrics(log)
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0: continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning("Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception, e:
log.exception('Unable to load custom check module %s' % module_spec)
def testSpeed(self):
# Pretend to be gmetad and serve a large piece of content
original_file = Fixtures.file('ganglia.txt')
subprocess.Popen("nc -l 8651 < %s" % original_file, shell=True)
# Wait for 1 second
time.sleep(1)
g = Ganglia(logging.getLogger(__file__))
parsed = StringIO(
g.check({
'ganglia_host': 'localhost',
'ganglia_port': 8651
}))
original = Fixtures.file('ganglia.txt')
x1 = tree.parse(parsed)
x2 = tree.parse(original)
# Cursory test
self.assertEquals([c.tag for c in x1.getroot()],
[c.tag for c in x2.getroot()])
def __init__(self, agentConfig, emitters):
self.agentConfig = agentConfig
self.plugins = None
self.emitters = emitters
self.os = None
self.checksLogger = logging.getLogger('checks')
socket.setdefaulttimeout(15)
self._apache = Apache(self.checksLogger)
self._nginx = Nginx(self.checksLogger)
self._disk = Disk(self.checksLogger)
self._io = IO()
self._load = Load(self.checksLogger)
self._memory = Memory(self.checksLogger)
self._network = Network(self.checksLogger)
self._processes = Processes()
self._cpu = Cpu()
self._couchdb = CouchDb(self.checksLogger)
self._mongodb = MongoDb(self.checksLogger)
self._mysql = MySql(self.checksLogger)
self._pgsql = PostgreSql(self.checksLogger)
self._rabbitmq = RabbitMq()
self._ganglia = Ganglia(self.checksLogger)
self._cassandra = Cassandra()
self._redis = Redis(self.checksLogger)
self._jvm = Jvm(self.checksLogger)
self._tomcat = Tomcat(self.checksLogger)
self._activemq = ActiveMQ(self.checksLogger)
self._solr = Solr(self.checksLogger)
self._memcache = Memcache(self.checksLogger)
self._dogstream = Dogstreams.init(self.checksLogger, self.agentConfig)
self._ddforwarder = DdForwarder(self.checksLogger, self.agentConfig)
# All new checks should be metrics checks:
self._metrics_checks = [
Cacti(self.checksLogger),
Redis(self.checksLogger),
Varnish(self.checksLogger),
ElasticSearch(self.checksLogger),
]
self._event_checks = [Hudson(), Nagios(socket.gethostname())]
self._resources_checks = [ResProcesses(self.checksLogger,self.agentConfig)]
self._ec2 = EC2(self.checksLogger)
class Collector(object):
"""
The collector is responsible for collecting data from each check and
passing it along to the emitters, who send it to their final destination.
"""
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval': int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60, # Wait for the checks to init
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval': int(agentConfig.get('agent_checks_interval', 10 * 60))
},
'processes': {
'start': time.time(),
'interval': int(agentConfig.get('processes_interval', 60))
}
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
# Unix System Checks
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
# Win32 System `Checks
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
# Old-style metric checks
self._ganglia = Ganglia(log)
self._dogstream = Dogstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning("Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception:
log.exception('Unable to load custom check module %s' % module_spec)
def stop(self):
"""
Tell the collector to stop at the next logical point.
"""
# This is called when the process is being killed, so
# try to stop the collector as soon as possible.
# Most importantly, don't try to submit to the emitters
# because the forwarder is quite possibly already killed
# in which case we'll get a misleading error in the logs.
# Best to not even try.
self.continue_running = False
for check in self.initialized_checks_d:
check.stop()
@staticmethod
def _stats_for_display(raw_stats):
return pprint.pformat(raw_stats, indent=4)
@log_exceptions(log)
def run(self, checksd=None, start_event=True, configs_reloaded=False):
"""
Collect data from each check and submit their data.
"""
log.debug("Found {num_checks} checks".format(num_checks=len(checksd['initialized_checks'])))
timer = Timer()
if not Platform.is_windows():
cpu_clock = time.clock()
self.run_count += 1
log.debug("Starting collection run #%s" % self.run_count)
if checksd:
self.initialized_checks_d = checksd['initialized_checks'] # is a list of AgentCheck instances
self.init_failed_checks_d = checksd['init_failed_checks'] # is of type {check_name: {error, traceback}}
payload = AgentPayload()
# Find the AgentMetrics check and pop it out
# This check must run at the end of the loop to collect info on agent performance
if not self._agent_metrics or configs_reloaded:
for check in self.initialized_checks_d:
if check.name == AGENT_METRICS_CHECK_NAME:
self._agent_metrics = check
self.initialized_checks_d.remove(check)
break
# Initialize payload
self._build_payload(payload)
metrics = payload['metrics']
events = payload['events']
service_checks = payload['service_checks']
# Run the system checks. Checks will depend on the OS
if Platform.is_windows():
# Win32 system checks
try:
metrics.extend(self._win32_system_checks['memory'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['cpu'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['network'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['io'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['proc'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['system'].check(self.agentConfig))
except Exception:
log.exception('Unable to fetch Windows system metrics.')
else:
# Unix system checks
sys_checks = self._unix_system_checks
load = sys_checks['load'].check(self.agentConfig)
payload.update(load)
system = sys_checks['system'].check(self.agentConfig)
payload.update(system)
memory = sys_checks['memory'].check(self.agentConfig)
if memory:
memstats = {
'memPhysUsed': memory.get('physUsed'),
'memPhysPctUsable': memory.get('physPctUsable'),
'memPhysFree': memory.get('physFree'),
'memPhysTotal': memory.get('physTotal'),
'memPhysUsable': memory.get('physUsable'),
'memSwapUsed': memory.get('swapUsed'),
'memSwapFree': memory.get('swapFree'),
'memSwapPctFree': memory.get('swapPctFree'),
'memSwapTotal': memory.get('swapTotal'),
'memCached': memory.get('physCached'),
'memBuffers': memory.get('physBuffers'),
'memShared': memory.get('physShared'),
'memSlab': memory.get('physSlab'),
'memPageTables': memory.get('physPageTables'),
'memSwapCached': memory.get('swapCached')
}
payload.update(memstats)
ioStats = sys_checks['io'].check(self.agentConfig)
if ioStats:
payload['ioStats'] = ioStats
processes = sys_checks['processes'].check(self.agentConfig)
payload.update({'processes': processes})
cpuStats = sys_checks['cpu'].check(self.agentConfig)
if cpuStats:
payload.update(cpuStats)
# Run old-style checks
gangliaData = self._ganglia.check(self.agentConfig)
dogstreamData = self._dogstream.check(self.agentConfig)
ddforwarderData = self._ddforwarder.check(self.agentConfig)
if gangliaData is not False and gangliaData is not None:
payload['ganglia'] = gangliaData
# dogstream
if dogstreamData:
dogstreamEvents = dogstreamData.get('dogstreamEvents', None)
if dogstreamEvents:
if 'dogstream' in payload['events']:
events['dogstream'].extend(dogstreamEvents)
else:
events['dogstream'] = dogstreamEvents
del dogstreamData['dogstreamEvents']
payload.update(dogstreamData)
# metrics about the forwarder
if ddforwarderData:
payload['datadog'] = ddforwarderData
# process collector of gohai (compliant with payload of legacy "resources checks")
if not Platform.is_windows() and self._should_send_additional_data('processes'):
gohai_processes = self._run_gohai_processes()
if gohai_processes:
try:
gohai_processes_json = json.loads(gohai_processes)
processes_payload = {
'snaps': [gohai_processes_json.get('processes')],
'format_version': 1
}
if self._is_first_run():
processes_payload['format_description'] = PROCESSES_FORMAT_DESCRIPTION
payload['resources'] = {
'processes': processes_payload,
'meta': {
'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
}
}
except Exception:
log.exception("Error running gohai processes collection")
# newer-style checks (not checks.d style)
for metrics_check in self._metrics_checks:
res = metrics_check.check(self.agentConfig)
if res:
metrics.extend(res)
# checks.d checks
check_statuses = []
for check in self.initialized_checks_d:
if not self.continue_running:
return
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_start_time = time.time()
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
check_stats = check._get_internal_profiling_stats()
# Collect metadata
current_check_metadata = check.get_service_metadata()
# Save metrics & events for the payload.
metrics.extend(current_check_metrics)
if current_check_events:
if check.name not in events:
events[check.name] = current_check_events
else:
events[check.name] += current_check_events
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name, instance_statuses, metric_count,
event_count, service_check_count, service_metadata=current_check_metadata,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats
)
# Service check for Agent checks failures
service_check_tags = ["check:%s" % check.name]
if check_status.status == STATUS_OK:
status = AgentCheck.OK
elif check_status.status == STATUS_ERROR:
status = AgentCheck.CRITICAL
check.service_check('datadog.agent.check_status', status, tags=service_check_tags)
# Collect the service checks and save them in the payload
current_check_service_checks = check.get_service_checks()
if current_check_service_checks:
service_checks.extend(current_check_service_checks)
service_check_count = len(current_check_service_checks)
# Update the check status with the correct service_check_count
check_status.service_check_count = service_check_count
check_statuses.append(check_status)
check_run_time = time.time() - check_start_time
log.debug("Check %s ran in %.2f s" % (check.name, check_run_time))
# Intrument check run timings if enabled.
if self.check_timings:
metric = 'datadog.agent.check_run_time'
meta = {'tags': ["check:%s" % check.name]}
metrics.append((metric, time.time(), check_run_time, meta))
for check_name, info in self.init_failed_checks_d.iteritems():
if not self.continue_running:
return
check_status = CheckStatus(check_name, None, None, None, None,
init_failed_error=info['error'],
init_failed_traceback=info['traceback'])
check_statuses.append(check_status)
# Add a service check for the agent
service_checks.append(create_service_check('datadog.agent.up', AgentCheck.OK,
hostname=self.hostname))
# Store the metrics and events in the payload.
payload['metrics'] = metrics
payload['events'] = events
payload['service_checks'] = service_checks
# Populate metadata
self._populate_payload_metadata(payload, check_statuses, start_event)
collect_duration = timer.step()
if self._agent_metrics:
metric_context = {
'collection_time': collect_duration,
'emit_time': self.emit_duration,
}
if not Platform.is_windows():
metric_context['cpu_time'] = time.clock() - cpu_clock
self._agent_metrics.set_metric_context(payload, metric_context)
self._agent_metrics.run()
agent_stats = self._agent_metrics.get_metrics()
payload['metrics'].extend(agent_stats)
if self.agentConfig.get('developer_mode'):
log.debug("\n Agent developer mode stats: \n {0}".format(
Collector._stats_for_display(agent_stats))
)
# Flush metadata for the Agent Metrics check. Otherwise they'll just accumulate and leak.
self._agent_metrics.get_service_metadata()
# Let's send our payload
emitter_statuses = payload.emit(log, self.agentConfig, self.emitters,
self.continue_running)
self.emit_duration = timer.step()
# Persist the status of the collection run.
try:
CollectorStatus(check_statuses, emitter_statuses,
self.hostname_metadata_cache).persist()
except Exception:
log.exception("Error persisting collector status")
if self.run_count <= FLUSH_LOGGING_INITIAL or self.run_count % FLUSH_LOGGING_PERIOD == 0:
log.info("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration, 2), round(self.emit_duration, 2)))
if self.run_count == FLUSH_LOGGING_INITIAL:
log.info("First flushes done, next flushes will be logged every %s flushes." %
FLUSH_LOGGING_PERIOD)
else:
log.debug("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration, 2), round(self.emit_duration, 2)))
return payload
@staticmethod
def run_single_check(check, verbose=True):
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
current_service_checks = check.get_service_checks()
current_service_metadata = check.get_service_metadata()
check_stats = check._get_internal_profiling_stats()
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
service_check_count = len(current_service_checks)
print "Metrics: \n{0}".format(pprint.pformat(current_check_metrics))
print "Events: \n{0}".format(pprint.pformat(current_check_events))
print "Service Checks: \n{0}".format(pprint.pformat(current_service_checks))
print "Service Metadata: \n{0}".format(pprint.pformat(current_service_metadata))
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name, instance_statuses, metric_count,
event_count, service_check_count,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats
)
return check_status
def _emit(self, payload):
""" Send the payload via the emitters. """
statuses = []
for emitter in self.emitters:
# Don't try to send to an emitter if we're stopping/
if not self.continue_running:
return statuses
name = emitter.__name__
emitter_status = EmitterStatus(name)
try:
emitter(payload, log, self.agentConfig)
except Exception, e:
log.exception("Error running emitter: %s" % emitter.__name__)
emitter_status = EmitterStatus(name, e)
statuses.append(emitter_status)
return statuses
class Collector(object):
"""
The collector is responsible for collecting data from each check and
passing it along to the emitters, who send it to their final destination.
"""
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval':
int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60, # Wait for the checks to init
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval':
int(agentConfig.get('agent_checks_interval', 10 * 60))
},
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
# Unix System Checks
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
# Win32 System `Checks
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
# Old-style metric checks
self._ganglia = Ganglia(log)
self._dogstream = Dogstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [
s.strip()
for s in self.agentConfig.get('custom_checks', '').split(',')
]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(
modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning(
"Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version"
)
except Exception:
log.exception('Unable to load custom check module %s' %
module_spec)
# Resource Checks
self._resources_checks = [ResProcesses(log, self.agentConfig)]
def stop(self):
"""
Tell the collector to stop at the next logical point.
"""
# This is called when the process is being killed, so
# try to stop the collector as soon as possible.
# Most importantly, don't try to submit to the emitters
# because the forwarder is quite possibly already killed
# in which case we'll get a misleading error in the logs.
# Best to not even try.
self.continue_running = False
for check in self.initialized_checks_d:
check.stop()
@staticmethod
def _stats_for_display(raw_stats):
return pprint.pformat(raw_stats, indent=4)
@log_exceptions(log)
def run(self, checksd=None, start_event=True, configs_reloaded=False):
"""
Collect data from each check and submit their data.
"""
log.debug("Found {num_checks} checks".format(
num_checks=len(checksd['initialized_checks'])))
timer = Timer()
if not Platform.is_windows():
cpu_clock = time.clock()
self.run_count += 1
log.debug("Starting collection run #%s" % self.run_count)
if checksd:
self.initialized_checks_d = checksd[
'initialized_checks'] # is a list of AgentCheck instances
self.init_failed_checks_d = checksd[
'init_failed_checks'] # is of type {check_name: {error, traceback}}
payload = AgentPayload()
# Find the AgentMetrics check and pop it out
# This check must run at the end of the loop to collect info on agent performance
if not self._agent_metrics or configs_reloaded:
for check in self.initialized_checks_d:
if check.name == AGENT_METRICS_CHECK_NAME:
self._agent_metrics = check
self.initialized_checks_d.remove(check)
break
# Initialize payload
self._build_payload(payload)
metrics = payload['metrics']
events = payload['events']
service_checks = payload['service_checks']
# Run the system checks. Checks will depend on the OS
if Platform.is_windows():
# Win32 system checks
try:
metrics.extend(self._win32_system_checks['memory'].check(
self.agentConfig))
metrics.extend(self._win32_system_checks['cpu'].check(
self.agentConfig))
metrics.extend(self._win32_system_checks['network'].check(
self.agentConfig))
metrics.extend(self._win32_system_checks['io'].check(
self.agentConfig))
metrics.extend(self._win32_system_checks['proc'].check(
self.agentConfig))
metrics.extend(self._win32_system_checks['system'].check(
self.agentConfig))
except Exception:
log.exception('Unable to fetch Windows system metrics.')
else:
# Unix system checks
sys_checks = self._unix_system_checks
load = sys_checks['load'].check(self.agentConfig)
payload.update(load)
system = sys_checks['system'].check(self.agentConfig)
payload.update(system)
memory = sys_checks['memory'].check(self.agentConfig)
if memory:
memstats = {
'memPhysUsed': memory.get('physUsed'),
'memPhysPctUsable': memory.get('physPctUsable'),
'memPhysFree': memory.get('physFree'),
'memPhysTotal': memory.get('physTotal'),
'memPhysUsable': memory.get('physUsable'),
'memSwapUsed': memory.get('swapUsed'),
'memSwapFree': memory.get('swapFree'),
'memSwapPctFree': memory.get('swapPctFree'),
'memSwapTotal': memory.get('swapTotal'),
'memCached': memory.get('physCached'),
'memBuffers': memory.get('physBuffers'),
'memShared': memory.get('physShared'),
'memSlab': memory.get('physSlab'),
'memPageTables': memory.get('physPageTables'),
'memSwapCached': memory.get('swapCached')
}
payload.update(memstats)
ioStats = sys_checks['io'].check(self.agentConfig)
if ioStats:
payload['ioStats'] = ioStats
processes = sys_checks['processes'].check(self.agentConfig)
payload.update({'processes': processes})
cpuStats = sys_checks['cpu'].check(self.agentConfig)
if cpuStats:
payload.update(cpuStats)
# Run old-style checks
gangliaData = self._ganglia.check(self.agentConfig)
dogstreamData = self._dogstream.check(self.agentConfig)
ddforwarderData = self._ddforwarder.check(self.agentConfig)
if gangliaData is not False and gangliaData is not None:
payload['ganglia'] = gangliaData
# dogstream
if dogstreamData:
dogstreamEvents = dogstreamData.get('dogstreamEvents', None)
if dogstreamEvents:
if 'dogstream' in payload['events']:
events['dogstream'].extend(dogstreamEvents)
else:
events['dogstream'] = dogstreamEvents
del dogstreamData['dogstreamEvents']
payload.update(dogstreamData)
# metrics about the forwarder
if ddforwarderData:
payload['datadog'] = ddforwarderData
# Resources checks
if not Platform.is_windows():
has_resource = False
for resources_check in self._resources_checks:
try:
resources_check.check()
snaps = resources_check.pop_snapshots()
if snaps:
has_resource = True
res_value = {
'snaps': snaps,
'format_version':
resources_check.get_format_version()
}
res_format = resources_check.describe_format_if_needed(
)
if res_format is not None:
res_value['format_description'] = res_format
payload['resources'][
resources_check.RESOURCE_KEY] = res_value
except Exception:
log.exception("Error running resource check %s" %
resources_check.RESOURCE_KEY)
if has_resource:
payload['resources']['meta'] = {
'api_key': self.agentConfig['api_key'],
'host': payload['internalHostname'],
}
# newer-style checks (not checks.d style)
for metrics_check in self._metrics_checks:
res = metrics_check.check(self.agentConfig)
if res:
metrics.extend(res)
# checks.d checks
check_statuses = []
for check in self.initialized_checks_d:
if not self.continue_running:
return
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_start_time = time.time()
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
check_stats = check._get_internal_profiling_stats()
# Collect metadata
current_check_metadata = check.get_service_metadata()
# Save metrics & events for the payload.
metrics.extend(current_check_metrics)
if current_check_events:
if check.name not in events:
events[check.name] = current_check_events
else:
events[check.name] += current_check_events
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name,
instance_statuses,
metric_count,
event_count,
service_check_count,
service_metadata=current_check_metadata,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats)
# Service check for Agent checks failures
service_check_tags = ["check:%s" % check.name]
if check_status.status == STATUS_OK:
status = AgentCheck.OK
elif check_status.status == STATUS_ERROR:
status = AgentCheck.CRITICAL
check.service_check('datadog.agent.check_status',
status,
tags=service_check_tags)
# Collect the service checks and save them in the payload
current_check_service_checks = check.get_service_checks()
if current_check_service_checks:
service_checks.extend(current_check_service_checks)
service_check_count = len(current_check_service_checks)
# Update the check status with the correct service_check_count
check_status.service_check_count = service_check_count
check_statuses.append(check_status)
check_run_time = time.time() - check_start_time
log.debug("Check %s ran in %.2f s" % (check.name, check_run_time))
# Intrument check run timings if enabled.
if self.check_timings:
metric = 'datadog.agent.check_run_time'
meta = {'tags': ["check:%s" % check.name]}
metrics.append((metric, time.time(), check_run_time, meta))
for check_name, info in self.init_failed_checks_d.iteritems():
if not self.continue_running:
return
check_status = CheckStatus(check_name,
None,
None,
None,
None,
init_failed_error=info['error'],
init_failed_traceback=info['traceback'])
check_statuses.append(check_status)
# Add a service check for the agent
service_checks.append(
create_service_check('datadog.agent.up',
AgentCheck.OK,
hostname=self.hostname))
# Store the metrics and events in the payload.
payload['metrics'] = metrics
payload['events'] = events
payload['service_checks'] = service_checks
# Populate metadata
self._populate_payload_metadata(payload, check_statuses, start_event)
collect_duration = timer.step()
if self._agent_metrics:
metric_context = {
'collection_time': collect_duration,
'emit_time': self.emit_duration,
}
if not Platform.is_windows():
metric_context['cpu_time'] = time.clock() - cpu_clock
self._agent_metrics.set_metric_context(payload, metric_context)
self._agent_metrics.run()
agent_stats = self._agent_metrics.get_metrics()
payload['metrics'].extend(agent_stats)
if self.agentConfig.get('developer_mode'):
log.debug("\n Agent developer mode stats: \n {0}".format(
Collector._stats_for_display(agent_stats)))
# Let's send our payload
emitter_statuses = payload.emit(log, self.agentConfig, self.emitters,
self.continue_running)
self.emit_duration = timer.step()
# Persist the status of the collection run.
try:
CollectorStatus(check_statuses, emitter_statuses,
self.hostname_metadata_cache).persist()
except Exception:
log.exception("Error persisting collector status")
if self.run_count <= FLUSH_LOGGING_INITIAL or self.run_count % FLUSH_LOGGING_PERIOD == 0:
log.info("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration,
2), round(self.emit_duration, 2)))
if self.run_count == FLUSH_LOGGING_INITIAL:
log.info(
"First flushes done, next flushes will be logged every %s flushes."
% FLUSH_LOGGING_PERIOD)
else:
log.debug(
"Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration,
2), round(self.emit_duration, 2)))
return payload
@staticmethod
def run_single_check(check, verbose=True):
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
current_service_checks = check.get_service_checks()
current_service_metadata = check.get_service_metadata()
check_stats = check._get_internal_profiling_stats()
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
service_check_count = len(current_service_checks)
print "Metrics: \n{0}".format(
pprint.pformat(current_check_metrics))
print "Events: \n{0}".format(pprint.pformat(current_check_events))
print "Service Checks: \n{0}".format(
pprint.pformat(current_service_checks))
print "Service Metadata: \n{0}".format(
pprint.pformat(current_service_metadata))
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(check.name,
instance_statuses,
metric_count,
event_count,
service_check_count,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME
or check.name,
check_stats=check_stats)
return check_status
def _emit(self, payload):
""" Send the payload via the emitters. """
statuses = []
for emitter in self.emitters:
# Don't try to send to an emitter if we're stopping/
if not self.continue_running:
return statuses
name = emitter.__name__
emitter_status = EmitterStatus(name)
try:
emitter(payload, log, self.agentConfig)
except Exception, e:
log.exception("Error running emitter: %s" % emitter.__name__)
emitter_status = EmitterStatus(name, e)
statuses.append(emitter_status)
return statuses
class Collector(object):
"""
The collector is responsible for collecting data from each check and
passing it along to the emitters, who send it to their final destination.
"""
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval': int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60, # Wait for the checks to init
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval': int(agentConfig.get('agent_checks_interval', 10 * 60))
},
'processes': {
'start': time.time(),
'interval': int(agentConfig.get('processes_interval', 60))
}
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
# Unix System Checks
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
# Win32 System `Checks
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
# Old-style metric checks
self._ganglia = Ganglia(log)
self._dogstream = Dogstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning("Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception:
log.exception('Unable to load custom check module %s' % module_spec)
def stop(self):
"""
Tell the collector to stop at the next logical point.
"""
# This is called when the process is being killed, so
# try to stop the collector as soon as possible.
# Most importantly, don't try to submit to the emitters
# because the forwarder is quite possibly already killed
# in which case we'll get a misleading error in the logs.
# Best to not even try.
self.continue_running = False
for check in self.initialized_checks_d:
check.stop()
@staticmethod
def _stats_for_display(raw_stats):
return pprint.pformat(raw_stats, indent=4)
@log_exceptions(log)
def run(self, checksd=None, start_event=True, configs_reloaded=False):
"""
Collect data from each check and submit their data.
"""
log.debug("Found {num_checks} checks".format(num_checks=len(checksd['initialized_checks'])))
timer = Timer()
if not Platform.is_windows():
cpu_clock = time.clock()
self.run_count += 1
log.debug("Starting collection run #%s" % self.run_count)
if checksd:
self.initialized_checks_d = checksd['initialized_checks'] # is a list of AgentCheck instances
self.init_failed_checks_d = checksd['init_failed_checks'] # is of type {check_name: {error, traceback}}
payload = AgentPayload()
# Find the AgentMetrics check and pop it out
# This check must run at the end of the loop to collect info on agent performance
if not self._agent_metrics or configs_reloaded:
for check in self.initialized_checks_d:
if check.name == AGENT_METRICS_CHECK_NAME:
self._agent_metrics = check
self.initialized_checks_d.remove(check)
break
# Initialize payload
self._build_payload(payload)
metrics = payload['metrics']
events = payload['events']
service_checks = payload['service_checks']
# Run the system checks. Checks will depend on the OS
if Platform.is_windows():
# Win32 system checks
try:
metrics.extend(self._win32_system_checks['memory'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['cpu'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['network'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['io'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['proc'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['system'].check(self.agentConfig))
except Exception:
log.exception('Unable to fetch Windows system metrics.')
else:
# Unix system checks
sys_checks = self._unix_system_checks
load = sys_checks['load'].check(self.agentConfig)
payload.update(load)
system = sys_checks['system'].check(self.agentConfig)
payload.update(system)
memory = sys_checks['memory'].check(self.agentConfig)
if memory:
memstats = {
'memPhysUsed': memory.get('physUsed'),
'memPhysPctUsable': memory.get('physPctUsable'),
'memPhysFree': memory.get('physFree'),
'memPhysTotal': memory.get('physTotal'),
'memPhysUsable': memory.get('physUsable'),
'memSwapUsed': memory.get('swapUsed'),
'memSwapFree': memory.get('swapFree'),
'memSwapPctFree': memory.get('swapPctFree'),
'memSwapTotal': memory.get('swapTotal'),
'memCached': memory.get('physCached'),
'memBuffers': memory.get('physBuffers'),
'memShared': memory.get('physShared'),
'memSlab': memory.get('physSlab'),
'memPageTables': memory.get('physPageTables'),
'memSwapCached': memory.get('swapCached')
}
payload.update(memstats)
ioStats = sys_checks['io'].check(self.agentConfig)
if ioStats:
payload['ioStats'] = ioStats
processes = sys_checks['processes'].check(self.agentConfig)
payload.update({'processes': processes})
cpuStats = sys_checks['cpu'].check(self.agentConfig)
if cpuStats:
payload.update(cpuStats)
# Run old-style checks
gangliaData = self._ganglia.check(self.agentConfig)
dogstreamData = self._dogstream.check(self.agentConfig)
ddforwarderData = self._ddforwarder.check(self.agentConfig)
if gangliaData is not False and gangliaData is not None:
payload['ganglia'] = gangliaData
# dogstream
if dogstreamData:
dogstreamEvents = dogstreamData.get('dogstreamEvents', None)
if dogstreamEvents:
if 'dogstream' in payload['events']:
events['dogstream'].extend(dogstreamEvents)
else:
events['dogstream'] = dogstreamEvents
del dogstreamData['dogstreamEvents']
payload.update(dogstreamData)
# metrics about the forwarder
if ddforwarderData:
payload['datadog'] = ddforwarderData
# process collector of gohai (compliant with payload of legacy "resources checks")
if not Platform.is_windows() and self._should_send_additional_data('processes'):
gohai_processes = self._run_gohai_processes()
if gohai_processes:
try:
gohai_processes_json = json.loads(gohai_processes)
processes_payload = {
'snaps': [gohai_processes_json.get('processes')],
'format_version': 1
}
if self._is_first_run():
processes_payload['format_description'] = PROCESSES_FORMAT_DESCRIPTION
payload['resources'] = {
'processes': processes_payload,
'meta': {
'host': payload['internalHostname'],
}
}
except Exception:
log.exception("Error running gohai processes collection")
# newer-style checks (not checks.d style)
for metrics_check in self._metrics_checks:
res = metrics_check.check(self.agentConfig)
if res:
metrics.extend(res)
# checks.d checks
check_statuses = []
for check in self.initialized_checks_d:
if not self.continue_running:
return
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_start_time = time.time()
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
check_stats = check._get_internal_profiling_stats()
# Collect metadata
current_check_metadata = check.get_service_metadata()
# Save metrics & events for the payload.
metrics.extend(current_check_metrics)
if current_check_events:
if check.name not in events:
events[check.name] = current_check_events
else:
events[check.name] += current_check_events
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name, instance_statuses, metric_count,
event_count, service_check_count, service_metadata=current_check_metadata,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats
)
# Service check for Agent checks failures
service_check_tags = ["check:%s" % check.name]
if check_status.status == STATUS_OK:
status = AgentCheck.OK
elif check_status.status == STATUS_ERROR:
status = AgentCheck.CRITICAL
check.service_check('datadog.agent.check_status', status, tags=service_check_tags)
# Collect the service checks and save them in the payload
current_check_service_checks = check.get_service_checks()
if current_check_service_checks:
service_checks.extend(current_check_service_checks)
service_check_count = len(current_check_service_checks)
# Update the check status with the correct service_check_count
check_status.service_check_count = service_check_count
check_statuses.append(check_status)
check_run_time = time.time() - check_start_time
log.debug("Check %s ran in %.2f s" % (check.name, check_run_time))
# Intrument check run timings if enabled.
if self.check_timings:
metric = 'datadog.agent.check_run_time'
meta = {'tags': ["check:%s" % check.name]}
metrics.append((metric, time.time(), check_run_time, meta))
for check_name, info in self.init_failed_checks_d.iteritems():
if not self.continue_running:
return
check_status = CheckStatus(check_name, None, None, None, None,
init_failed_error=info['error'],
init_failed_traceback=info['traceback'])
check_statuses.append(check_status)
# Add a service check for the agent
service_checks.append(create_service_check('datadog.agent.up', AgentCheck.OK,
hostname=self.hostname))
# Store the metrics and events in the payload.
payload['metrics'] = metrics
payload['events'] = events
payload['service_checks'] = service_checks
# Populate metadata
self._populate_payload_metadata(payload, check_statuses, start_event)
collect_duration = timer.step()
if self._agent_metrics:
metric_context = {
'collection_time': collect_duration,
'emit_time': self.emit_duration,
}
if not Platform.is_windows():
metric_context['cpu_time'] = time.clock() - cpu_clock
self._agent_metrics.set_metric_context(payload, metric_context)
self._agent_metrics.run()
agent_stats = self._agent_metrics.get_metrics()
payload['metrics'].extend(agent_stats)
if self.agentConfig.get('developer_mode'):
log.debug("\n Agent developer mode stats: \n {0}".format(
Collector._stats_for_display(agent_stats))
)
# Flush metadata for the Agent Metrics check. Otherwise they'll just accumulate and leak.
self._agent_metrics.get_service_metadata()
# Let's send our payload
emitter_statuses = payload.emit(log, self.agentConfig, self.emitters,
self.continue_running)
self.emit_duration = timer.step()
# Persist the status of the collection run.
try:
CollectorStatus(check_statuses, emitter_statuses,
self.hostname_metadata_cache).persist()
except Exception:
log.exception("Error persisting collector status")
if self.run_count <= FLUSH_LOGGING_INITIAL or self.run_count % FLUSH_LOGGING_PERIOD == 0:
log.info("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration, 2), round(self.emit_duration, 2)))
if self.run_count == FLUSH_LOGGING_INITIAL:
log.info("First flushes done, next flushes will be logged every %s flushes." %
FLUSH_LOGGING_PERIOD)
else:
log.debug("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration, 2), round(self.emit_duration, 2)))
return payload
@staticmethod
def run_single_check(check, verbose=True):
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
current_service_checks = check.get_service_checks()
current_service_metadata = check.get_service_metadata()
check_stats = check._get_internal_profiling_stats()
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
service_check_count = len(current_service_checks)
print "Metrics: \n{0}".format(pprint.pformat(current_check_metrics))
print "Events: \n{0}".format(pprint.pformat(current_check_events))
print "Service Checks: \n{0}".format(pprint.pformat(current_service_checks))
print "Service Metadata: \n{0}".format(pprint.pformat(current_service_metadata))
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name, instance_statuses, metric_count,
event_count, service_check_count,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats
)
return check_status
def _emit(self, payload):
""" Send the payload via the emitters. """
statuses = []
for emitter in self.emitters:
# Don't try to send to an emitter if we're stopping/
if not self.continue_running:
return statuses
name = emitter.__name__
emitter_status = EmitterStatus(name)
try:
emitter(payload, log, self.agentConfig)
except Exception as e:
log.exception("Error running emitter: %s" % emitter.__name__)
emitter_status = EmitterStatus(name, e)
statuses.append(emitter_status)
return statuses
def _is_first_run(self):
return self.run_count <= 1
def _build_payload(self, payload):
"""
Build the payload skeleton, so it contains all of the generic payload data.
"""
now = time.time()
payload['collection_timestamp'] = now
payload['os'] = self.os
payload['python'] = sys.version
payload['agentVersion'] = self.agentConfig['version']
payload['apiKey'] = self.agentConfig['api_key']
payload['events'] = {}
payload['metrics'] = []
payload['service_checks'] = []
payload['resources'] = {}
payload['internalHostname'] = self.hostname
payload['uuid'] = get_uuid()
payload['host-tags'] = {}
payload['external_host_tags'] = {}
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 _get_hostname_metadata(self):
"""
Returns a dictionnary that contains hostname metadata.
"""
metadata = EC2.get_metadata(self.agentConfig)
if metadata.get('hostname'):
metadata['ec2-hostname'] = metadata.get('hostname')
del metadata['hostname']
if self.agentConfig.get('hostname'):
metadata['agent-hostname'] = self.agentConfig.get('hostname')
else:
try:
metadata["socket-hostname"] = socket.gethostname()
except Exception:
pass
try:
metadata["socket-fqdn"] = socket.getfqdn()
except Exception:
pass
metadata["hostname"] = self.hostname
metadata["timezones"] = sanitize_tzname(time.tzname)
# Add cloud provider aliases
host_aliases = GCE.get_host_aliases(self.agentConfig)
if host_aliases:
metadata['host_aliases'] = host_aliases
return metadata
def _should_send_additional_data(self, data_name):
if self._is_first_run():
return True
# If the interval has passed, send the metadata again
now = time.time()
if now - self.push_times[data_name]['start'] >= self.push_times[data_name]['interval']:
log.debug('%s interval has passed. Sending it.' % data_name)
self.push_times[data_name]['start'] = now
return True
return False
def _run_gohai_metadata(self):
return self._run_gohai(['--exclude', 'processes'])
def _run_gohai_processes(self):
return self._run_gohai(['--only', 'processes'])
def _run_gohai(self, options):
output = None
try:
if not Platform.is_windows():
command = "gohai"
else:
command = "gohai\gohai.exe"
output, err, _ = get_subprocess_output([command] + options, log)
if err:
log.warning("GOHAI LOG | {0}".format(err))
except OSError as e:
if e.errno == 2: # file not found, expected when install from source
log.info("gohai file not found")
else:
log.warning("Unexpected OSError when running gohai %s", e)
except Exception as e:
log.warning("gohai command failed with error %s", e)
return output
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig["system_stats"] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get("check_timings")
self.push_times = {
"host_metadata": {"start": time.time(), "interval": int(agentConfig.get("metadata_interval", 4 * 60 * 60))},
"external_host_tags": {
"start": time.time() - 3 * 60, # Wait for the checks to init
"interval": int(agentConfig.get("external_host_tags", 5 * 60)),
},
"agent_checks": {"start": time.time(), "interval": int(agentConfig.get("agent_checks_interval", 10 * 60))},
"processes": {"start": time.time(), "interval": int(agentConfig.get("processes_interval", 60))},
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
# Unix System Checks
self._unix_system_checks = {
"io": u.IO(log),
"load": u.Load(log),
"memory": u.Memory(log),
"processes": u.Processes(log),
"cpu": u.Cpu(log),
"system": u.System(log),
}
# Win32 System `Checks
self._win32_system_checks = {
"io": w32.IO(log),
"proc": w32.Processes(log),
"memory": w32.Memory(log),
"network": w32.Network(log),
"cpu": w32.Cpu(log),
"system": w32.System(log),
}
# Old-style metric checks
self._ganglia = Ganglia(log) if self.agentConfig.get("ganglia_host", "") != "" else None
self._dogstream = None if self.agentConfig.get("dogstreams") is None else Dogstreams.init(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get("custom_checks", "").split(",")]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, "Check")(log))
log.info("Registered custom check %s" % module_spec)
log.warning(
"Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version"
)
except Exception:
log.exception("Unable to load custom check module %s" % module_spec)
def __init__(self, agentConfig, emitters, systemStats):
self.agentConfig = agentConfig
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = getOS()
self.plugins = None
self.emitters = emitters
self.metadata_interval = int(
agentConfig.get('metadata_interval', 10 * 60))
self.metadata_start = time.time()
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
# Unix System Checks
self._unix_system_checks = {
'disk': u.Disk(checks_logger),
'io': u.IO(),
'load': u.Load(checks_logger),
'memory': u.Memory(checks_logger),
'network': u.Network(checks_logger),
'processes': u.Processes(),
'cpu': u.Cpu(checks_logger)
}
# Win32 System `Checks
self._win32_system_checks = {
'disk': w32.Disk(checks_logger),
'io': w32.IO(checks_logger),
'proc': w32.Processes(checks_logger),
'memory': w32.Memory(checks_logger),
'network': w32.Network(checks_logger),
'cpu': w32.Cpu(checks_logger)
}
# Old-style metric checks
self._couchdb = CouchDb(checks_logger)
self._mongodb = MongoDb(checks_logger)
self._mysql = MySql(checks_logger)
self._rabbitmq = RabbitMq()
self._ganglia = Ganglia(checks_logger)
self._cassandra = Cassandra()
self._dogstream = Dogstreams.init(checks_logger, self.agentConfig)
self._ddforwarder = DdForwarder(checks_logger, self.agentConfig)
self._ec2 = EC2(checks_logger)
# Metric Checks
self._metrics_checks = [
ElasticSearch(checks_logger),
Jvm(checks_logger),
Tomcat(checks_logger),
ActiveMQ(checks_logger),
Solr(checks_logger),
WMICheck(checks_logger),
Memcache(checks_logger),
]
# Custom metric checks
for module_spec in [
s.strip()
for s in self.agentConfig.get('custom_checks', '').split(',')
]:
if len(module_spec) == 0: continue
try:
self._metrics_checks.append(
modules.load(module_spec, 'Check')(checks_logger))
logger.info("Registered custom check %s" % module_spec)
except Exception, e:
logger.exception('Unable to load custom check module %s' %
module_spec)
class checks(object):
def __init__(self, agentConfig, emitters):
self.agentConfig = agentConfig
self.plugins = None
self.emitters = emitters
self.os = None
self.checksLogger = logging.getLogger('checks')
socket.setdefaulttimeout(15)
self._apache = Apache(self.checksLogger)
self._nginx = Nginx(self.checksLogger)
self._disk = Disk(self.checksLogger)
self._io = IO()
self._load = Load(self.checksLogger)
self._memory = Memory(self.checksLogger)
self._network = Network(self.checksLogger)
self._processes = Processes()
self._cpu = Cpu()
self._couchdb = CouchDb(self.checksLogger)
self._mongodb = MongoDb(self.checksLogger)
self._mysql = MySql(self.checksLogger)
self._pgsql = PostgreSql(self.checksLogger)
self._rabbitmq = RabbitMq()
self._ganglia = Ganglia(self.checksLogger)
self._cassandra = Cassandra()
self._redis = Redis(self.checksLogger)
self._jvm = Jvm(self.checksLogger)
self._tomcat = Tomcat(self.checksLogger)
self._activemq = ActiveMQ(self.checksLogger)
self._solr = Solr(self.checksLogger)
self._memcache = Memcache(self.checksLogger)
self._dogstream = Dogstreams.init(self.checksLogger, self.agentConfig)
self._ddforwarder = DdForwarder(self.checksLogger, self.agentConfig)
# All new checks should be metrics checks:
self._metrics_checks = [
Cacti(self.checksLogger),
Redis(self.checksLogger),
Varnish(self.checksLogger),
ElasticSearch(self.checksLogger),
]
self._event_checks = [Hudson(), Nagios(socket.gethostname())]
self._resources_checks = [ResProcesses(self.checksLogger,self.agentConfig)]
self._ec2 = EC2(self.checksLogger)
#
# Checks - FIXME migrating to the new Check interface is a WIP
#
@recordsize
def getApacheStatus(self):
return self._apache.check(self.agentConfig)
@recordsize
def getCouchDBStatus(self):
return self._couchdb.check(self.agentConfig)
@recordsize
def getDiskUsage(self):
return self._disk.check(self.agentConfig)
@recordsize
def getIOStats(self):
return self._io.check(self.checksLogger, self.agentConfig)
@recordsize
def getLoadAvrgs(self):
return self._load.check(self.agentConfig)
@recordsize
def getMemoryUsage(self):
return self._memory.check(self.agentConfig)
@recordsize
def getMongoDBStatus(self):
return self._mongodb.check(self.agentConfig)
@recordsize
def getMySQLStatus(self):
return self._mysql.check(self.agentConfig)
@recordsize
def getPgSQLStatus(self):
return self._pgsql.check(self.agentConfig)
@recordsize
def getNetworkTraffic(self):
return self._network.check(self.agentConfig)
@recordsize
def getNginxStatus(self):
return self._nginx.check(self.agentConfig)
@recordsize
def getProcesses(self):
return self._processes.check(self.checksLogger, self.agentConfig)
@recordsize
def getRabbitMQStatus(self):
return self._rabbitmq.check(self.checksLogger, self.agentConfig)
@recordsize
def getGangliaData(self):
return self._ganglia.check(self.agentConfig)
@recordsize
def getCassandraData(self):
return self._cassandra.check(self.checksLogger, self.agentConfig)
@recordsize
def getJvmData(self):
return self._jvm.check(self.agentConfig)
@recordsize
def getTomcatData(self):
return self._tomcat.check(self.agentConfig)
@recordsize
def getActiveMQData(self):
return self._activemq.check(self.agentConfig)
@recordsize
def getSolrData(self):
return self._solr.check(self.agentConfig)
@recordsize
def getMemcacheData(self):
return self._memcache.check(self.agentConfig)
@recordsize
def getDogstreamData(self):
return self._dogstream.check(self.agentConfig)
@recordsize
def getDdforwarderData(self):
return self._ddforwarder.check(self.agentConfig)
@recordsize
def getCPUStats(self):
return self._cpu.check(self.checksLogger, self.agentConfig)
@recordsize
def get_metadata(self):
metadata = self._ec2.get_metadata()
if metadata.get('hostname'):
metadata['ec2-hostname'] = metadata.get('hostname')
if self.agentConfig.get('hostname'):
metadata['agent-hostname'] = self.agentConfig.get('hostname')
try:
metadata["hostname"] = socket.gethostname()
except:
pass
try:
metadata["fqdn"] = socket.getfqdn()
except:
pass
return metadata
def doChecks(self, firstRun=False, systemStats=False):
"""Actual work
"""
self.checksLogger.info("Starting checks")
apacheStatus = self.getApacheStatus()
diskUsage = self.getDiskUsage()
loadAvrgs = self.getLoadAvrgs()
memory = self.getMemoryUsage()
mysqlStatus = self.getMySQLStatus()
pgsqlStatus = self.getPgSQLStatus()
networkTraffic = self.getNetworkTraffic()
nginxStatus = self.getNginxStatus()
processes = self.getProcesses()
rabbitmq = self.getRabbitMQStatus()
mongodb = self.getMongoDBStatus()
couchdb = self.getCouchDBStatus()
ioStats = self.getIOStats()
cpuStats = self.getCPUStats()
gangliaData = self.getGangliaData()
cassandraData = self.getCassandraData()
jvmData = self.getJvmData()
tomcatData = self.getTomcatData()
activeMQData = self.getActiveMQData()
solrData = self.getSolrData()
memcacheData = self.getMemcacheData()
dogstreamData = self.getDogstreamData()
ddforwarderData = self.getDdforwarderData()
checksData = {
'collection_timestamp': time.time(),
'os' : self.os,
'python': sys.version,
'agentVersion' : self.agentConfig['version'],
'loadAvrg1' : loadAvrgs['1'],
'loadAvrg5' : loadAvrgs['5'],
'loadAvrg15' : loadAvrgs['15'],
'memPhysUsed' : memory.get('physUsed'),
'memPhysFree' : memory.get('physFree'),
'memPhysTotal' : memory.get('physTotal'),
'memPhysUsable' : memory.get('physUsable'),
'memSwapUsed' : memory.get('swapUsed'),
'memSwapFree' : memory.get('swapFree'),
'memSwapTotal' : memory.get('swapTotal'),
'memCached' : memory.get('physCached'),
'memBuffers': memory.get('physBuffers'),
'memShared': memory.get('physShared'),
'networkTraffic' : networkTraffic,
'processes' : processes,
'apiKey': self.agentConfig['apiKey'],
'events': {},
'resources': {},
}
if diskUsage is not False and len(diskUsage) == 2:
checksData["diskUsage"] = diskUsage[0]
checksData["inodes"] = diskUsage[1]
if cpuStats is not False and cpuStats is not None:
checksData.update(cpuStats)
if gangliaData is not False and gangliaData is not None:
checksData['ganglia'] = gangliaData
if cassandraData is not False and cassandraData is not None:
checksData['cassandra'] = cassandraData
# Apache Status
if apacheStatus:
checksData.update(apacheStatus)
# MySQL Status
if mysqlStatus:
checksData.update(mysqlStatus)
# PostgreSQL status
if pgsqlStatus:
checksData['postgresql'] = pgsqlStatus
# Nginx Status
if nginxStatus:
checksData.update(nginxStatus)
# RabbitMQ
if rabbitmq:
checksData['rabbitMQ'] = rabbitmq
# MongoDB
if mongodb:
if mongodb.has_key('events'):
checksData['events']['Mongo'] = mongodb['events']['Mongo']
del mongodb['events']
checksData['mongoDB'] = mongodb
# CouchDB
if couchdb:
checksData['couchDB'] = couchdb
if ioStats:
checksData['ioStats'] = ioStats
if jvmData:
checksData['jvm'] = jvmData
if tomcatData:
checksData['tomcat'] = tomcatData
if activeMQData:
checksData['activemq'] = activeMQData
if solrData:
checksData['solr'] = solrData
if memcacheData:
checksData['memcache'] = memcacheData
if dogstreamData:
dogstreamEvents = dogstreamData.get('dogstreamEvents', None)
if dogstreamEvents:
if 'dogstream' in checksData['events']:
checksData['events']['dogstream'].extend(dogstreamEvents)
else:
checksData['events']['dogstream'] = dogstreamEvents
del dogstreamData['dogstreamEvents']
checksData.update(dogstreamData)
if ddforwarderData:
checksData['datadog'] = ddforwarderData
# Include server indentifiers
checksData['internalHostname'] = gethostname(self.agentConfig)
checksData['uuid'] = getUuid()
self.checksLogger.debug('doChecks: added uuid %s' % checksData['uuid'])
# Process the event checks.
for event_check in self._event_checks:
event_data = event_check.check(self.checksLogger, self.agentConfig)
if event_data:
checksData['events'][event_check.key] = event_data
# Include system stats on first postback
if firstRun:
checksData['systemStats'] = systemStats
# Add static tags from the configuration file
if self.agentConfig['tags'] is not None:
checksData['tags'] = self.agentConfig['tags']
# Also post an event in the newsfeed
checksData['events']['System'] = [{'api_key': self.agentConfig['apiKey'],
'host': checksData['internalHostname'],
'timestamp': int(time.mktime(datetime.datetime.now().timetuple())),
'event_type':'Agent Startup',
'msg_text': 'Version %s' % get_version()
}]
# Collect metadata
checksData['meta'] = self.get_metadata()
# Resources checks
has_resource = False
for resources_check in self._resources_checks:
resources_check.check()
snaps = resources_check.pop_snapshots()
if snaps:
has_resource = True
res_value = { 'snaps': snaps,
'format_version': resources_check.get_format_version() }
res_format = resources_check.describe_format_if_needed()
if res_format is not None:
res_value['format_description'] = res_format
checksData['resources'][resources_check.RESOURCE_KEY] = res_value
if has_resource:
checksData['resources']['meta'] = {
'api_key': self.agentConfig['apiKey'],
'host': checksData['internalHostname'],
}
metrics = []
for metrics_check in self._metrics_checks:
res = metrics_check.check(self.agentConfig)
if res:
metrics.extend(res)
checksData['metrics'] = metrics
# Send back data
self.checksLogger.debug("checksData: %s" % checksData)
for emitter in self.emitters:
emitter(checksData, self.checksLogger, self.agentConfig)
self.checksLogger.info("Checks done")
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval': int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60, # Wait for the checks to init
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval': int(agentConfig.get('agent_checks_interval', 10 * 60))
},
'processes': {
'start': time.time(),
'interval': int(agentConfig.get('processes_interval', 60))
}
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
if Platform.is_linux() and psutil is not None:
procfs_path = agentConfig.get('procfs_path', '/proc').rstrip('/')
psutil.PROCFS_PATH = procfs_path
# Unix System Checks
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
# Win32 System `Checks
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
# Old-style metric checks
self._ganglia = Ganglia(log) if self.agentConfig.get('ganglia_host', '') != '' else None
self._dogstream = None if self.agentConfig.get('dogstreams') is None else Dogstreams.init(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning("Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception:
log.exception('Unable to load custom check module %s' % module_spec)
class Collector(object):
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.ip = get_ip(agentConfig)
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
self.agentConfig['system_stats'] = systemStats
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval': int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60,
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval': int(agentConfig.get('agent_checks_interval', 10 * 60))
},
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
self._ganglia = Ganglia(log)
self._monitorstream = monitorstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
self._agent_metrics = None
self._metrics_checks = []
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning(
"Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception:
log.exception('Unable to load custom check module %s' % module_spec)
def stop(self):
self.continue_running = False
for check in self.initialized_checks_d:
check.stop()
@staticmethod
def _stats_for_display(raw_stats):
return pprint.pformat(raw_stats, indent=4)
@log_exceptions(log)
def run(self, checksd=None, start_event=True, configs_reloaded=False):
log.debug("Found {num_checks} checks".format(num_checks=len(checksd['initialized_checks'])))
timer = Timer()
if not Platform.is_windows():
cpu_clock = time.clock()
self.run_count += 1
log.debug("Starting collection run #%s" % self.run_count)
if checksd:
self.initialized_checks_d = checksd['initialized_checks']
self.init_failed_checks_d = checksd['init_failed_checks']
payload = AgentPayload()
if not self._agent_metrics or configs_reloaded:
for check in self.initialized_checks_d:
if check.name == AGENT_METRICS_CHECK_NAME:
self._agent_metrics = check
self.initialized_checks_d.remove(check)
break
self._build_payload(payload)
metrics = payload['metrics']
events = payload['events']
service_checks = payload['service_checks']
if Platform.is_windows():
try:
metrics.extend(self._win32_system_checks['memory'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['cpu'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['network'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['io'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['proc'].check(self.agentConfig))
metrics.extend(self._win32_system_checks['system'].check(self.agentConfig))
except Exception:
log.exception('Unable to fetch Windows system metrics.')
else:
sys_checks = self._unix_system_checks
load = sys_checks['load'].check(self.agentConfig)
payload.update(load)
system = sys_checks['system'].check(self.agentConfig)
payload.update(system)
memory = sys_checks['memory'].check(self.agentConfig)
if memory:
memstats = {
'memPhysUsed': memory.get('physUsed'),
'memPhysPctUsage': memory.get('physPctUsage'),
'memPhysFree': memory.get('physFree'),
'memPhysTotal': memory.get('physTotal'),
'memPhysUsable': memory.get('physUsable'),
'memSwapUsed': memory.get('swapUsed'),
'memSwapFree': memory.get('swapFree'),
'memSwapPctFree': memory.get('swapPctFree'),
'memSwapTotal': memory.get('swapTotal'),
'memCached': memory.get('physCached'),
'memBuffers': memory.get('physBuffers'),
'memShared': memory.get('physShared'),
'memSlab': memory.get('physSlab'),
'memPageTables': memory.get('physPageTables'),
'memSwapCached': memory.get('swapCached')
}
payload.update(memstats)
ioStats = sys_checks['io'].check(self.agentConfig)
if ioStats:
payload['ioStats'] = ioStats
processes = sys_checks['processes'].check(self.agentConfig)
payload.update({'processes': processes})
cpuStats = sys_checks['cpu'].check(self.agentConfig)
if cpuStats:
payload.update(cpuStats)
gangliaData = self._ganglia.check(self.agentConfig)
monitorstreamData = self._monitorstream.check(self.agentConfig)
ddforwarderData = self._ddforwarder.check(self.agentConfig)
if gangliaData is not False and gangliaData is not None:
payload['ganglia'] = gangliaData
if monitorstreamData:
monitorstreamEvents = monitorstreamData.get('monitorstreamEvents', None)
if monitorstreamEvents:
if 'monitorstream' in payload['events']:
events['monitorstream'].extend(monitorstreamEvents)
else:
events['monitorstream'] = monitorstreamEvents
del monitorstreamData['monitorstreamEvents']
payload.update(monitorstreamData)
if ddforwarderData:
payload['datamonitor'] = ddforwarderData
for metrics_check in self._metrics_checks:
res = metrics_check.check(self.agentConfig)
if res:
metrics.extend(res)
check_statuses = []
for check in self.initialized_checks_d:
if not self.continue_running:
return
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_start_time = time.time()
check_stats = None
try:
instance_statuses = check.run()
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
check_stats = check._get_internal_profiling_stats()
current_check_metadata = check.get_service_metadata()
metrics.extend(current_check_metrics)
if current_check_events:
if check.name not in events:
events[check.name] = current_check_events
else:
events[check.name] += current_check_events
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name, instance_statuses, metric_count,
event_count, service_check_count, service_metadata=current_check_metadata,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats
)
service_check_tags = ["check:%s" % check.name]
if check_status.status == STATUS_OK:
status = AgentCheck.OK
elif check_status.status == STATUS_ERROR:
status = AgentCheck.CRITICAL
check.service_check('datamonitor.agent.check_status', status, tags=service_check_tags)
current_check_service_checks = check.get_service_checks()
if current_check_service_checks:
service_checks.extend(current_check_service_checks)
service_check_count = len(current_check_service_checks)
check_status.service_check_count = service_check_count
check_statuses.append(check_status)
check_run_time = time.time() - check_start_time
log.debug("Check %s ran in %.2f s" % (check.name, check_run_time))
if self.check_timings:
metric = 'datamonitor.agent.check_run_time'
meta = {'tags': ["check:%s" % check.name]}
metrics.append((metric, time.time(), check_run_time, meta))
for check_name, info in self.init_failed_checks_d.iteritems():
if not self.continue_running:
return
check_status = CheckStatus(check_name, None, None, None, None,
init_failed_error=info['error'],
init_failed_traceback=info['error'])
check_statuses.append(check_status)
service_checks.append(create_service_check('datamonitor.agent.up', AgentCheck.OK,
hostname=self.hostname))
payload['metrics'] = metrics
payload['events'] = events
payload['service_checks'] = service_checks
self._populate_payload_metadata(payload, check_statuses, start_event)
collect_duration = timer.step()
if self._agent_metrics:
metric_context = {
'collection_time': collect_duration,
'emit_time': self.emit_duration,
}
if not Platform.is_windows():
metric_context['cpu_time'] = time.clock() - cpu_clock
self._agent_metrics.set_metric_context(payload, metric_context)
self._agent_metrics.run()
agent_stats = self._agent_metrics.get_metrics()
payload['metrics'].extend(agent_stats)
if self.agentConfig.get('developer_mode'):
log.debug("\n Agent developer mode stats: \n {0}".format(
Collector._stats_for_display(agent_stats))
)
self._agent_metrics.get_service_metadata()
emitter_statuses = payload.emit(log, self.agentConfig, self.emitters,
self.continue_running)
self.emit_duration = timer.step()
try:
CollectorStatus(check_statuses, emitter_statuses,
self.hostname_metadata_cache).persist()
except Exception:
log.exception("Error persisting collector status")
if self.run_count <= FLUSH_LOGGING_INITIAL or self.run_count % FLUSH_LOGGING_PERIOD == 0:
log.info("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration, 2), round(self.emit_duration, 2)))
if self.run_count == FLUSH_LOGGING_INITIAL:
log.info("First flushes done, next flushes will be logged every %s flushes." %
FLUSH_LOGGING_PERIOD)
else:
log.debug("Finished run #%s. Collection time: %ss. Emit time: %ss" %
(self.run_count, round(collect_duration, 2), round(self.emit_duration, 2)))
return payload
@staticmethod
def run_single_check(check, verbose=True):
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_stats = None
try:
instance_statuses = check.run()
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
current_service_checks = check.get_service_checks()
current_service_metadata = check.get_service_metadata()
check_stats = check._get_internal_profiling_stats()
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
service_check_count = len(current_service_checks)
print "Metrics: \n{0}".format(pprint.pformat(current_check_metrics))
print "Events: \n{0}".format(pprint.pformat(current_check_events))
print "Service Checks: \n{0}".format(pprint.pformat(current_service_checks))
print "Service Metadata: \n{0}".format(pprint.pformat(current_service_metadata))
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name, instance_statuses, metric_count,
event_count, service_check_count,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats
)
return check_status
def _emit(self, payload):
statuses = []
for emitter in self.emitters:
if not self.continue_running:
return statuses
name = emitter.__name__
emitter_status = EmitterStatus(name)
try:
emitter(payload, log, self.agentConfig)
except Exception, e:
log.exception("Error running emitter: %s" % emitter.__name__)
emitter_status = EmitterStatus(name, e)
statuses.append(emitter_status)
return statuses
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.ip = get_ip(agentConfig)
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
self.agentConfig['system_stats'] = systemStats
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval': int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60,
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval': int(agentConfig.get('agent_checks_interval', 10 * 60))
},
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
self._ganglia = Ganglia(log)
self._monitorstream = monitorstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
self._agent_metrics = None
self._metrics_checks = []
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning(
"Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception:
log.exception('Unable to load custom check module %s' % module_spec)
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig['system_stats'] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get('check_timings')
self.push_times = {
'host_metadata': {
'start': time.time(),
'interval': int(agentConfig.get('metadata_interval', 4 * 60 * 60))
},
'external_host_tags': {
'start': time.time() - 3 * 60, # Wait for the checks to init
'interval': int(agentConfig.get('external_host_tags', 5 * 60))
},
'agent_checks': {
'start': time.time(),
'interval': int(agentConfig.get('agent_checks_interval', 10 * 60))
},
'processes': {
'start': time.time(),
'interval': int(agentConfig.get('processes_interval', 60))
}
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
# Unix System Checks
self._unix_system_checks = {
'io': u.IO(log),
'load': u.Load(log),
'memory': u.Memory(log),
'processes': u.Processes(log),
'cpu': u.Cpu(log),
'system': u.System(log)
}
# Win32 System `Checks
self._win32_system_checks = {
'io': w32.IO(log),
'proc': w32.Processes(log),
'memory': w32.Memory(log),
'network': w32.Network(log),
'cpu': w32.Cpu(log),
'system': w32.System(log)
}
# Old-style metric checks
self._ganglia = Ganglia(log)
self._dogstream = Dogstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get('custom_checks', '').split(',')]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, 'Check')(log))
log.info("Registered custom check %s" % module_spec)
log.warning("Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version")
except Exception:
log.exception('Unable to load custom check module %s' % module_spec)
class Collector(object):
"""
The collector is responsible for collecting data from each check and
passing it along to the emitters, who send it to their final destination.
"""
def __init__(self, agentConfig, emitters, systemStats, hostname):
self.emit_duration = None
self.agentConfig = agentConfig
self.hostname = hostname
# system stats is generated by config.get_system_stats
self.agentConfig["system_stats"] = systemStats
# agent config is used during checks, system_stats can be accessed through the config
self.os = get_os()
self.plugins = None
self.emitters = emitters
self.check_timings = agentConfig.get("check_timings")
self.push_times = {
"host_metadata": {"start": time.time(), "interval": int(agentConfig.get("metadata_interval", 4 * 60 * 60))},
"external_host_tags": {
"start": time.time() - 3 * 60, # Wait for the checks to init
"interval": int(agentConfig.get("external_host_tags", 5 * 60)),
},
"agent_checks": {"start": time.time(), "interval": int(agentConfig.get("agent_checks_interval", 10 * 60))},
}
socket.setdefaulttimeout(15)
self.run_count = 0
self.continue_running = True
self.hostname_metadata_cache = None
self.initialized_checks_d = []
self.init_failed_checks_d = {}
# Unix System Checks
self._unix_system_checks = {
"io": u.IO(log),
"load": u.Load(log),
"memory": u.Memory(log),
"processes": u.Processes(log),
"cpu": u.Cpu(log),
"system": common.System(log),
}
# Win32 System `Checks
self._win32_system_checks = {
"io": w32.IO(log),
"proc": w32.Processes(log),
"memory": w32.Memory(log),
"network": w32.Network(log),
"cpu": w32.Cpu(log),
"system": common.System(log),
}
# Old-style metric checks
self._ganglia = Ganglia(log)
self._dogstream = Dogstreams.init(log, self.agentConfig)
self._ddforwarder = DdForwarder(log, self.agentConfig)
# Agent performance metrics check
self._agent_metrics = None
self._metrics_checks = []
# Custom metric checks
for module_spec in [s.strip() for s in self.agentConfig.get("custom_checks", "").split(",")]:
if len(module_spec) == 0:
continue
try:
self._metrics_checks.append(modules.load(module_spec, "Check")(log))
log.info("Registered custom check %s" % module_spec)
log.warning(
"Old format custom checks are deprecated. They should be moved to the checks.d interface as old custom checks will be removed in a next version"
)
except Exception:
log.exception("Unable to load custom check module %s" % module_spec)
# Resource Checks
self._resources_checks = [ResProcesses(log, self.agentConfig)]
def stop(self):
"""
Tell the collector to stop at the next logical point.
"""
# This is called when the process is being killed, so
# try to stop the collector as soon as possible.
# Most importantly, don't try to submit to the emitters
# because the forwarder is quite possibly already killed
# in which case we'll get a misleading error in the logs.
# Best to not even try.
self.continue_running = False
for check in self.initialized_checks_d:
check.stop()
@staticmethod
def _stats_for_display(raw_stats):
return pprint.pformat(raw_stats, indent=4)
@log_exceptions(log)
def run(self, checksd=None, start_event=True, configs_reloaded=False):
"""
Collect data from each check and submit their data.
"""
log.debug("Found {num_checks} checks".format(num_checks=len(checksd["initialized_checks"])))
timer = Timer()
if not Platform.is_windows():
cpu_clock = time.clock()
self.run_count += 1
log.debug("Starting collection run #%s" % self.run_count)
if checksd:
self.initialized_checks_d = checksd["initialized_checks"] # is a list of AgentCheck instances
self.init_failed_checks_d = checksd["init_failed_checks"] # is of type {check_name: {error, traceback}}
payload = AgentPayload()
# Find the AgentMetrics check and pop it out
# This check must run at the end of the loop to collect info on agent performance
if not self._agent_metrics or configs_reloaded:
for check in self.initialized_checks_d:
if check.name == AGENT_METRICS_CHECK_NAME:
self._agent_metrics = check
self.initialized_checks_d.remove(check)
break
# Initialize payload
self._build_payload(payload)
metrics = payload["metrics"]
events = payload["events"]
service_checks = payload["service_checks"]
# Run the system checks. Checks will depend on the OS
if Platform.is_windows():
# Win32 system checks
try:
metrics.extend(self._win32_system_checks["memory"].check(self.agentConfig))
metrics.extend(self._win32_system_checks["cpu"].check(self.agentConfig))
metrics.extend(self._win32_system_checks["network"].check(self.agentConfig))
metrics.extend(self._win32_system_checks["io"].check(self.agentConfig))
metrics.extend(self._win32_system_checks["proc"].check(self.agentConfig))
except Exception:
log.exception("Unable to fetch Windows system metrics.")
else:
# Unix system checks
sys_checks = self._unix_system_checks
load = sys_checks["load"].check(self.agentConfig)
payload.update(load)
system = sys_checks["system"].check(self.agentConfig)
payload.update(system)
memory = sys_checks["memory"].check(self.agentConfig)
if memory:
memstats = {
"memPhysUsed": memory.get("physUsed"),
"memPhysPctUsable": memory.get("physPctUsable"),
"memPhysFree": memory.get("physFree"),
"memPhysTotal": memory.get("physTotal"),
"memPhysUsable": memory.get("physUsable"),
"memSwapUsed": memory.get("swapUsed"),
"memSwapFree": memory.get("swapFree"),
"memSwapPctFree": memory.get("swapPctFree"),
"memSwapTotal": memory.get("swapTotal"),
"memCached": memory.get("physCached"),
"memBuffers": memory.get("physBuffers"),
"memShared": memory.get("physShared"),
"memSlab": memory.get("physSlab"),
"memPageTables": memory.get("physPageTables"),
"memSwapCached": memory.get("swapCached"),
}
payload.update(memstats)
ioStats = sys_checks["io"].check(self.agentConfig)
if ioStats:
payload["ioStats"] = ioStats
processes = sys_checks["processes"].check(self.agentConfig)
payload.update({"processes": processes})
cpuStats = sys_checks["cpu"].check(self.agentConfig)
if cpuStats:
payload.update(cpuStats)
# Run old-style checks
gangliaData = self._ganglia.check(self.agentConfig)
dogstreamData = self._dogstream.check(self.agentConfig)
ddforwarderData = self._ddforwarder.check(self.agentConfig)
if gangliaData is not False and gangliaData is not None:
payload["ganglia"] = gangliaData
# dogstream
if dogstreamData:
dogstreamEvents = dogstreamData.get("dogstreamEvents", None)
if dogstreamEvents:
if "dogstream" in payload["events"]:
events["dogstream"].extend(dogstreamEvents)
else:
events["dogstream"] = dogstreamEvents
del dogstreamData["dogstreamEvents"]
payload.update(dogstreamData)
# metrics about the forwarder
if ddforwarderData:
payload["datadog"] = ddforwarderData
# Resources checks
if not Platform.is_windows():
has_resource = False
for resources_check in self._resources_checks:
try:
resources_check.check()
snaps = resources_check.pop_snapshots()
if snaps:
has_resource = True
res_value = {"snaps": snaps, "format_version": resources_check.get_format_version()}
res_format = resources_check.describe_format_if_needed()
if res_format is not None:
res_value["format_description"] = res_format
payload["resources"][resources_check.RESOURCE_KEY] = res_value
except Exception:
log.exception("Error running resource check %s" % resources_check.RESOURCE_KEY)
if has_resource:
payload["resources"]["meta"] = {
"api_key": self.agentConfig["api_key"],
"host": payload["internalHostname"],
}
# newer-style checks (not checks.d style)
for metrics_check in self._metrics_checks:
res = metrics_check.check(self.agentConfig)
if res:
metrics.extend(res)
# checks.d checks
check_statuses = []
for check in self.initialized_checks_d:
if not self.continue_running:
return
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_start_time = time.time()
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
check_stats = check._get_internal_profiling_stats()
# Collect metadata
current_check_metadata = check.get_service_metadata()
# Save metrics & events for the payload.
metrics.extend(current_check_metrics)
if current_check_events:
if check.name not in events:
events[check.name] = current_check_events
else:
events[check.name] += current_check_events
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name,
instance_statuses,
metric_count,
event_count,
service_check_count,
service_metadata=current_check_metadata,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats,
)
# Service check for Agent checks failures
service_check_tags = ["check:%s" % check.name]
if check_status.status == STATUS_OK:
status = AgentCheck.OK
elif check_status.status == STATUS_ERROR:
status = AgentCheck.CRITICAL
check.service_check("datadog.agent.check_status", status, tags=service_check_tags)
# Collect the service checks and save them in the payload
current_check_service_checks = check.get_service_checks()
if current_check_service_checks:
service_checks.extend(current_check_service_checks)
service_check_count = len(current_check_service_checks)
# Update the check status with the correct service_check_count
check_status.service_check_count = service_check_count
check_statuses.append(check_status)
check_run_time = time.time() - check_start_time
log.debug("Check %s ran in %.2f s" % (check.name, check_run_time))
# Intrument check run timings if enabled.
if self.check_timings:
metric = "datadog.agent.check_run_time"
meta = {"tags": ["check:%s" % check.name]}
metrics.append((metric, time.time(), check_run_time, meta))
for check_name, info in self.init_failed_checks_d.iteritems():
if not self.continue_running:
return
check_status = CheckStatus(
check_name,
None,
None,
None,
None,
init_failed_error=info["error"],
init_failed_traceback=info["traceback"],
)
check_statuses.append(check_status)
# Add a service check for the agent
service_checks.append(create_service_check("datadog.agent.up", AgentCheck.OK, hostname=self.hostname))
# Store the metrics and events in the payload.
payload["metrics"] = metrics
payload["events"] = events
payload["service_checks"] = service_checks
# Populate metadata
self._populate_payload_metadata(payload, check_statuses, start_event)
collect_duration = timer.step()
if self._agent_metrics:
metric_context = {"collection_time": collect_duration, "emit_time": self.emit_duration}
if not Platform.is_windows():
metric_context["cpu_time"] = time.clock() - cpu_clock
self._agent_metrics.set_metric_context(payload, metric_context)
self._agent_metrics.run()
agent_stats = self._agent_metrics.get_metrics()
payload["metrics"].extend(agent_stats)
if self.agentConfig.get("developer_mode"):
log.debug("\n Agent developer mode stats: \n {0}".format(Collector._stats_for_display(agent_stats)))
# Let's send our payload
emitter_statuses = payload.emit(log, self.agentConfig, self.emitters, self.continue_running)
self.emit_duration = timer.step()
# Persist the status of the collection run.
try:
CollectorStatus(check_statuses, emitter_statuses, self.hostname_metadata_cache).persist()
except Exception:
log.exception("Error persisting collector status")
if self.run_count <= FLUSH_LOGGING_INITIAL or self.run_count % FLUSH_LOGGING_PERIOD == 0:
log.info(
"Finished run #%s. Collection time: %ss. Emit time: %ss"
% (self.run_count, round(collect_duration, 2), round(self.emit_duration, 2))
)
if self.run_count == FLUSH_LOGGING_INITIAL:
log.info("First flushes done, next flushes will be logged every %s flushes." % FLUSH_LOGGING_PERIOD)
else:
log.debug(
"Finished run #%s. Collection time: %ss. Emit time: %ss"
% (self.run_count, round(collect_duration, 2), round(self.emit_duration, 2))
)
return payload
@staticmethod
def run_single_check(check, verbose=True):
log.info("Running check %s" % check.name)
instance_statuses = []
metric_count = 0
event_count = 0
service_check_count = 0
check_stats = None
try:
# Run the check.
instance_statuses = check.run()
# Collect the metrics and events.
current_check_metrics = check.get_metrics()
current_check_events = check.get_events()
current_service_checks = check.get_service_checks()
current_service_metadata = check.get_service_metadata()
check_stats = check._get_internal_profiling_stats()
# Save the status of the check.
metric_count = len(current_check_metrics)
event_count = len(current_check_events)
service_check_count = len(current_service_checks)
print "Metrics: \n{0}".format(pprint.pformat(current_check_metrics))
print "Events: \n{0}".format(pprint.pformat(current_check_events))
print "Service Checks: \n{0}".format(pprint.pformat(current_service_checks))
print "Service Metadata: \n{0}".format(pprint.pformat(current_service_metadata))
except Exception:
log.exception("Error running check %s" % check.name)
check_status = CheckStatus(
check.name,
instance_statuses,
metric_count,
event_count,
service_check_count,
library_versions=check.get_library_info(),
source_type_name=check.SOURCE_TYPE_NAME or check.name,
check_stats=check_stats,
)
return check_status
def _emit(self, payload):
""" Send the payload via the emitters. """
statuses = []
for emitter in self.emitters:
# Don't try to send to an emitter if we're stopping/
if not self.continue_running:
return statuses
name = emitter.__name__
emitter_status = EmitterStatus(name)
try:
emitter(payload, log, self.agentConfig)
except Exception, e:
log.exception("Error running emitter: %s" % emitter.__name__)
emitter_status = EmitterStatus(name, e)
statuses.append(emitter_status)
return statuses