def main():
"""Main program.
Parse arguments, start webserver to serve /metrics.
"""
args = ARGS.parse_args()
if args.level >= 2:
logging.basicConfig(level=logging.DEBUG)
elif args.level == 1:
logging.basicConfig(level=logging.INFO)
elif args.level == 0:
logging.basicConfig(level=logging.WARNING)
LOG.debug('Starting HTTP server')
httpd_thread = start_http_server(args.port, addr=args.addr)
LOG.debug('Registering StatsPluginCollector')
REGISTRY.register(StatsPluginCollector(args.endpoint))
if not args.no_procstats:
LOG.debug('Registering ProcessCollector')
REGISTRY.register(ProcessCollector(
pid=lambda: get_ts_pid(args.pidfile),
namespace='trafficserver'))
LOG.info('Listening on :{port}'.format(port=args.port))
# Wait for the webserver
httpd_thread.join()
def test_enabled(self):
self.assertEqual(1, REGISTRY.get_sample_value('python_gc_enabled'))
try:
gc.disable()
self.assertEqual(0, REGISTRY.get_sample_value('python_gc_enabled'))
finally:
gc.enable()
def _assert_task_states(self, states, cnt):
for state in states:
assert REGISTRY.get_sample_value(
'celery_tasks', labels=dict(state=state)) == cnt
task_by_name_label = dict(state=state, name=self.task)
assert REGISTRY.get_sample_value(
'celery_tasks_by_name', labels=task_by_name_label) == cnt
def test_debug(self):
self.assertEqual(0, REGISTRY.get_sample_value('python_gc_debug'))
try:
gc.set_debug(gc.DEBUG_STATS)
self.assertEqual(gc.DEBUG_STATS, REGISTRY.get_sample_value(
'python_gc_enabled'))
finally:
gc.set_debug(0)
def ttl_watchdog_unregister_old_metrics(now):
for (name, last_update) in list(GAUGES_LAST_UPDATE.items()):
if now - last_update > GAUGES_TTL:
REGISTRY.unregister(GAUGES[name])
del GAUGES[name]
del GAUGES_LAST_UPDATE[name]
for (other_name, label_values) in list(GAUGES_LABELS_LAST_UPDATE.keys()):
if name == other_name:
del GAUGES_LABELS_LAST_UPDATE[(name, label_values)]
def main():
"""
Symbols list contain a list of pairs which describes stock symbols as used by Google API.
Each element should be 'EXCHANGE:SYMBOL' examples:
[ 'NASDAQ:GOOG', 'NASDAQ:CSCO', 'NYSE:IBM', 'BCBA:YPFD' ]
"""
start_http_server(int(sys.argv[1]))
REGISTRY.register(QuoteCollector())
while True: time.sleep(1)
def test_rootLoggerExports(self):
logging.error('There was an error.')
self.assertEquals(
1, REGISTRY.get_sample_value('python_logging_messages_total',
labels={'logger': 'test_levels',
'level': 'ERROR'}))
def test_set_zero_on_queue_length_when_an_channel_layer_error_occurs_during_queue_read(self):
instance = QueueLenghtMonitoringThread(app=self.app, queue_list=['noqueue'])
instance.measure_queues_length()
sample = REGISTRY.get_sample_value('celery_queue_length', {'queue_name':'noqueue'})
self.assertEqual(0.0, sample)
def test_prometheus_metrics_guage():
metrics = PrometheusMetricsFactory(namespace='test')
gauge = metrics.create_gauge(name='jaeger:test_gauge',
tags={'result': 'ok'})
gauge(1)
after = REGISTRY.get_sample_value('test_jaeger:test_gauge',
{'result': 'ok'})
assert 1 == after
def test_setLevel(self):
logger = logging.getLogger('test_setLevel')
logger.setLevel(logging.CRITICAL)
logger.debug('debug message')
logger.info('info message')
logger.warn('warn message')
logger.error('error message')
logger.critical('critical message')
for level in ('DEBUG', 'INFO', 'WARNING', 'ERROR'):
self.assertEquals(
None,
REGISTRY.get_sample_value('python_logging_messages_total',
labels={'logger': 'test_setLevel',
'level': level}))
self.assertEquals(
1, REGISTRY.get_sample_value('python_logging_messages_total',
labels={'logger': 'test_levels',
'level': 'CRITICAL'}))
def prometheus_exporter():
registry = REGISTRY
encoder, content_type = exposition.choose_encoder(
request.headers.get('Accept'))
if 'name[]' in request.args:
registry = REGISTRY.restricted_registry(
request.args.get('name[]'))
body = encoder(registry)
return Response(body, content_type=content_type)
def main(args):
port = int(args.port)
with open(args.metrics_file_path) as metrics_file:
metrics_yaml = metrics_file.read()
metrics = parse_aws_metrics(metrics_yaml)
collector = AwsMetricsCollector(metrics,
boto3.Session(region_name=args.region))
REGISTRY.register(collector)
start_http_server(port)
print("Serving at port: %s" % port)
while True:
try:
collector.update()
time.sleep(args.period_seconds)
except KeyboardInterrupt:
print("Caught SIGTERM - stopping...")
break
print("Done.")
def test_request_latency_count(self):
method, path = 'TEST', 'PATH'
before = REGISTRY.get_sample_value(
'http_requests_latency_seconds_count', {
'method': method,
'path': path
}) or 0
trigger_metrics(method, path, wait=True)
after = REGISTRY.get_sample_value(
'http_requests_latency_seconds_count', {
'method': method,
'path': path
})
assert after is not None
assert after - before == 1
def _app():
class Collector:
def __init__(self):
self.metrics = []
def set_metrics(self, metrics):
self.metrics = metrics
def collect(self):
return self.metrics
app = vadvisor.app.rest.app
app.collector = Collector()
app.metricStore = MetricStore()
REGISTRY.register(LibvirtCollector(collector=app.collector))
return app
def main(args):
port = int(args.port)
with open(args.metrics_file_path) as metrics_file:
metrics_yaml = metrics_file.read()
metrics = parse_aws_metrics(metrics_yaml)
collector = AwsMetricsCollector(metrics, boto3.client("sts"), args.assume_role_arn, "prometheusAssumeRole", args.duration_seconds)
REGISTRY.register(collector)
start_http_server(port)
print("Serving at port: %s" % port)
while True:
try:
print("Starting the collection again : ", datetime.datetime.now())
collector.update()
time.sleep(args.period_seconds)
except KeyboardInterrupt:
print("Caught SIGTERM - stopping...")
break
print("Done.")
def main():
"""Main entry point"""
port = int(os.getenv("PORT", "9877"))
configFile = os.getenv("CONFIG", "config.yaml")
if os.getenv('KUBERNETES_SERVICE_HOST'):
config.load_incluster_config()
else:
config.load_kube_config()
with open(configFile) as f:
resources = yaml.load(f, Loader=yaml.FullLoader)
start_http_server(port, registry=REGISTRY)
REGISTRY.register(CustomCollector(resources))
while True:
time.sleep(1)
def test_assign_report_slice_new(self):
"""Test the assign report slice function with only a new report slice."""
self.report_slice.state = ReportSlice.NEW
self.report_slice.save()
self.processor.report_or_slice = None
self.processor.assign_object()
self.assertEqual(self.processor.report_or_slice, self.report_slice)
queued_slices = REGISTRY.get_sample_value('queued_report_slices')
self.assertEqual(queued_slices, 1)
def make_rest_app(libvirtConnection):
# start libvirt event broker
broker = LibvirtEventBroker()
Greenlet(broker.run).start()
app.eventBroker = broker
# Attach event store to broker
app.eventStore = EventStore()
def store_events():
q = queue.Queue()
broker.subscribe(q)
while True:
app.eventStore.put(q.get())
Greenlet(store_events).start()
# Create metric collector
app.conn = libvirtConnection
app.collector = Collector(app.conn)
# Register prometheus metrics
REGISTRY.register(LibvirtCollector(app.collector))
# For statsd debugging
app.statsd = StatsdCollector(app.collector)
# Collect metrics every second and store them in the metrics store
app.metricStore = MetricStore()
def store_metrics():
while True:
try:
app.metricStore.put(app.collector.collect())
except Exception as e:
logging.error(e)
sleep(5)
sleep(1)
Greenlet(store_metrics).start()
# Add gzip support
mime_types = ['application/json', 'text/plain']
return gzip(mime_types=mime_types, compress_level=9)(app)
def create_app() -> "app":
app = Flask(__name__)
# Unregister all collectors.
collectors = list(REGISTRY._collector_to_names.keys())
print(f"before unregister collectors={collectors}")
for collector in collectors:
REGISTRY.unregister(collector)
print(
f"after unregister collectors={list(REGISTRY._collector_to_names.keys())}"
)
# Import default collectors.
from prometheus_client import (gc_collector, platform_collector,
process_collector)
# Re-register default collectors.
process_collector.ProcessCollector()
platform_collector.PlatformCollector()
gc_collector.GCCollector()
@app.route("/")
def home():
return "Hello World!"
@app.route("/path/<page_name>")
def other_page(page_name):
return page_name
@app.route("/to/exclude")
def exclude():
return "Exclude me!"
@app.route("/server_error")
def server_error():
raise Exception("Test")
return "will ever get here"
@app.route("/ignored")
@Instrumentator.do_not_track()
def ignored():
return "HALLO"
return app
def make_rest_app(libvirtConnection):
# start libvirt event broker
broker = LibvirtEventBroker()
Greenlet(broker.run).start()
app.eventBroker = broker
# Attach event store to broker
app.eventStore = EventStore()
def store_events():
q = queue.Queue()
broker.subscribe(q)
while True:
app.eventStore.put(q.get())
Greenlet(store_events).start()
# Create metric collector
app.conn = libvirtConnection
app.collector = Collector(app.conn)
# Register prometheus metrics
REGISTRY.register(LibvirtCollector(app.collector))
# For statsd debugging
app.statsd = StatsdCollector(app.collector)
# Collect metrics every second and store them in the metrics store
app.metricStore = MetricStore()
def store_metrics():
while True:
try:
app.metricStore.put(app.collector.collect())
except Exception as e:
logging.error(e)
sleep(5)
sleep(1)
Greenlet(store_metrics).start()
# Add gzip support
mime_types = ['application/json', 'text/plain']
return gzip(mime_types=mime_types, compress_level=9)(app)
def main():
opts = parse_opts(sys.argv[1:])
init_logger(opts.log_level)
scheduler = Scheduler()
def sigterm_handler(signum, frame):
if scheduler and signal.SIGTERM == signum:
scheduler.shutdown()
signal.signal(signal.SIGTERM, sigterm_handler)
sonarqube_client = SonarQubeClient(opts.url, opts.user_token, **{"verify": opts.ignore_ssl})
sonar_collector = SonarQubeCollector(sonarqube_client)
REGISTRY.register(sonar_collector)
scheduler.schedule(sonar_collector, int(opts.interval))
scheduler.start()
start_http_server(int(opts.port))
sys.exit(scheduler.wait())
def test_set_zero_on_queue_length_when_an_channel_layer_error_occurs_during_queue_read(
self):
instance = QueueLengthMonitoringThread(app=self.app,
queue_list=['noqueue'])
instance.measure_queues_length()
sample = REGISTRY.get_sample_value('celery_queue_length',
{'queue_name': 'noqueue'})
self.assertEqual(0.0, sample)
def test_info_gauge():
dct = {
"foo": "bar",
"hello": "friend",
}
telemetry.info(dct, "test_info_gauge")
assert (REGISTRY.get_sample_value(
f"{s.PROMETHEUS_NAMESPACE}_test_info_gauge", dct) == 1)
def test_rootLoggerExports(self):
logging.error('There was an error.')
self.assertEquals(
1,
REGISTRY.get_sample_value('python_logging_messages_total',
labels={
'logger': 'test_levels',
'level': 'ERROR'
}))
def test_latency(icmp_mocked_host):
"""
Test if ping succeeds and generates the right metrics.
"""
clean_prometheus_registry()
reload(monitor)
before_up = REGISTRY.get_sample_value("internet_monitor_up")
before_ping_total = REGISTRY.get_sample_value(
"internet_monitor_ping_total")
before_ping_packet_loss_total = REGISTRY.get_sample_value(
"internet_monitor_ping_packet_loss_total")
before_ping_jitter_seconds = REGISTRY.get_sample_value(
"internet_monitor_ping_jitter_seconds")
before_ping_latency_seconds = REGISTRY.get_sample_value(
"internet_monitor_ping_latency_seconds")
assert before_up == 0
assert before_ping_total == 0
assert before_ping_packet_loss_total == 0
assert before_ping_jitter_seconds == 0
with patch("im.monitor.ping") as mock_ping:
mock_ping.return_value = icmp_mocked_host
monitor.latency("1.1.1.1")
after_up = REGISTRY.get_sample_value("internet_monitor_up")
after_ping_total = REGISTRY.get_sample_value("internet_monitor_ping_total")
after_ping_packet_loss_total = REGISTRY.get_sample_value(
"internet_monitor_ping_packet_loss_total")
after_ping_jitter_seconds = REGISTRY.get_sample_value(
"internet_monitor_ping_jitter_seconds")
after_ping_latency_seconds = REGISTRY.get_sample_value(
"internet_monitor_ping_latency_seconds_sum")
assert after_up == 1
assert after_ping_total == 1
assert after_ping_packet_loss_total == icmp_mocked_host.packet_loss
assert (after_ping_jitter_seconds == (icmp_mocked_host.max_rtt -
icmp_mocked_host.min_rtt) / 1000)
assert after_ping_latency_seconds == icmp_mocked_host.avg_rtt / 1000
def test_prometheus_metrics_counter():
metrics = PrometheusMetricsFactory(namespace='test')
counter1 = metrics.create_counter(name='jaeger:test_counter',
tags={'result': 'ok'})
counter1(1)
counter2 = metrics.create_counter(name='jaeger:test_counter',
tags={'result': 'ok'})
counter2(1)
after = REGISTRY.get_sample_value('test_jaeger:test_counter',
{'result': 'ok'})
assert 2 == after
def test_prometheus_metrics_counter():
metrics = PrometheusMetricsFactory(namespace='test')
counter1 = metrics.create_counter(name='jaeger:test_counter',
tags={'result': 'ok'})
counter1(1)
counter2 = metrics.create_counter(name='jaeger:test_counter',
tags={'result': 'ok'})
counter2(1)
after = REGISTRY.get_sample_value('test_jaeger:test_counter',
{'result': 'ok'})
assert 2 == after
def test_can_measure_queue_length(self):
celery_app = get_celery_app(queue='realqueue')
sample_task = SampleTask()
sample_task.app = celery_app
monitoring_thread_instance = QueueLenghtMonitoringThread(celery_app, queue_list=['realqueue'])
sample_task.delay()
monitoring_thread_instance.measure_queues_length()
sample = REGISTRY.get_sample_value('celery_queue_length', {'queue_name':'realqueue'})
self.assertEqual(1.0, sample)
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--config.file", dest="config_file", default="aci.yml")
parser.add_argument("--web.listen-port", dest="web_listen_port", type=int, default=DEFAULT_PORT)
parser.add_argument("--web.listen-address", dest="web_listen_address", type=str, default="")
args = parser.parse_args()
config = load_config(args.config_file)
aci_collector = AciCollector(config)
REGISTRY.register(aci_collector)
if hasattr(signal, 'SIGHUP'):
sighup_handler = get_sighup_handler(aci_collector, args.config_file)
signal.signal(signal.SIGHUP, sighup_handler)
start_http_server(args.web_listen_port, args.web_listen_address)
while True:
time.sleep(9001)
def test_add_probability_filter_passed_no_nodes_metric(
prometheus_noop_scenario):
"""
Ensures that add_probability_filter_passed_no_nodes_metric is called when
the filter decides to pass no nodes based on a probability
"""
assert prometheus_noop_scenario.name == "test scenario"
random.seed(6) # make the tests deterministic
candidates = [dummy_object()]
before = REGISTRY.get_sample_value(
PROBABILITY_FILTER_NOT_PASSED_METRIC_NAME)
assert before == 0
criterion = {"probabilityPassAll": 0.00000001}
prometheus_noop_scenario.filter_probability(candidates, criterion)
after = REGISTRY.get_sample_value(
PROBABILITY_FILTER_NOT_PASSED_METRIC_NAME)
assert after == 1
def test_can_measure_queue_length(self):
celery_app = get_celery_app(queue='realqueue')
sample_task = SampleTask()
sample_task.app = celery_app
monitoring_thread_instance = QueueLengthMonitoringThread(celery_app, queue_list=['realqueue'])
sample_task.delay()
monitoring_thread_instance.measure_queues_length()
sample = REGISTRY.get_sample_value('celery_queue_length', {'queue_name':'realqueue'})
self.assertEqual(1.0, sample)
def main(port, config_file=None):
"""
Main function.
Parse config, create Aws connections to read data and create a web-server for prometheus
:param port: The http port the server will listen for incoming http requests
:type port: int
:param config_file: The path of the config file, optional. If none, look for config in the script folder
:type config_file: str|None
:return: The exit code
:rtype: int
"""
try:
if not port:
port = DEFAULT_PORT
if not config_file:
config_file = os.path.join(
os.path.dirname(os.path.realpath(__file__)), "config.yml")
if not os.path.exists(config_file):
raise ExporterError("Unable to load config file '" +
str(config_file) + "'")
with open(config_file, "r") as cfg_fh:
cfg_content = cfg_fh.read()
collector = BkpCollector(yaml.load(cfg_content))
REGISTRY.register(collector)
start_http_server(port)
while True:
time.sleep(0.1)
except KeyboardInterrupt:
print("\nExiting, please wait...")
return 0
except SystemExit:
raise
except ExporterError as error:
sys.stderr.write(error.message)
sys.stderr.write("\n")
sys.stderr.flush()
return 1
def test_workers_count(self):
assert REGISTRY.get_sample_value('celery_workers') == 0
with patch.object(self.app.control, 'ping') as mock_ping:
w = WorkerMonitoringThread(app=self.app)
mock_ping.return_value = []
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 0
mock_ping.return_value = [0] # 1 worker
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 1
mock_ping.return_value = [0, 0] # 2 workers
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 2
mock_ping.return_value = []
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 0
def start_webserver(conf):
token_manager = TokenManager(conf.get('tableau_user'),
conf.get('tableau_password'),
conf['site'],
conf['server_host'],
conf['api_version'],
token_name=conf.get('tableau_token_name'),
token_secret=conf.get('tableau_token_secret'))
REGISTRY.register(
TableauMetricsCollector(token_manager,
verify_ssl=conf.get('verify_ssl', False)))
# Start up the server to expose the metrics.
root = Resource()
root.putChild(b'metrics', MetricsResource())
factory = Site(root)
logger.info('Starting webserver on {}'.format(conf['exporter_port']))
reactor.listenTCP(conf['exporter_port'], factory)
reactor.run()
def __init__(self, port=8000, counters={}):
"""Start the http server for scraping.
The port where you open should be scraped by prometheus
e.g port 8000 could have:
scrape_configs:
- job_name: 'local'
scrape_interval: 5s
static_configs:
- targets: ['localhost:8000']
The counters dict given with will be used to keep track of counters and
will be given to prometheus when scraped
"""
start_http_server(port)
# keep data for scrape_interval * scrape_amount
# (scrape_interval is found in /etc/prometheus/prometheus.yml,
# when writing it is 5)
self.scrape_amount = 60
self.scrape_count = self.scrape_amount // 2
self.data = [[] for _ in range(self.scrape_amount)]
self.counters = counters
REGISTRY.register(self)
def test_workers_count(self):
assert REGISTRY.get_sample_value('celery_workers') == 0
with patch.object(self.app.control, 'ping') as mock_ping:
w = WorkerMonitoringThread(app=self.app)
mock_ping.return_value = []
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 0
mock_ping.return_value = [0] # 1 worker
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 1
mock_ping.return_value = [0, 0] # 2 workers
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 2
mock_ping.return_value = []
w.update_workers_count()
assert REGISTRY.get_sample_value('celery_workers') == 0
def test_download_speed(http_mocked_response):
"""
Test if download succeeds and calculates the appropriate size and successfull job execution.
"""
clean_prometheus_registry()
reload(monitor)
before_download_duration_seconds_sum = REGISTRY.get_sample_value(
"internet_monitor_download_duration_seconds_sum")
before_download_size_bytes = REGISTRY.get_sample_value(
"internet_monitor_download_size_bytes")
before_download_total = REGISTRY.get_sample_value(
"internet_monitor_download_total")
assert before_download_duration_seconds_sum == 0
assert before_download_size_bytes == 0
assert before_download_total == 0
with patch("im.monitor.requests") as mocked_requests:
mocked_requests.get.return_value = http_mocked_response
monitor.download_speed("http://localhost")
after_download_duration_seconds_sum = REGISTRY.get_sample_value(
"internet_monitor_download_duration_seconds_sum")
after_download_size_bytes = REGISTRY.get_sample_value(
"internet_monitor_download_size_bytes")
after_download_total = REGISTRY.get_sample_value(
"internet_monitor_download_total")
assert isinstance(after_download_duration_seconds_sum, float)
assert after_download_size_bytes == len(http_mocked_response.content)
assert after_download_total == 1
def test_upload_speed_failures(http_mocked_response):
"""
Test if upload fails and failure metrics are updated accordingly.
"""
clean_prometheus_registry()
reload(monitor)
before_upload_duration_seconds_sum = REGISTRY.get_sample_value(
"internet_monitor_upload_duration_seconds_sum")
before_upload_failures_total = REGISTRY.get_sample_value(
"internet_monitor_upload_failures_total")
before_upload_total = REGISTRY.get_sample_value(
"internet_monitor_upload_total")
assert before_upload_duration_seconds_sum == 0
assert before_upload_failures_total == 0
assert before_upload_total == 0
with patch("im.monitor.requests") as mocked_requests:
mocked_requests.post.side_effect = Exception("boom!")
monitor.upload_speed("http://localhost")
after_upload_duration_seconds_sum = REGISTRY.get_sample_value(
"internet_monitor_upload_duration_seconds_sum")
after_upload_failures_total = REGISTRY.get_sample_value(
"internet_monitor_upload_failures_total")
after_upload_total = REGISTRY.get_sample_value(
"internet_monitor_upload_total")
assert isinstance(after_upload_duration_seconds_sum, float)
assert after_upload_failures_total == 1
assert after_upload_total == 1
def test_latency_failure(icmp_mocked_host):
"""
Test if ping fails and generates the right metrics.
"""
clean_prometheus_registry()
reload(monitor)
before_up = REGISTRY.get_sample_value("internet_monitor_up")
before_ping_total = REGISTRY.get_sample_value(
"internet_monitor_ping_total")
before_ping_failures_total = REGISTRY.get_sample_value(
"internet_monitor_ping_failures_total")
assert before_up == 0
assert before_ping_total == 0
assert before_ping_failures_total == 0
with patch("im.monitor.ping") as mock_ping:
mock_ping.side_effect = Exception("boom!")
monitor.latency("1.1.1.1")
after_up = REGISTRY.get_sample_value("internet_monitor_up")
after_ping_total = REGISTRY.get_sample_value("internet_monitor_ping_total")
before_ping_failures_total = REGISTRY.get_sample_value(
"internet_monitor_ping_failures_total")
assert after_up == 0
assert after_ping_total == 1
assert before_ping_failures_total == 1
def test_converge_complete(self):
"""
At the end of a convergence iteration, ``_CONVERGE_COMPLETE`` is updated
to the current time.
"""
interval = 45
reactor = MemoryReactorClock()
deploy_config = DeploymentConfiguration(
domain=u"s4.example.com",
kubernetes_namespace=u"testing",
subscription_manager_endpoint=URL.from_text(
u"http://localhost:8000"),
s3_access_key_id=u"access key id",
s3_secret_key=u"secret key",
introducer_image=u"introducer:abcdefgh",
storageserver_image=u"storageserver:abcdefgh",
)
state_path = FilePath(self.mktemp().decode("ascii"))
state_path.makedirs()
subscription_client = memory_client(
state_path,
deploy_config.domain,
)
k8s_client = KubeClient(k8s=memory_kubernetes().client())
aws_region = FakeAWSServiceRegion(
access_key=deploy_config.s3_access_key_id,
secret_key=deploy_config.s3_secret_key,
)
d = aws_region.get_route53_client().create_hosted_zone(
u"foo",
deploy_config.domain,
)
self.successResultOf(d)
service = _convergence_service(
reactor,
interval,
deploy_config,
subscription_client,
k8s_client,
aws_region,
)
service.startService()
reactor.advance(interval)
last_completed = next(
iter(
list(metric.samples[-1][-1] for metric in REGISTRY.collect()
if metric.name == u"s4_last_convergence_succeeded")))
self.assertThat(reactor.seconds(), Equals(last_completed))
def test_converge_complete(self):
"""
At the end of a convergence iteration, ``_CONVERGE_COMPLETE`` is updated
to the current time.
"""
interval = 45
reactor = MemoryReactorClock()
deploy_config = DeploymentConfiguration(
domain=u"s4.example.com",
kubernetes_namespace=u"testing",
subscription_manager_endpoint=URL.from_text(u"http://localhost:8000"),
s3_access_key_id=u"access key id",
s3_secret_key=u"secret key",
introducer_image=u"introducer:abcdefgh",
storageserver_image=u"storageserver:abcdefgh",
)
state_path = FilePath(self.mktemp().decode("ascii"))
state_path.makedirs()
subscription_client = memory_client(
state_path,
deploy_config.domain,
)
k8s_client = KubeClient(k8s=memory_kubernetes().client())
aws_region = FakeAWSServiceRegion(
access_key=deploy_config.s3_access_key_id,
secret_key=deploy_config.s3_secret_key,
)
d = aws_region.get_route53_client().create_hosted_zone(
u"foo", deploy_config.domain,
)
self.successResultOf(d)
service = _convergence_service(
reactor,
interval,
deploy_config,
subscription_client,
k8s_client,
aws_region,
)
service.startService()
reactor.advance(interval)
last_completed = next(iter(list(
metric.samples[-1][-1]
for metric
in REGISTRY.collect()
if metric.name == u"s4_last_convergence_succeeded"
)))
self.assertThat(reactor.seconds(), Equals(last_completed))
def main():
parser = argparse.ArgumentParser()
parser.add_argument('-l',
'--listen',
metavar='ADDRESS',
help='Listen on this address',
default=':8000')
parser.add_argument('-u',
'--uri',
default='/',
help='The URI to check for POSTs coming from Druid')
parser.add_argument('-d',
'--debug',
action='store_true',
help='Enable debug logging')
parser.add_argument('-e',
'--encoding',
default='utf-8',
help='Encoding of the Druid POST JSON data.')
args = parser.parse_args()
if args.debug:
logging.basicConfig(level=logging.DEBUG)
else:
logging.basicConfig(level=logging.WARNING)
collect_metrics_from = []
address, port = args.listen.split(':', 1)
log.info('Starting druid_exporter on %s:%s', address, port)
druid_collector = collector.DruidCollector()
REGISTRY.register(druid_collector)
prometheus_app = make_wsgi_app()
druid_wsgi_app = DruidWSGIApp(args.uri, druid_collector, prometheus_app,
args.encoding)
httpd = make_server(address, int(port), druid_wsgi_app)
httpd.serve_forever()
def __init__(self, rdc_ip_port, field_ids, update_freq, max_keep_age,
max_keep_samples, gpu_indexes, rdc_unauth,
enable_plugin_monitoring):
group_name = "rdc_prometheus_plugin_group"
field_group_name = "rdc_prometheus_plugin_fieldgroup"
if rdc_unauth:
RdcReader.__init__(self,
ip_port=rdc_ip_port,
field_ids=field_ids,
update_freq=update_freq,
max_keep_age=max_keep_age,
max_keep_samples=max_keep_samples,
gpu_indexes=gpu_indexes,
field_group_name=field_group_name,
gpu_group_name=group_name,
root_ca=None)
else:
RdcReader.__init__(self,
ip_port=rdc_ip_port,
field_ids=field_ids,
update_freq=update_freq,
max_keep_age=max_keep_age,
max_keep_samples=max_keep_samples,
gpu_indexes=gpu_indexes,
field_group_name=field_group_name,
gpu_group_name=group_name)
# Supress internal metrics from prometheus_client
if enable_plugin_monitoring == False:
REGISTRY.unregister(PROCESS_COLLECTOR)
REGISTRY.unregister(PLATFORM_COLLECTOR)
# Create the guages
self.guages = {}
for fid in self.field_ids:
field_name = self.rdc_util.field_id_string(fid).lower()
self.guages[fid] = Gauge(field_name,
field_name,
labelnames=['gpu_index'])
def start_metrics_server():
"""
Starts a http server on a port specified in the configuration file
and exposes Prometheus metrics on it.
Also removes GC_COLLECTOR metrics because they are not really needed.
"""
# Remove garbage collection metrics
REGISTRY.unregister(GC_COLLECTOR)
# Gather configurations
config = Configuration().values
ip = config.service.ip
metrics_port = config.service.metrics_port
# Start server
start_wsgi_server(metrics_port)
# Log
Logger() \
.event(category="runnable", action="run metrics") \
.server(ip=ip, port=metrics_port) \
.out(severity=Severity.INFO)
def test_request_latency_sum(self):
method, path = 'TEST', 'PATH'
before = REGISTRY.get_sample_value('http_requests_latency_seconds_sum',
{
'method': method,
'path': path
}) or 0
trigger_metrics(method, path, wait=True)
after = REGISTRY.get_sample_value('http_requests_latency_seconds_sum',
{
'method': method,
'path': path
})
assert after is not None
diff = after - before
# Check the difference is roughly in the ballpark of what we expect.
assert (diff >= 2) and (diff <= 3)
def test_combined_size_no_labels():
app = create_app()
Instrumentator().add(
metrics.combined_size(
should_include_handler=False,
should_include_method=False,
should_include_status=False,
)).instrument(app)
client = TestClient(app)
client.get("/")
assert REGISTRY.get_sample_value("http_combined_size_bytes_sum", {}) == 14
def cli(**settings):
"""Kayako metrics exporter for Prometheus"""
if settings['version']:
click.echo('Version %s' % kayako_exporter.__version__)
return
if not settings['url']:
click.echo('Please provide Kayako API URL')
return
if not settings['login']:
click.echo('Please provide Kayako username')
return
if not settings['password']:
click.echo('Please provide Kayako account password')
return
if settings['verbose']:
handler = logging.StreamHandler()
logger.addHandler(handler)
logger.setLevel(logging.DEBUG)
handler.setFormatter(logging.Formatter('[%(asctime)s] %(message)s', "%Y-%m-%d %H:%M:%S"))
click.echo('Exporter for {base_url}, user: {login}, password: ***'.format(
base_url=settings['url'].rstrip('/'),
login=settings['login'],
password=settings['password']
))
REGISTRY.register(KayakoCollector(
base_url=settings['url'].rstrip('/'),
login=settings['login'],
password=settings['password'],
department_ids=settings['department_id'],
))
httpd = HTTPServer(('', int(settings['port'])), MetricsHandler)
click.echo('Exporting Kayako metrics on http://0.0.0.0:{}'.format(settings['port']))
httpd.serve_forever()
def getMetricVector(self, metric_name):
"""Returns the values for all labels of a given metric.
The result is returned as a list of (labels, value) tuples,
where `labels` is a dict.
This is quite a hack since it relies on the internal
representation of the prometheus_client, and it should
probably be provided as a function there instead.
"""
all_metrics = REGISTRY.collect()
output = []
for metric in all_metrics:
for n, l, value in metric._samples:
if n == metric_name:
output.append((l, value))
return output
def test_thresholds(self):
self.assertTrue(REGISTRY.get_sample_value(
'python_gc_threshold', labels={'generation': '0'}) is not None)
self.assertTrue(REGISTRY.get_sample_value(
'python_gc_threshold', labels={'generation': '1'}) is not None)
self.assertTrue(REGISTRY.get_sample_value(
'python_gc_threshold', labels={'generation': '2'}) is not None)
original_thresholds = gc.get_threshold()
try:
gc.disable()
gc.set_threshold(42, 43, 44)
self.assertEqual(42, REGISTRY.get_sample_value(
'python_gc_threshold', labels={'generation': '0'}))
self.assertEqual(43, REGISTRY.get_sample_value(
'python_gc_threshold', labels={'generation': '1'}))
self.assertEqual(44, REGISTRY.get_sample_value(
'python_gc_threshold', labels={'generation': '2'}))
finally:
gc.set_threshold(*original_thresholds)
gc.enable()
def test_tasks_events(self):
task_uuid = uuid()
hostname = 'myhost'
local_received = time()
latency_before_started = 123.45
runtime = 234.5
m = MonitorThread(app=self.app)
self._assert_task_states(celery.states.ALL_STATES, 0)
assert REGISTRY.get_sample_value('celery_task_latency_count') == 0
assert REGISTRY.get_sample_value('celery_task_latency_sum') == 0
m._process_event(Event(
'task-received', uuid=task_uuid, name=self.task,
args='()', kwargs='{}', retries=0, eta=None, hostname=hostname,
clock=0,
local_received=local_received))
self._assert_all_states({celery.states.RECEIVED})
m._process_event(Event(
'task-started', uuid=task_uuid, hostname=hostname,
clock=1, name=self.task,
local_received=local_received + latency_before_started))
self._assert_all_states({celery.states.STARTED})
m._process_event(Event(
'task-succeeded', uuid=task_uuid, result='42',
runtime=runtime, hostname=hostname, clock=2,
local_received=local_received + latency_before_started + runtime))
self._assert_all_states({celery.states.SUCCESS})
assert REGISTRY.get_sample_value('celery_task_latency_count') == 1
self.assertAlmostEqual(REGISTRY.get_sample_value(
'celery_task_latency_sum'), latency_before_started)
assert REGISTRY.get_sample_value(
'celery_tasks_runtime_seconds_count',
labels=dict(name=self.task)) == 1
assert REGISTRY.get_sample_value(
'celery_tasks_runtime_seconds_sum',
labels=dict(name=self.task)) == 234.5
def test_collection_time(self):
current_total = REGISTRY.get_sample_value(
'python_gc_collection_process_time_total_s')
gc.collect()
self.assertTrue(current_total < REGISTRY.get_sample_value(
'python_gc_collection_process_time_total_s'))
def test_initial_metric_values(self):
self._assert_task_states(celery.states.ALL_STATES, 0)
assert REGISTRY.get_sample_value('celery_workers') == 0
assert REGISTRY.get_sample_value('celery_task_latency_count') == 0
assert REGISTRY.get_sample_value('celery_task_latency_sum') == 0
def getMetric(self, metric_name, **labels):
return REGISTRY.get_sample_value(metric_name, labels=labels)
def cli(**settings):
"""Zabbix metrics exporter for Prometheus
Use config file to map zabbix metrics names/labels into prometheus.
Config below transfroms this:
local.metric[uwsgi,workers,myapp,busy] = 8
local.metric[uwsgi,workers,myapp,idle] = 6
into familiar Prometheus gauges:
uwsgi_workers{instance="host1",app="myapp",status="busy"} 8
uwsgi_workers{instance="host1",app="myapp",status="idle"} 6
YAML:
\b
metrics:
- key: 'local.metric[uwsgi,workers,*,*]'
name: 'uwsgi_workers'
labels:
app: $1
status: $2
reject:
- 'total'
"""
if settings['version']:
click.echo('Version %s' % zabbix_exporter.__version__)
return
if not validate_settings(settings):
return
if settings['config']:
exporter_config = yaml.safe_load(open(settings['config']))
else:
exporter_config = {}
base_logger = logging.getLogger('zabbix_exporter')
handler = logging.StreamHandler()
base_logger.addHandler(handler)
base_logger.setLevel(logging.ERROR)
handler.setFormatter(logging.Formatter('[%(asctime)s] %(message)s', "%Y-%m-%d %H:%M:%S"))
if settings['verbose']:
base_logger.setLevel(logging.DEBUG)
collector = ZabbixCollector(
base_url=settings['url'].rstrip('/'),
login=settings['login'],
password=settings['password'],
verify_tls=settings['verify_tls'],
timeout=settings['timeout'],
**exporter_config
)
if settings['dump_metrics']:
return dump_metrics(collector)
REGISTRY.register(collector)
httpd = HTTPServer(('', int(settings['port'])), MetricsHandler)
click.echo('Exporter for {base_url}, user: {login}, password: ***'.format(
base_url=settings['url'].rstrip('/'),
login=settings['login'],
password=settings['password']
))
if settings['return_server']:
return httpd
click.echo('Exporting Zabbix metrics on http://0.0.0.0:{}'.format(settings['port']))
httpd.serve_forever()
def test_prometheus_metrics_counter_without_tags():
metrics = PrometheusMetricsFactory()
counter = metrics.create_counter(name='jaeger:test_counter_no_tags')
counter(1)
after = REGISTRY.get_sample_value('jaeger:test_counter_no_tags')
assert 1 == after
# DagRun metrics
dag_duration = GaugeMetricFamily(
'airflow_dag_run_duration',
'Duration of currently running dag_runs in seconds',
labels=['dag_id', 'run_id']
)
driver = Session.bind.driver
for dag in get_dag_duration_info():
if driver == 'mysqldb' or driver == 'pysqlite':
dag_duration.add_metric([dag.dag_id, dag.run_id], dag.duration)
else:
dag_duration.add_metric([dag.dag_id, dag.run_id], dag.duration.seconds)
yield dag_duration
REGISTRY.register(MetricsCollector())
class Metrics(BaseView):
@expose('/')
def index(self):
return Response(generate_latest(), mimetype='text/plain')
ADMIN_VIEW = Metrics(category="Prometheus exporter", name="metrics")
class AirflowPrometheusPlugins(AirflowPlugin):
'''plugin for show metrics'''
name = "airflow_prometheus_plugin"
operators = []
def register_slaves_metrics_collector(cls, get_slaves: Callable[[], List['app.master.slave.Slave']]):
if not cls._slaves_collector_is_registered:
REGISTRY.register(SlavesCollector(get_slaves))
cls._slaves_collector_is_registered = True
def test_prometheus_metrics_gauge_without_tags():
metrics = PrometheusMetricsFactory()
gauge = metrics.create_gauge(name='jaeger:test_gauge_no_tags')
gauge(1)
after = REGISTRY.get_sample_value('jaeger:test_gauge_no_tags')
assert 1 == after
