def test_metric_limit_gauges(self, aggregator):
check = LimitedCheck()
assert check.get_warnings() == []
for i in range(0, 10):
check.gauge("metric", 0)
assert len(check.get_warnings()) == 0
assert len(aggregator.metrics("metric")) == 10
for i in range(0, 10):
check.gauge("metric", 0)
assert len(check.get_warnings()) == 1
assert len(aggregator.metrics("metric")) == 10
def test_metric_limit_instance_config(self, aggregator):
instances = [{
"max_returned_metrics": 42,
}]
check = AgentCheck("test", {}, instances)
assert check.get_warnings() == []
for i in range(0, 42):
check.gauge("metric", 0)
assert len(check.get_warnings()) == 0
assert len(aggregator.metrics("metric")) == 42
check.gauge("metric", 0)
assert len(check.get_warnings()) == 1
assert len(aggregator.metrics("metric")) == 42
def test_metric_limit_count(self, aggregator):
check = LimitedCheck()
assert check.get_warnings() == []
# Multiple calls for a single set of (metric_name, tags) should not trigger
for i in range(0, 20):
check.count("metric", 0, hostname="host-single")
assert len(check.get_warnings()) == 0
assert len(aggregator.metrics("metric")) == 20
# Multiple sets of tags should trigger
# Only 9 new sets of tags should pass through
for i in range(0, 20):
check.count("metric", 0, hostname="host-{}".format(i))
assert len(check.get_warnings()) == 1
assert len(aggregator.metrics("metric")) == 29
def test_metric_limit_instance_config_zero(self, aggregator):
instances = [{
"max_returned_metrics": 0,
}]
check = LimitedCheck("test", {}, instances)
assert len(check.get_warnings()) == 1
for i in range(0, 42):
check.gauge("metric", 0)
assert len(check.get_warnings()) == 1 # get_warnings resets the array
assert len(aggregator.metrics("metric")) == 10
def test_success(self, aggregator):
instance = INSTANCES['main']
c = Envoy(self.CHECK_NAME, None, {}, [instance])
c.check(instance)
metrics_collected = 0
for metric in METRICS.keys():
metrics_collected += len(aggregator.metrics(METRIC_PREFIX +
metric))
assert metrics_collected >= 250
def test_invalid_metric(aggregator, poll_mock):
"""
Testing that invalid values of metrics are discarded
"""
bad_metric_instance = {
'prometheus_url': 'http://localhost:10249/metrics',
'namespace': 'prometheus',
'metrics': [{
'metric1': 'renamed.metric1'
}, 'metric2', 'metric3'],
'send_histograms_buckets': True
}
c = PrometheusCheck('prometheus', None, {}, [bad_metric_instance])
c.check(bad_metric_instance)
assert aggregator.metrics('metric3') == []
def test_success(self, aggregator):
instance = self.INSTANCES['main']
c = Envoy(self.CHECK_NAME, None, {}, [instance])
with mock.patch('requests.get',
return_value=response('multiple_services')):
c.check(instance)
metrics_collected = 0
for metric in METRICS.keys():
metrics_collected += len(aggregator.metrics(METRIC_PREFIX +
metric))
num_metrics = len(
response('multiple_services').content.decode().splitlines())
num_metrics -= sum(c.unknown_metrics.values()) + sum(
c.unknown_tags.values())
assert 4150 <= metrics_collected == num_metrics
def test_check(aggregator, spin_up_elastic):
elastic_check = ESCheck(CHECK_NAME, {}, {})
default_tags = ["url:http://{0}:{1}".format(HOST, PORT)]
for _ in xrange(2):
try:
elastic_check.check(CONFIG)
except Exception:
time.sleep(1)
elastic_check.check(CONFIG)
expected_metrics = dict(ESCheck.STATS_METRICS)
ESCheck.CLUSTER_HEALTH_METRICS.update(ESCheck.CLUSTER_PENDING_TASKS)
expected_metrics.update(ESCheck.CLUSTER_HEALTH_METRICS)
instance = elastic_check.get_instance_config(CONFIG)
es_version = elastic_check._get_es_version(instance)
assert es_version == get_es_version()
if es_version < [5, 0, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_PRE_5_0_0)
if es_version >= [0, 90, 5]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_0_90_5)
if es_version >= [0, 90, 10]:
expected_metrics.update(ESCheck.JVM_METRICS_POST_0_90_10)
else:
expected_metrics.update(ESCheck.JVM_METRICS_PRE_0_90_10)
else:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_PRE_0_90_5)
expected_metrics.update(ESCheck.JVM_METRICS_PRE_0_90_10)
if es_version >= [1, 0, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_1_0_0)
if es_version < [2, 0, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_PRE_2_0)
if es_version >= [0, 90, 5]:
expected_metrics.update(
ESCheck.ADDITIONAL_METRICS_POST_0_90_5_PRE_2_0)
# Currently has issues in test framework
# if es_version >= [1, 0, 0]:
# expected_metrics.update(ESCheck.ADDITIONAL_METRICS_1_x)
if es_version >= [1, 3, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_1_3_0)
if es_version >= [1, 4, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_1_4_0)
if es_version >= [1, 5, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_1_5_0)
if es_version >= [1, 6, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_1_6_0)
if es_version >= [2, 0, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_2_0)
if es_version >= [2, 1, 0]:
expected_metrics.update(ESCheck.ADDITIONAL_METRICS_POST_2_1)
if os.environ.get("DD_ELASTIC_LOCAL_HOSTNAME"):
local_hostname = os.environ.get("DD_ELASTIC_LOCAL_HOSTNAME")
elif es_version < [2, 0, 0]:
local_hostname = socket.gethostname()
else:
local_hostname = '127.0.0.1'
contexts = [(CONF_HOSTNAME, default_tags + TAGS),
(local_hostname, default_tags + TAGS)]
stats_keys = (set(expected_metrics.keys()) -
set(ESCheck.CLUSTER_HEALTH_METRICS.keys()) -
set(ESCheck.CLUSTER_PENDING_TASKS.keys()))
# index_writer_max_memory metric was removed in v5
deprecated_metrics = [
'elasticsearch.indices.segments.index_writer_max_memory_in_bytes'
]
for m_name, desc in set(expected_metrics.iteritems()):
for hostname, m_tags in contexts:
m_tags = m_tags + CLUSTER_TAG
if (m_name in ESCheck.CLUSTER_HEALTH_METRICS
and hostname == local_hostname):
hostname = CONF_HOSTNAME
if m_name in stats_keys:
m_tags = m_tags + [u"node_name:batman"]
m_tags.sort()
if desc[0] == "gauge":
aggregator.metrics(m_name)
if es_version < [2, 0, 0]:
aggregator.assert_metric(m_name, tags=m_tags)
elif not (m_name in deprecated_metrics):
aggregator.assert_metric(m_name)
good_sc_tags = ['host:{0}'.format(HOST), 'port:{0}'.format(PORT)]
aggregator.assert_service_check(
'elasticsearch.can_connect',
status=ESCheck.OK, # OK
tags=good_sc_tags + TAGS)
# Assert service metadata
# self.assertServiceMetadata(['version'], count=3)
# FIXME: 0.90.13 returns randomly a red status instead of yellow,
# so we don't do a coverage test for it
# Remove me when we stop supporting 0.90.x (not supported anymore by ES)
if get_es_version() != [0, 90, 13]:
# Warning because elasticsearch status should be yellow, according to
# http://chrissimpson.co.uk/elasticsearch-yellow-cluster-status-explained.html
aggregator.assert_service_check('elasticsearch.cluster_health')
