def test_counter(self): stats = MetricsAggregator('myhost') # Track some counters. stats.submit_packets('my.first.counter:1|c') stats.submit_packets('my.first.counter:5|c') stats.submit_packets('my.second.counter:1|c') stats.submit_packets('my.third.counter:3|c') # Ensure they roll up nicely. metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 3 first, second, third = metrics nt.assert_equals(first['metric'], 'my.first.counter') nt.assert_equals(first['points'][0][1], 6) nt.assert_equals(first['host'], 'myhost') nt.assert_equals(second['metric'], 'my.second.counter') nt.assert_equals(second['points'][0][1], 1) nt.assert_equals(third['metric'], 'my.third.counter') nt.assert_equals(third['points'][0][1], 3) # Ensure that counters reset to zero. metrics = self.sort_metrics(stats.flush()) first, second, third = metrics nt.assert_equals(first['metric'], 'my.first.counter') nt.assert_equals(first['points'][0][1], 0) nt.assert_equals(second['metric'], 'my.second.counter') nt.assert_equals(second['points'][0][1], 0) nt.assert_equals(third['metric'], 'my.third.counter') nt.assert_equals(third['points'][0][1], 0)
def test_dogstatsd_aggregation_perf(self): ma = MetricsAggregator('my.host') for _ in xrange(self.FLUSH_COUNT): for i in xrange(self.LOOPS_PER_FLUSH): for j in xrange(self.METRIC_COUNT): # metrics ma.submit_packets('counter.%s:%s|c' % (j, i)) ma.submit_packets('gauge.%s:%s|g' % (j, i)) ma.submit_packets('histogram.%s:%s|h' % (j, i)) ma.submit_packets('set.%s:%s|s' % (j, 1.0)) # tagged metrics ma.submit_packets('counter.%s:%s|c|#tag1,tag2' % (j, i)) ma.submit_packets('gauge.%s:%s|g|#tag1,tag2' % (j, i)) ma.submit_packets('histogram.%s:%s|h|#tag1,tag2' % (j, i)) ma.submit_packets('set.%s:%s|s|#tag1,tag2' % (j, i)) # sampled metrics ma.submit_packets('counter.%s:%s|c|@0.5' % (j, i)) ma.submit_packets('gauge.%s:%s|g|@0.5' % (j, i)) ma.submit_packets('histogram.%s:%s|h|@0.5' % (j, i)) ma.submit_packets('set.%s:%s|s|@0.5' % (j, i)) ma.flush()
def test_gauge(self): stats = MetricsAggregator('myhost') # Track some counters. stats.submit_packets('my.first.gauge:1|g') stats.submit_packets('my.first.gauge:5|g') stats.submit_packets('my.second.gauge:1.5|g') # Ensure that gauges roll up correctly. metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 2 first, second = metrics nt.assert_equals(first['metric'], 'my.first.gauge') nt.assert_equals(first['points'][0][1], 5) nt.assert_equals(first['host'], 'myhost') nt.assert_equals(second['metric'], 'my.second.gauge') nt.assert_equals(second['points'][0][1], 1.5) # Ensure that old gauges get dropped due to old timestamps stats.gauge('my.first.gauge', 5) stats.gauge('my.first.gauge', 1, timestamp=1000000000) stats.gauge('my.second.gauge', 20, timestamp=1000000000) metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 1 first = metrics[0] nt.assert_equals(first['metric'], 'my.first.gauge') nt.assert_equals(first['points'][0][1], 5) nt.assert_equals(first['host'], 'myhost')
def test_gauge(self): stats = MetricsAggregator('myhost') # Track some counters. stats.submit_packets('my.first.gauge:1|g') stats.submit_packets('my.first.gauge:5|g') stats.submit_packets('my.second.gauge:1.5|g') # Ensure that gauges roll up correctly. metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 2 first, second = metrics nt.assert_equals(first['metric'], 'my.first.gauge') nt.assert_equals(first['points'][0][1], 5) nt.assert_equals(first['host'], 'myhost') nt.assert_equals(second['metric'], 'my.second.gauge') nt.assert_equals(second['points'][0][1], 1.5) # Ensure that old gauges get dropped due to old timestamps stats.gauge('my.first.gauge', 5) stats.gauge('my.first.gauge', 1, timestamp=1000000000) stats.gauge('my.second.gauge', 20, timestamp=1000000000) metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 1 first = metrics[0] nt.assert_equals(first['metric'], 'my.first.gauge') nt.assert_equals(first['points'][0][1], 5) nt.assert_equals(first['host'], 'myhost')
def test_histogram(self): stats = MetricsAggregator('myhost') # Sample all numbers between 1-100 many times. This # means our percentiles should be relatively close to themselves. percentiles = range(100) random.shuffle(percentiles) # in place for i in percentiles: for j in xrange(20): for type_ in ['h', 'ms']: m = 'my.p:%s|%s' % (i, type_) stats.submit_packets(m) metrics = self.sort_metrics(stats.flush()) def assert_almost_equal(i, j, e=1): # Floating point math? assert abs(i - j) <= e, "%s %s %s" % (i, j, e) nt.assert_equal(len(metrics), 5) p95, pavg, pcount, pmax, pmed = self.sort_metrics(metrics) nt.assert_equal(p95['metric'], 'my.p.95percentile') assert_almost_equal(p95['points'][0][1], 95, 10) assert_almost_equal(pmax['points'][0][1], 99, 1) assert_almost_equal(pmed['points'][0][1], 50, 2) assert_almost_equal(pavg['points'][0][1], 50, 2) assert_almost_equal(pcount['points'][0][1], 4000, 0) # 100 * 20 * 2 nt.assert_equals(p95['host'], 'myhost') # Ensure that histograms are reset. metrics = self.sort_metrics(stats.flush()) assert not metrics
def test_histogram(self): stats = MetricsAggregator('myhost') # Sample all numbers between 1-100 many times. This # means our percentiles should be relatively close to themselves. percentiles = range(100) random.shuffle(percentiles) # in place for i in percentiles: for j in xrange(20): for type_ in ['h', 'ms']: m = 'my.p:%s|%s' % (i, type_) stats.submit_packets(m) metrics = self.sort_metrics(stats.flush()) def assert_almost_equal(i, j, e=1): # Floating point math? assert abs(i - j) <= e, "%s %s %s" % (i, j, e) nt.assert_equal(len(metrics), 5) p95, pavg, pcount, pmax, pmed = self.sort_metrics(metrics) nt.assert_equal(p95['metric'], 'my.p.95percentile') assert_almost_equal(p95['points'][0][1], 95, 10) assert_almost_equal(pmax['points'][0][1], 99, 1) assert_almost_equal(pmed['points'][0][1], 50, 2) assert_almost_equal(pavg['points'][0][1], 50, 2) assert_almost_equal(pcount['points'][0][1], 4000, 0) # 100 * 20 * 2 nt.assert_equals(p95['host'], 'myhost') # Ensure that histograms are reset. metrics = self.sort_metrics(stats.flush()) assert not metrics
def test_counter(self): stats = MetricsAggregator('myhost') # Track some counters. stats.submit_packets('my.first.counter:1|c') stats.submit_packets('my.first.counter:5|c') stats.submit_packets('my.second.counter:1|c') stats.submit_packets('my.third.counter:3|c') # Ensure they roll up nicely. metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 3 first, second, third = metrics nt.assert_equals(first['metric'], 'my.first.counter') nt.assert_equals(first['points'][0][1], 6) nt.assert_equals(first['host'], 'myhost') nt.assert_equals(second['metric'], 'my.second.counter') nt.assert_equals(second['points'][0][1], 1) nt.assert_equals(third['metric'], 'my.third.counter') nt.assert_equals(third['points'][0][1], 3) # Ensure that counters reset to zero. metrics = self.sort_metrics(stats.flush()) first, second, third = metrics nt.assert_equals(first['metric'], 'my.first.counter') nt.assert_equals(first['points'][0][1], 0) nt.assert_equals(second['metric'], 'my.second.counter') nt.assert_equals(second['points'][0][1], 0) nt.assert_equals(third['metric'], 'my.third.counter') nt.assert_equals(third['points'][0][1], 0)
def test_checksd_aggregation_perf(self): ma = MetricsAggregator('my.host') for _ in xrange(self.FLUSH_COUNT): for i in xrange(self.LOOPS_PER_FLUSH): # Counters for j in xrange(self.METRIC_COUNT): ma.increment('counter.%s' % j, i) ma.gauge('gauge.%s' % j, i) ma.histogram('histogram.%s' % j, i) ma.set('set.%s' % j, float(i)) ma.flush()
def test_rate(self): stats = MetricsAggregator('myhost') stats.submit_packets('my.rate:10|_dd-r') # Sleep 1 second so the time interval > 0 time.sleep(1) stats.submit_packets('my.rate:40|_dd-r') # Check that the rate is calculated correctly metrics = stats.flush() nt.assert_equal(len(metrics), 1) m = metrics[0] nt.assert_equals(m['metric'], 'my.rate') nt.assert_equals(m['points'][0][1], 30) # Assert that no more rates are given assert not stats.flush()
def test_tags(self): stats = MetricsAggregator('myhost') stats.submit_packets('gauge:1|c') stats.submit_packets('gauge:2|c|@1') stats.submit_packets('gauge:4|c|#tag1,tag2') stats.submit_packets( 'gauge:8|c|#tag2,tag1') # Should be the same as above stats.submit_packets('gauge:16|c|#tag3,tag4') metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 3 first, second, third = metrics nt.assert_equal(first['metric'], 'gauge') nt.assert_equal(first['dimensions'], None) nt.assert_equal(first['points'][0][1], 3) nt.assert_equal(first['host'], 'myhost') nt.assert_equal(second['metric'], 'gauge') nt.assert_equal(second['dimensions'], ('tag1', 'tag2')) nt.assert_equal(second['points'][0][1], 12) nt.assert_equal(second['host'], 'myhost') nt.assert_equal(third['metric'], 'gauge') nt.assert_equal(third['dimensions'], ('tag3', 'tag4')) nt.assert_equal(third['points'][0][1], 16) nt.assert_equal(third['host'], 'myhost')
def test_tags(self): stats = MetricsAggregator('myhost') stats.submit_packets('gauge:1|c') stats.submit_packets('gauge:2|c|@1') stats.submit_packets('gauge:4|c|#tag1,tag2') stats.submit_packets('gauge:8|c|#tag2,tag1') # Should be the same as above stats.submit_packets('gauge:16|c|#tag3,tag4') metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 3 first, second, third = metrics nt.assert_equal(first['metric'], 'gauge') nt.assert_equal(first['dimensions'], None) nt.assert_equal(first['points'][0][1], 3) nt.assert_equal(first['host'], 'myhost') nt.assert_equal(second['metric'], 'gauge') nt.assert_equal(second['dimensions'], ('tag1', 'tag2')) nt.assert_equal(second['points'][0][1], 12) nt.assert_equal(second['host'], 'myhost') nt.assert_equal(third['metric'], 'gauge') nt.assert_equal(third['dimensions'], ('tag3', 'tag4')) nt.assert_equal(third['points'][0][1], 16) nt.assert_equal(third['host'], 'myhost')
def test_rate(self): stats = MetricsAggregator('myhost') stats.submit_packets('my.rate:10|_dd-r') # Sleep 1 second so the time interval > 0 time.sleep(1) stats.submit_packets('my.rate:40|_dd-r') # Check that the rate is calculated correctly metrics = stats.flush() nt.assert_equal(len(metrics), 1) m = metrics[0] nt.assert_equals(m['metric'], 'my.rate') nt.assert_equals(m['points'][0][1], 30) # Assert that no more rates are given assert not stats.flush()
def test_string_sets(self): stats = MetricsAggregator('myhost') stats.submit_packets('my.set:string|s') stats.submit_packets('my.set:sets|s') stats.submit_packets('my.set:sets|s') stats.submit_packets('my.set:test|s') stats.submit_packets('my.set:test|s') stats.submit_packets('my.set:test|s') # Assert that it's treated normally. metrics = stats.flush() nt.assert_equal(len(metrics), 1) m = metrics[0] nt.assert_equal(m['metric'], 'my.set') nt.assert_equal(m['points'][0][1], 3) # Assert there are no more sets assert not stats.flush()
def test_scientific_notation(self): stats = MetricsAggregator('myhost', interval=10) stats.submit_packets('test.scinot:9.512901e-05|g') metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 1 ts, val = metrics[0].get('points')[0] nt.assert_almost_equal(val, 9.512901e-05)
def test_scientific_notation(self): stats = MetricsAggregator('myhost', interval=10) stats.submit_packets('test.scinot:9.512901e-05|g') metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 1 ts, val = metrics[0].get('points')[0] nt.assert_almost_equal(val, 9.512901e-05)
def test_string_sets(self): stats = MetricsAggregator('myhost') stats.submit_packets('my.set:string|s') stats.submit_packets('my.set:sets|s') stats.submit_packets('my.set:sets|s') stats.submit_packets('my.set:test|s') stats.submit_packets('my.set:test|s') stats.submit_packets('my.set:test|s') # Assert that it's treated normally. metrics = stats.flush() nt.assert_equal(len(metrics), 1) m = metrics[0] nt.assert_equal(m['metric'], 'my.set') nt.assert_equal(m['points'][0][1], 3) # Assert there are no more sets assert not stats.flush()
def test_sampled_counter(self): # Submit a sampled counter. stats = MetricsAggregator('myhost') stats.submit_packets('sampled.counter:1|c|@0.5') metrics = stats.flush() assert len(metrics) == 1 m = metrics[0] assert m['metric'] == 'sampled.counter' nt.assert_equal(m['points'][0][1], 2)
def test_sampled_counter(self): # Submit a sampled counter. stats = MetricsAggregator('myhost') stats.submit_packets('sampled.counter:1|c|@0.5') metrics = stats.flush() assert len(metrics) == 1 m = metrics[0] assert m['metric'] == 'sampled.counter' nt.assert_equal(m['points'][0][1], 2)
def test_metrics_expiry(self): # Ensure metrics eventually expire and stop submitting. stats = MetricsAggregator('myhost', expiry_seconds=1) stats.submit_packets('test.counter:123|c') # Ensure points keep submitting assert stats.flush() assert stats.flush() time.sleep(0.5) assert stats.flush() # Now sleep for longer than the expiry window and ensure # no points are submitted time.sleep(2) m = stats.flush() assert not m, str(m) # If we submit again, we're all good. stats.submit_packets('test.counter:123|c') assert stats.flush()
def test_metrics_expiry(self): # Ensure metrics eventually expire and stop submitting. stats = MetricsAggregator('myhost', expiry_seconds=1) stats.submit_packets('test.counter:123|c') # Ensure points keep submitting assert stats.flush() assert stats.flush() time.sleep(0.5) assert stats.flush() # Now sleep for longer than the expiry window and ensure # no points are submitted time.sleep(2) m = stats.flush() assert not m, str(m) # If we submit again, we're all good. stats.submit_packets('test.counter:123|c') assert stats.flush()
def test_sampled_histogram(self): # Submit a sampled histogram. stats = MetricsAggregator('myhost') stats.submit_packets('sampled.hist:5|h|@0.5') # Assert we scale up properly. metrics = self.sort_metrics(stats.flush()) p95, pavg, pcount, pmax, pmed = self.sort_metrics(metrics) nt.assert_equal(pcount['points'][0][1], 2) for p in [p95, pavg, pmed, pmax]: nt.assert_equal(p['points'][0][1], 5)
def test_histogram_normalization(self): stats = MetricsAggregator('myhost', interval=10) for i in range(5): stats.submit_packets('h1:1|h') for i in range(20): stats.submit_packets('h2:1|h') metrics = self.sort_metrics(stats.flush()) _, _, h1count, _, _, _, _, h2count, _, _ = metrics nt.assert_equal(h1count['points'][0][1], 0.5) nt.assert_equal(h2count['points'][0][1], 2)
def test_gauge_sample_rate(self): stats = MetricsAggregator('myhost') # Submit a sampled gauge metric. stats.submit_packets('sampled.gauge:10|g|@0.1') # Assert that it's treated normally. metrics = stats.flush() nt.assert_equal(len(metrics), 1) m = metrics[0] nt.assert_equal(m['metric'], 'sampled.gauge') nt.assert_equal(m['points'][0][1], 10)
def test_sampled_histogram(self): # Submit a sampled histogram. stats = MetricsAggregator('myhost') stats.submit_packets('sampled.hist:5|h|@0.5') # Assert we scale up properly. metrics = self.sort_metrics(stats.flush()) p95, pavg, pcount, pmax, pmed = self.sort_metrics(metrics) nt.assert_equal(pcount['points'][0][1], 2) for p in [p95, pavg, pmed, pmax]: nt.assert_equal(p['points'][0][1], 5)
def test_batch_submission(self): # Submit a sampled histogram. stats = MetricsAggregator('myhost') metrics = ['counter:1|c', 'counter:1|c', 'gauge:1|g'] packet = "\n".join(metrics) stats.submit_packets(packet) metrics = self.sort_metrics(stats.flush()) nt.assert_equal(2, len(metrics)) counter, gauge = metrics assert counter['points'][0][1] == 2 assert gauge['points'][0][1] == 1
def test_gauge_sample_rate(self): stats = MetricsAggregator('myhost') # Submit a sampled gauge metric. stats.submit_packets('sampled.gauge:10|g|@0.1') # Assert that it's treated normally. metrics = stats.flush() nt.assert_equal(len(metrics), 1) m = metrics[0] nt.assert_equal(m['metric'], 'sampled.gauge') nt.assert_equal(m['points'][0][1], 10)
def test_histogram_normalization(self): stats = MetricsAggregator('myhost', interval=10) for i in range(5): stats.submit_packets('h1:1|h') for i in range(20): stats.submit_packets('h2:1|h') metrics = self.sort_metrics(stats.flush()) _, _, h1count, _, _, _, _, h2count, _, _ = metrics nt.assert_equal(h1count['points'][0][1], 0.5) nt.assert_equal(h2count['points'][0][1], 2)
def test_batch_submission(self): # Submit a sampled histogram. stats = MetricsAggregator('myhost') metrics = [ 'counter:1|c', 'counter:1|c', 'gauge:1|g' ] packet = "\n".join(metrics) stats.submit_packets(packet) metrics = self.sort_metrics(stats.flush()) nt.assert_equal(2, len(metrics)) counter, gauge = metrics assert counter['points'][0][1] == 2 assert gauge['points'][0][1] == 1
def test_histogram_counter(self): # Test whether histogram.count == increment # same deal with a sample rate cnt = 100000 for run in [1, 2]: stats = MetricsAggregator('myhost') for i in xrange(cnt): if run == 2: stats.submit_packets('test.counter:1|c|@0.5') stats.submit_packets('test.hist:1|ms|@0.5') else: stats.submit_packets('test.counter:1|c') stats.submit_packets('test.hist:1|ms') metrics = self.sort_metrics(stats.flush()) assert len(metrics) > 0 nt.assert_equal([m['points'][0][1] for m in metrics if m['metric'] == 'test.counter'], [cnt * run]) nt.assert_equal([m['points'][0][1] for m in metrics if m['metric'] == 'test.hist.count'], [cnt * run])
def test_counter_normalization(self): stats = MetricsAggregator('myhost', interval=10) # Assert counters are normalized. stats.submit_packets('int:1|c') stats.submit_packets('int:4|c') stats.submit_packets('int:15|c') stats.submit_packets('float:5|c') metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 2 floatc, intc = metrics nt.assert_equal(floatc['metric'], 'float') nt.assert_equal(floatc['points'][0][1], 0.5) nt.assert_equal(floatc['host'], 'myhost') nt.assert_equal(intc['metric'], 'int') nt.assert_equal(intc['points'][0][1], 2) nt.assert_equal(intc['host'], 'myhost')
def test_counter_normalization(self): stats = MetricsAggregator('myhost', interval=10) # Assert counters are normalized. stats.submit_packets('int:1|c') stats.submit_packets('int:4|c') stats.submit_packets('int:15|c') stats.submit_packets('float:5|c') metrics = self.sort_metrics(stats.flush()) assert len(metrics) == 2 floatc, intc = metrics nt.assert_equal(floatc['metric'], 'float') nt.assert_equal(floatc['points'][0][1], 0.5) nt.assert_equal(floatc['host'], 'myhost') nt.assert_equal(intc['metric'], 'int') nt.assert_equal(intc['points'][0][1], 2) nt.assert_equal(intc['host'], 'myhost')
def test_histogram_counter(self): # Test whether histogram.count == increment # same deal with a sample rate cnt = 100000 for run in [1, 2]: stats = MetricsAggregator('myhost') for i in xrange(cnt): if run == 2: stats.submit_packets('test.counter:1|c|@0.5') stats.submit_packets('test.hist:1|ms|@0.5') else: stats.submit_packets('test.counter:1|c') stats.submit_packets('test.hist:1|ms') metrics = self.sort_metrics(stats.flush()) assert len(metrics) > 0 nt.assert_equal([ m['points'][0][1] for m in metrics if m['metric'] == 'test.counter' ], [cnt * run]) nt.assert_equal([ m['points'][0][1] for m in metrics if m['metric'] == 'test.hist.count' ], [cnt * run])
class AgentCheck(object): def __init__(self, name, init_config, agent_config, instances=None): """Initialize a new check. :param name: The name of the check :param init_config: The config for initializing the check :param agent_config: The global configuration for the agent :param instances: A list of configuration objects for each instance. """ self.name = name self.init_config = init_config self.agent_config = agent_config self.hostname = monagent.common.util.get_hostname(agent_config) self.log = logging.getLogger('%s.%s' % (__name__, name)) self.aggregator = MetricsAggregator( self.hostname, recent_point_threshold=agent_config.get('recent_point_threshold', None)) self.events = [] self.instances = instances or [] self.warnings = [] self.library_versions = None def instance_count(self): """Return the number of instances that are configured for this check. """ return len(self.instances) def gauge(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None, timestamp=None): """Record the value of a gauge, with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value of the gauge :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric :param timestamp: (optional) The timestamp for this metric value """ self.aggregator.gauge(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name, timestamp) def increment(self, metric, value=1, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Increment a counter with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value to increment by :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.increment(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def decrement(self, metric, value=-1, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Decrement a counter with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value to decrement by :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.decrement(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def rate(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Submit a point for a metric that will be calculated as a rate on flush. Values will persist across each call to `check` if there is not enough point to generate a rate on the flush. :param metric: The name of the metric :param value: The value of the rate :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.rate(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def histogram(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Sample a histogram value, with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value to sample for the histogram :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.histogram(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def set(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Sample a set value, with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value for the set :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.set(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def _set_dimensions(self, dimensions): new_dimensions = { 'component': 'monasca-agent', 'service': 'monitoring' } if dimensions is not None: new_dimensions.update(dimensions.copy()) return new_dimensions def event(self, event): """Save an event. :param event: The event payload as a dictionary. Has the following structure: { "timestamp": int, the epoch timestamp for the event, "event_type": string, the event time name, "api_key": string, the api key of the account to associate the event with, "msg_title": string, the title of the event, "msg_text": string, the text body of the event, "alert_type": (optional) string, one of ('error', 'warning', 'success', 'info'). Defaults to 'info'. "source_type_name": (optional) string, the source type name, "host": (optional) string, the name of the host, "dimensions": (optional) a dictionary of dimensions to associate with this event } """ if event.get('api_key') is None: event['api_key'] = self.agent_config['api_key'] self.events.append(event) def has_events(self): """Check whether the check has saved any events @return whether or not the check has saved any events @rtype boolean """ return len(self.events) > 0 def get_metrics(self, prettyprint=False): """Get all metrics, including the ones that are tagged. @return the list of samples @rtype list of Measurement objects from monagent.common.metrics """ if prettyprint: metrics = self.aggregator.flush() for metric in metrics: print(" Timestamp: {}".format(metric.timestamp)) print(" Name: {}".format(metric.name)) print(" Value: {}".format(metric.value)) if (metric.delegated_tenant): print(" Delegtd ID: {}".format(metric.delegated_tenant)) print(" Dimensions: ", end='') line = 0 for name in metric.dimensions: if line != 0: print(" " * 13, end='') print("{0}={1}".format(name, metric.dimensions[name])) line += 1 print("-" * 24) return self.aggregator.flush() def get_events(self): """Return a list of the events saved by the check, if any @return the list of events saved by this check @rtype list of event dictionaries """ events = self.events self.events = [] return events def has_warnings(self): """Check whether the instance run created any warnings. """ return len(self.warnings) > 0 def warning(self, warning_message): """Add a warning message that will be printed in the info page :param warning_message: String. Warning message to be displayed """ self.warnings.append(warning_message) def get_library_info(self): if self.library_versions is not None: return self.library_versions try: self.library_versions = self.get_library_versions() except NotImplementedError: pass def get_library_versions(self): """Should return a string that shows which version of the needed libraries are used """ raise NotImplementedError def get_warnings(self): """Return the list of warnings messages to be displayed in the info page. """ warnings = self.warnings self.warnings = [] return warnings def run(self): """Run all instances. """ instance_statuses = [] for i, instance in enumerate(self.instances): try: self.check(instance) if self.has_warnings(): instance_status = monagent.common.check_status.InstanceStatus( i, monagent.common.check_status.STATUS_WARNING, warnings=self.get_warnings()) else: instance_status = monagent.common.check_status.InstanceStatus( i, monagent.common.check_status.STATUS_OK) except Exception as e: self.log.exception("Check '%s' instance #%s failed" % (self.name, i)) instance_status = monagent.common.check_status.InstanceStatus( i, monagent.common.check_status.STATUS_ERROR, error=e, tb=traceback.format_exc()) instance_statuses.append(instance_status) return instance_statuses def check(self, instance): """Overriden by the check class. This will be called to run the check. :param instance: A dict with the instance information. This will vary depending on your config structure. """ raise NotImplementedError() @staticmethod def stop(): """To be executed when the agent is being stopped to clean ressources. """ pass @classmethod def from_yaml(cls, path_to_yaml=None, agentConfig=None, yaml_text=None, check_name=None): """A method used for testing your check without running the agent. """ if hasattr(yaml, 'CLoader'): Loader = yaml.CLoader else: Loader = yaml.Loader if path_to_yaml: check_name = os.path.basename(path_to_yaml).split('.')[0] try: f = open(path_to_yaml) except IOError: raise Exception('Unable to open yaml config: %s' % path_to_yaml) yaml_text = f.read() f.close() config = yaml.load(yaml_text, Loader=Loader) check = cls(check_name, config.get('init_config') or {}, agentConfig or {}) return check, config.get('instances', []) @staticmethod def normalize(metric, prefix=None): """Turn a metric into a well-formed metric name prefix.b.c :param metric The metric name to normalize :param prefix A prefix to to add to the normalized name, default None """ name = re.sub(r"[,\+\*\-/()\[\]{}]", "_", metric) # Eliminate multiple _ name = re.sub(r"__+", "_", name) # Don't start/end with _ name = re.sub(r"^_", "", name) name = re.sub(r"_$", "", name) # Drop ._ and _. name = re.sub(r"\._", ".", name) name = re.sub(r"_\.", ".", name) if prefix is not None: return prefix + "." + name else: return name @staticmethod def read_config(instance, key, message=None, cast=None, optional=False): val = instance.get(key) if val is None: if optional is False: message = message or 'Must provide `%s` value in instance config' % key raise Exception(message) else: return val if cast is None: return val else: return cast(val)
class AgentCheck(object): def __init__(self, name, init_config, agent_config, instances=None): """Initialize a new check. :param name: The name of the check :param init_config: The config for initializing the check :param agent_config: The global configuration for the agent :param instances: A list of configuration objects for each instance. """ self.name = name self.init_config = init_config self.agent_config = agent_config self.hostname = monagent.common.util.get_hostname(agent_config) self.log = logging.getLogger('%s.%s' % (__name__, name)) self.aggregator = MetricsAggregator(self.hostname, recent_point_threshold=agent_config.get('recent_point_threshold', None)) self.events = [] self.instances = instances or [] self.warnings = [] self.library_versions = None def instance_count(self): """Return the number of instances that are configured for this check. """ return len(self.instances) def gauge(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None, timestamp=None): """Record the value of a gauge, with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value of the gauge :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric :param timestamp: (optional) The timestamp for this metric value """ self.aggregator.gauge(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name, timestamp) def increment(self, metric, value=1, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Increment a counter with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value to increment by :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.increment(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def decrement(self, metric, value=-1, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Decrement a counter with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value to decrement by :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.decrement(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def rate(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Submit a point for a metric that will be calculated as a rate on flush. Values will persist across each call to `check` if there is not enough point to generate a rate on the flush. :param metric: The name of the metric :param value: The value of the rate :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.rate(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def histogram(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Sample a histogram value, with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value to sample for the histogram :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.histogram(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def set(self, metric, value, dimensions=None, delegated_tenant=None, hostname=None, device_name=None): """Sample a set value, with optional dimensions, hostname and device name. :param metric: The name of the metric :param value: The value for the set :param dimensions: (optional) A dictionary of dimensions for this metric :param delegated_tenant: (optional) Submit metrics on behalf of this tenant ID. :param hostname: (optional) A hostname for this metric. Defaults to the current hostname. :param device_name: (optional) The device name for this metric """ self.aggregator.set(metric, value, self._set_dimensions(dimensions), delegated_tenant, hostname, device_name) def _set_dimensions(self, dimensions): new_dimensions = {'component': 'monasca-agent', 'service': 'monitoring'} if dimensions is not None: new_dimensions.update(dimensions.copy()) return new_dimensions def event(self, event): """Save an event. :param event: The event payload as a dictionary. Has the following structure: { "timestamp": int, the epoch timestamp for the event, "event_type": string, the event time name, "api_key": string, the api key of the account to associate the event with, "msg_title": string, the title of the event, "msg_text": string, the text body of the event, "alert_type": (optional) string, one of ('error', 'warning', 'success', 'info'). Defaults to 'info'. "source_type_name": (optional) string, the source type name, "host": (optional) string, the name of the host, "dimensions": (optional) a dictionary of dimensions to associate with this event } """ if event.get('api_key') is None: event['api_key'] = self.agent_config['api_key'] self.events.append(event) def has_events(self): """Check whether the check has saved any events @return whether or not the check has saved any events @rtype boolean """ return len(self.events) > 0 def get_metrics(self, prettyprint=False): """Get all metrics, including the ones that are tagged. @return the list of samples @rtype list of Measurement objects from monagent.common.metrics """ if prettyprint: metrics = self.aggregator.flush() for metric in metrics: print(" Timestamp: {}".format(metric.timestamp)) print(" Name: {}".format(metric.name)) print(" Value: {}".format(metric.value)) if (metric.delegated_tenant): print(" Delegtd ID: {}".format(metric.delegated_tenant)) print(" Dimensions: ", end='') line = 0 for name in metric.dimensions: if line != 0: print(" " * 13, end='') print("{0}={1}".format(name, metric.dimensions[name])) line += 1 print("-" * 24) return self.aggregator.flush() def get_events(self): """Return a list of the events saved by the check, if any @return the list of events saved by this check @rtype list of event dictionaries """ events = self.events self.events = [] return events def has_warnings(self): """Check whether the instance run created any warnings. """ return len(self.warnings) > 0 def warning(self, warning_message): """Add a warning message that will be printed in the info page :param warning_message: String. Warning message to be displayed """ self.warnings.append(warning_message) def get_library_info(self): if self.library_versions is not None: return self.library_versions try: self.library_versions = self.get_library_versions() except NotImplementedError: pass def get_library_versions(self): """Should return a string that shows which version of the needed libraries are used """ raise NotImplementedError def get_warnings(self): """Return the list of warnings messages to be displayed in the info page. """ warnings = self.warnings self.warnings = [] return warnings def run(self): """Run all instances. """ instance_statuses = [] for i, instance in enumerate(self.instances): try: self.check(instance) if self.has_warnings(): instance_status = monagent.common.check_status.InstanceStatus(i, monagent.common.check_status.STATUS_WARNING, warnings=self.get_warnings()) else: instance_status = monagent.common.check_status.InstanceStatus(i, monagent.common.check_status.STATUS_OK) except Exception as e: self.log.exception("Check '%s' instance #%s failed" % (self.name, i)) instance_status = monagent.common.check_status.InstanceStatus(i, monagent.common.check_status.STATUS_ERROR, error=e, tb=traceback.format_exc()) instance_statuses.append(instance_status) return instance_statuses def check(self, instance): """Overriden by the check class. This will be called to run the check. :param instance: A dict with the instance information. This will vary depending on your config structure. """ raise NotImplementedError() @staticmethod def stop(): """To be executed when the agent is being stopped to clean ressources. """ pass @classmethod def from_yaml(cls, path_to_yaml=None, agentConfig=None, yaml_text=None, check_name=None): """A method used for testing your check without running the agent. """ if hasattr(yaml, 'CLoader'): Loader = yaml.CLoader else: Loader = yaml.Loader if path_to_yaml: check_name = os.path.basename(path_to_yaml).split('.')[0] try: f = open(path_to_yaml) except IOError: raise Exception('Unable to open yaml config: %s' % path_to_yaml) yaml_text = f.read() f.close() config = yaml.load(yaml_text, Loader=Loader) check = cls(check_name, config.get('init_config') or {}, agentConfig or {}) return check, config.get('instances', []) @staticmethod def normalize(metric, prefix=None): """Turn a metric into a well-formed metric name prefix.b.c :param metric The metric name to normalize :param prefix A prefix to to add to the normalized name, default None """ name = re.sub(r"[,\+\*\-/()\[\]{}]", "_", metric) # Eliminate multiple _ name = re.sub(r"__+", "_", name) # Don't start/end with _ name = re.sub(r"^_", "", name) name = re.sub(r"_$", "", name) # Drop ._ and _. name = re.sub(r"\._", ".", name) name = re.sub(r"_\.", ".", name) if prefix is not None: return prefix + "." + name else: return name @staticmethod def read_config(instance, key, message=None, cast=None, optional=False): val = instance.get(key) if val is None: if optional is False: message = message or 'Must provide `%s` value in instance config' % key raise Exception(message) else: return val if cast is None: return val else: return cast(val)