def translate_to_collector(
metric_records: Sequence[MetricRecord], ) -> Sequence[metrics_pb2.Metric]:
collector_metrics = []
for metric_record in metric_records:
label_values = []
label_keys = []
for label_tuple in metric_record.labels:
label_keys.append(metrics_pb2.LabelKey(key=label_tuple[0]))
label_values.append(
metrics_pb2.LabelValue(has_value=label_tuple[1] is not None,
value=label_tuple[1]))
metric_descriptor = metrics_pb2.MetricDescriptor(
name=metric_record.metric.name,
description=metric_record.metric.description,
unit=metric_record.metric.unit,
type=get_collector_metric_type(metric_record.metric),
label_keys=label_keys,
)
timeseries = metrics_pb2.TimeSeries(
label_values=label_values,
points=[get_collector_point(metric_record)],
)
collector_metrics.append(
metrics_pb2.Metric(metric_descriptor=metric_descriptor,
timeseries=[timeseries]))
return collector_metrics
def _get_time_series_list_proto(series_list):
return [
metrics_pb2.TimeSeries(
start_timestamp=utils.proto_ts_from_datetime_str(
series.start_timestamp),
label_values=_get_label_values_proto(series.label_values),
points=_get_points_proto(series.points)) for series in series_list
]
def translate_to_collector(
export_records: Sequence[ExportRecord],
exporter_start_timestamp: Timestamp,
) -> Sequence[metrics_pb2.Metric]:
collector_metrics = []
for export_record in export_records:
label_values = []
label_keys = []
for label_tuple in export_record.labels:
label_keys.append(metrics_pb2.LabelKey(key=label_tuple[0]))
label_values.append(
metrics_pb2.LabelValue(
has_value=label_tuple[1] is not None,
value=str(label_tuple[1]),
)
)
metric_descriptor = metrics_pb2.MetricDescriptor(
name=export_record.instrument.name,
description=export_record.instrument.description,
unit=export_record.instrument.unit,
type=get_collector_metric_type(export_record.instrument),
label_keys=label_keys,
)
# If cumulative and stateful, explicitly set the start_timestamp to
# exporter start time.
if export_record.instrument.meter.processor.stateful:
start_timestamp = exporter_start_timestamp
else:
start_timestamp = None
timeseries = metrics_pb2.TimeSeries(
label_values=label_values,
points=[get_collector_point(export_record)],
start_timestamp=start_timestamp,
)
collector_metrics.append(
metrics_pb2.Metric(
metric_descriptor=metric_descriptor,
timeseries=[timeseries],
resource=get_resource(export_record),
)
)
return collector_metrics
def test_export_view_data(self):
v_data = view_data_module.ViewData(view=VIDEO_SIZE_VIEW,
start_time=TEST_TIME_STR,
end_time=TEST_TIME_STR)
v_data.record(context=tag_map_module.TagMap(), value=2, timestamp=None)
view_data = [v_data]
view_data = [metric_utils.view_data_to_metric(view_data[0], TEST_TIME)]
handler = mock.Mock(spec=ocagent.ExportRpcHandler)
ocagent.StatsExporter(handler).export_metrics(view_data)
self.assertEqual(
handler.send.call_args[0][0].metrics[0].metric_descriptor,
metrics_pb2.MetricDescriptor(
name=VIDEO_SIZE_VIEW_NAME,
description='processed video size over time',
unit='By',
type=metrics_pb2.MetricDescriptor.CUMULATIVE_DISTRIBUTION,
label_keys=[metrics_pb2.LabelKey(key=FRONTEND_KEY)]))
self.assertEqual(
handler.send.call_args[0][0].metrics[0].timeseries[0],
metrics_pb2.TimeSeries(
start_timestamp=timestamp_pb2.Timestamp(seconds=1545699723,
nanos=4000),
label_values=[metrics_pb2.LabelValue(has_value=False)],
points=[
metrics_pb2.Point(
timestamp=timestamp_pb2.Timestamp(seconds=1545699723,
nanos=4000),
distribution_value=metrics_pb2.DistributionValue(
sum=2,
count=1,
bucket_options=metrics_pb2.DistributionValue.
BucketOptions(
explicit=metrics_pb2.DistributionValue.
BucketOptions.Explicit(
bounds=[16.0 * MiB, 256.0 * MiB])),
buckets=[
metrics_pb2.DistributionValue.Bucket(count=1),
metrics_pb2.DistributionValue.Bucket(),
metrics_pb2.DistributionValue.Bucket(),
]))
]))