def _extract_header(time_series):
"""Return a copy of time_series with the points removed."""
return TimeSeries(
metric=time_series.metric,
resource=time_series.resource,
metric_kind=time_series.metric_kind,
value_type=time_series.value_type,
)
def _query_iter_get_result():
METRIC_LABELS = list({'instance_name': name} for name in INSTANCE_NAMES)
RESOURCE_LABELS = list({
'project_id': PROJECT,
'zone': zone,
'instance_id': instance_id,
} for zone, instance_id in zip(INSTANCE_ZONES, INSTANCE_IDS))
for metric_labels, resource_labels in zip(METRIC_LABELS, RESOURCE_LABELS):
yield TimeSeries(
metric=Metric(type=METRIC_TYPE, labels=metric_labels),
resource=MonitoredResource(type=RESOURCE_TYPE, labels=resource_labels),
metric_kind='GAUGE',
value_type='DOUBLE',
points=[],
)
def _format_metric(self, key, value, unit, buffer=False, **dimensions):
series = TimeSeries()
series.metrics.type = 'custom.googleapis.com/custodian/policy/%s' % key
# Google controlled vocabulary with artificial limitations on resource type
# there's not uch useful we can utilize.
series.resource.type = 'global'
# series.resource.labels['project_id'] =
point = series.points.add()
if unit == 'Count':
point.value.int64 = value
elif unit == 'Seconds':
point.value.double_value = value
now = time.time()
point.interval.end_time.seconds = ns = int(now)
point.interval.end_time.nanos = int((now - ns) * 10**9)
return series
def get_time_series(metric_descriptor, value): series = TimeSeries() series.metric.type = metric_descriptor.type labels = series.metric.labels labels['workflow_id'] = WORKFLOW_ID labels['task_call_name'] = TASK_CALL_NAME labels['task_call_index'] = TASK_CALL_INDEX labels['task_call_attempt'] = TASK_CALL_ATTEMPT labels['cpu_count'] = CPU_COUNT_LABEL labels['mem_size'] = MEMORY_SIZE_LABEL labels['disk_size'] = DISK_SIZE_LABEL series.resource.type = 'gce_instance' series.resource.labels['zone'] = ZONE series.resource.labels['instance_id'] = INSTANCE point = series.points.add(value=value) point.interval.end_time.seconds = int(time()) return series
def get_time_series(metric_descriptor, value):
global last_time
series = TimeSeries()
series.metric.type = metric_descriptor.type
labels = series.metric.labels
labels['workflow_id'] = WORKFLOW_ID
labels['task_call_name'] = TASK_CALL_NAME
labels['task_call_index'] = TASK_CALL_INDEX
labels['task_call_attempt'] = TASK_CALL_ATTEMPT
labels['cpu_count'] = CPU_COUNT_LABEL
labels['mem_size'] = MEMORY_SIZE_LABEL
labels['disk_size'] = DISK_SIZE_LABEL
labels['preemptible'] = PREEMPTIBLE_LABEL
series.resource.type = 'gce_instance'
series.resource.labels['zone'] = MACHINE['zone']
series.resource.labels['instance_id'] = MACHINE['name']
point = series.points.add(value=value)
end_time = max(time(), last_time + REPORT_TIME_SEC_MIN)
point.interval.end_time.seconds = int(end_time)
return series
def deserialize(cls, data: str) -> "Result":
ts = TimeSeries.from_json(data)
return cls(ts=ts)
def serialize(obj) -> str:
return json.dumps(TimeSeries.to_dict(obj.ts))
def deserialize(cls, data: str) -> "GCPResult":
b = base64.b64decode(data.encode("utf-8"))
ts = TimeSeries.deserialize(b)
return cls(ts=ts)
def serialize(obj):
b = TimeSeries.serialize(obj.ts)
return base64.b64encode(b).decode("utf-8")