def get_instance_cpu_usage(self, instance):
metric_name = self.METRIC_NAMES[
self.config.datasource]['instance_cpu_usage']
if self.config.datasource == "ceilometer":
return self.ceilometer.statistic_aggregation(
resource_id=instance.uuid,
meter_name=metric_name,
period=self.period,
aggregate='avg')
elif self.config.datasource == "monasca":
statistics = self.monasca.statistic_aggregation(
meter_name=metric_name,
dimensions=dict(resource_id=instance.uuid),
period=self.period,
aggregate='avg')
cpu_usage = None
for stat in statistics:
avg_col_idx = stat['columns'].index('avg')
values = [r[avg_col_idx] for r in stat['statistics']]
value = float(sum(values)) / len(values)
cpu_usage = value
return cpu_usage
raise exception.UnsupportedDataSource(
strategy=self.name, datasource=self.config.datasource)
def statistic_aggregation(self,
resource_id=None,
meter_name=None,
period=300,
granularity=300,
dimensions=None,
aggregation='avg',
group_by='*'):
"""Representing a statistic aggregate by operators
:param resource_id: id of resource to list statistics for.
This param isn't used in Monasca datasource.
:param meter_name: meter names of which we want the statistics.
:param period: Sampling `period`: In seconds. If no period is given,
only one aggregate statistic is returned. If given, a
faceted result will be returned, divided into given
periods. Periods with no data are ignored.
:param granularity: frequency of marking metric point, in seconds.
This param isn't used in Ceilometer datasource.
:param dimensions: dimensions (dict).
:param aggregation: Should be either 'avg', 'count', 'min' or 'max'.
:param group_by: list of columns to group the metrics to be returned.
:return: A list of dict with each dict being a distinct result row
"""
if dimensions is None:
raise exception.UnsupportedDataSource(datasource='Monasca')
stop_time = datetime.datetime.utcnow()
start_time = stop_time - datetime.timedelta(seconds=(int(period)))
if aggregation == 'mean':
aggregation = 'avg'
raw_kwargs = dict(
name=meter_name,
start_time=start_time.isoformat(),
end_time=stop_time.isoformat(),
dimensions=dimensions,
period=period,
statistics=aggregation,
group_by=group_by,
)
kwargs = {k: v for k, v in raw_kwargs.items() if k and v}
statistics = self.query_retry(f=self.monasca.metrics.list_statistics,
**kwargs)
cpu_usage = None
for stat in statistics:
avg_col_idx = stat['columns'].index(aggregation)
values = [r[avg_col_idx] for r in stat['statistics']]
value = float(sum(values)) / len(values)
cpu_usage = value
return cpu_usage
def get_node_cpu_usage(self, node):
metric_name = self.METRIC_NAMES[
self.config.datasource]['host_cpu_usage']
if self.config.datasource == "ceilometer":
resource_id = "%s_%s" % (node.uuid, node.hostname)
return self.ceilometer.statistic_aggregation(
resource_id=resource_id,
meter_name=metric_name,
period=self.period,
aggregate='avg',
)
elif self.config.datasource == "gnocchi":
resource_id = "%s_%s" % (node.uuid, node.hostname)
stop_time = datetime.datetime.utcnow()
start_time = stop_time - datetime.timedelta(
seconds=int(self.period))
return self.gnocchi.statistic_aggregation(
resource_id=resource_id,
metric=metric_name,
granularity=self.granularity,
start_time=start_time,
stop_time=stop_time,
aggregation='mean')
elif self.config.datasource == "monasca":
statistics = self.monasca.statistic_aggregation(
meter_name=metric_name,
dimensions=dict(hostname=node.uuid),
period=self.period,
aggregate='avg')
cpu_usage = None
for stat in statistics:
avg_col_idx = stat['columns'].index('avg')
values = [r[avg_col_idx] for r in stat['statistics']]
value = float(sum(values)) / len(values)
cpu_usage = value
return cpu_usage
raise exception.UnsupportedDataSource(
strategy=self.name, datasource=self.config.datasource)