ironic-lib/ironic_lib/metrics.py

303 lines
9.3 KiB
Python

# Copyright 2016 Rackspace Hosting
# All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import abc
import functools
import random
import time
from ironic_lib.common.i18n import _
class Timer(object):
"""A timer decorator and context manager.
This metric type times the decorated method or code running inside the
context manager, and emits the time as the metric value. It is bound to
this MetricLogger. For example::
from ironic_lib import metrics_utils
METRICS = metrics_utils.get_metrics_logger()
@METRICS.timer('foo')
def foo(bar, baz):
print bar, baz
with METRICS.timer('foo'):
do_something()
"""
def __init__(self, metrics, name):
"""Init the decorator / context manager.
:param metrics: The metric logger
:param name: The metric name
"""
if not isinstance(name, str):
raise TypeError(_("The metric name is expected to be a string. "
"Value is %s") % name)
self.metrics = metrics
self.name = name
self._start = None
def __call__(self, f):
@functools.wraps(f)
def wrapped(*args, **kwargs):
start = _time()
result = f(*args, **kwargs)
duration = _time() - start
# Log the timing data (in ms)
self.metrics.send_timer(self.metrics.get_metric_name(self.name),
duration * 1000)
return result
return wrapped
def __enter__(self):
self._start = _time()
def __exit__(self, exc_type, exc_val, exc_tb):
duration = _time() - self._start
# Log the timing data (in ms)
self.metrics.send_timer(self.metrics.get_metric_name(self.name),
duration * 1000)
class Counter(object):
"""A counter decorator and context manager.
This metric type increments a counter every time the decorated method or
context manager is executed. It is bound to this MetricLogger. For
example::
from ironic_lib import metrics_utils
METRICS = metrics_utils.get_metrics_logger()
@METRICS.counter('foo')
def foo(bar, baz):
print bar, baz
with METRICS.counter('foo'):
do_something()
"""
def __init__(self, metrics, name, sample_rate):
"""Init the decorator / context manager.
:param metrics: The metric logger
:param name: The metric name
:param sample_rate: Probabilistic rate at which the values will be sent
"""
if not isinstance(name, str):
raise TypeError(_("The metric name is expected to be a string. "
"Value is %s") % name)
if (sample_rate is not None
and (sample_rate < 0.0 or sample_rate > 1.0)):
msg = _("sample_rate is set to %s. Value must be None "
"or in the interval [0.0, 1.0]") % sample_rate
raise ValueError(msg)
self.metrics = metrics
self.name = name
self.sample_rate = sample_rate
def __call__(self, f):
@functools.wraps(f)
def wrapped(*args, **kwargs):
self.metrics.send_counter(
self.metrics.get_metric_name(self.name),
1, sample_rate=self.sample_rate)
result = f(*args, **kwargs)
return result
return wrapped
def __enter__(self):
self.metrics.send_counter(self.metrics.get_metric_name(self.name),
1, sample_rate=self.sample_rate)
def __exit__(self, exc_type, exc_val, exc_tb):
pass
class Gauge(object):
"""A gauge decorator.
This metric type returns the value of the decorated method as a metric
every time the method is executed. It is bound to this MetricLogger. For
example::
from ironic_lib import metrics_utils
METRICS = metrics_utils.get_metrics_logger()
@METRICS.gauge('foo')
def add_foo(bar, baz):
return (bar + baz)
"""
def __init__(self, metrics, name):
"""Init the decorator / context manager.
:param metrics: The metric logger
:param name: The metric name
"""
if not isinstance(name, str):
raise TypeError(_("The metric name is expected to be a string. "
"Value is %s") % name)
self.metrics = metrics
self.name = name
def __call__(self, f):
@functools.wraps(f)
def wrapped(*args, **kwargs):
result = f(*args, **kwargs)
self.metrics.send_gauge(self.metrics.get_metric_name(self.name),
result)
return result
return wrapped
def _time():
"""Wraps time.time() for simpler testing."""
return time.time()
class MetricLogger(object, metaclass=abc.ABCMeta):
"""Abstract class representing a metrics logger.
A MetricLogger sends data to a backend (noop or statsd).
The data can be a gauge, a counter, or a timer.
The data sent to the backend is composed of:
- a full metric name
- a numeric value
The format of the full metric name is:
_prefix<delim>name
where:
- _prefix: [global_prefix<delim>][uuid<delim>][host_name<delim>]prefix
- name: the name of this metric
- <delim>: the delimiter. Default is '.'
"""
def __init__(self, prefix='', delimiter='.'):
"""Init a MetricLogger.
:param prefix: Prefix for this metric logger. This string will prefix
all metric names.
:param delimiter: Delimiter used to generate the full metric name.
"""
self._prefix = prefix
self._delimiter = delimiter
def get_metric_name(self, name):
"""Get the full metric name.
The format of the full metric name is:
_prefix<delim>name
where:
- _prefix: [global_prefix<delim>][uuid<delim>][host_name<delim>]
prefix
- name: the name of this metric
- <delim>: the delimiter. Default is '.'
:param name: The metric name.
:return: The full metric name, with logger prefix, as a string.
"""
if not self._prefix:
return name
return self._delimiter.join([self._prefix, name])
def send_gauge(self, name, value):
"""Send gauge metric data.
Gauges are simple values.
The backend will set the value of gauge 'name' to 'value'.
:param name: Metric name
:param value: Metric numeric value that will be sent to the backend
"""
self._gauge(name, value)
def send_counter(self, name, value, sample_rate=None):
"""Send counter metric data.
Counters are used to count how many times an event occurred.
The backend will increment the counter 'name' by the value 'value'.
Optionally, specify sample_rate in the interval [0.0, 1.0] to
sample data probabilistically where::
P(send metric data) = sample_rate
If sample_rate is None, then always send metric data, but do not
have the backend send sample rate information (if supported).
:param name: Metric name
:param value: Metric numeric value that will be sent to the backend
:param sample_rate: Probabilistic rate at which the values will be
sent. Value must be None or in the interval [0.0, 1.0].
"""
if (sample_rate is None or random.random() < sample_rate):
return self._counter(name, value,
sample_rate=sample_rate)
def send_timer(self, name, value):
"""Send timer data.
Timers are used to measure how long it took to do something.
:param m_name: Metric name
:param m_value: Metric numeric value that will be sent to the backend
"""
self._timer(name, value)
def timer(self, name):
return Timer(self, name)
def counter(self, name, sample_rate=None):
return Counter(self, name, sample_rate)
def gauge(self, name):
return Gauge(self, name)
@abc.abstractmethod
def _gauge(self, name, value):
"""Abstract method for backends to implement gauge behavior."""
@abc.abstractmethod
def _counter(self, name, value, sample_rate=None):
"""Abstract method for backends to implement counter behavior."""
@abc.abstractmethod
def _timer(self, name, value):
"""Abstract method for backends to implement timer behavior."""
class NoopMetricLogger(MetricLogger):
"""Noop metric logger that throws away all metric data."""
def _gauge(self, name, value):
pass
def _counter(self, name, value, sample_rate=None):
pass
def _timer(self, m_name, value):
pass