monitoring

Documentation

Overview

Package monitoring provides monitoring functionality.

Index

• Constants
• func LatencyBuckets() []float64
• func PercentileBuckets(inc int64) []float64
• func SetStartSpan(f startSpanFunc)
• func StartSpan(ctx context.Context, name string) (context.Context, func())
• type Counter
• type Gauge
• type Histogram
• type InertDistribution
  • func (m *InertDistribution) Info(labelVals ...string) (uint64, float64)
  • func (m *InertDistribution) Observe(val float64, labelVals ...string)
• type InertFloat
  • func (m *InertFloat) Add(val float64, labelVals ...string)
  • func (m *InertFloat) Dec(labelVals ...string)
  • func (m *InertFloat) Inc(labelVals ...string)
  • func (m *InertFloat) Set(val float64, labelVals ...string)
  • func (m *InertFloat) Value(labelVals ...string) float64
• type InertMetricFactory
  • func (imf InertMetricFactory) NewCounter(name, help string, labelNames ...string) Counter
  • func (imf InertMetricFactory) NewGauge(name, help string, labelNames ...string) Gauge
  • func (imf InertMetricFactory) NewHistogram(name, help string, labelNames ...string) Histogram
  • func (imf InertMetricFactory) NewHistogramWithBuckets(name, help string, _ []float64, labelNames ...string) Histogram
• type MetricFactory
• type RPCStatsInterceptor
  • func NewRPCStatsInterceptor(timeSource clock.TimeSource, prefix string, mf MetricFactory) *RPCStatsInterceptor
  • func (r *RPCStatsInterceptor) Interceptor() grpc.UnaryServerInterceptor

Constants

const (
	// TreeIDLabel is the monitoring label used to represent a tree ID.
	// TODO(codingllama): Consider using TreeIDLabel in place of log ID.
	TreeIDLabel = "tree_id"
)

Functions

func LatencyBuckets

func LatencyBuckets() []float64

LatencyBuckets returns a reasonable range of histogram upper limits for most latency-in-seconds usecases.

func PercentileBuckets

func PercentileBuckets(inc int64) []float64

PercentileBuckets returns a range of buckets for 0.0-100.0% use cases. in specified integer increments. The increment must be at least 1%, which prevents creating very large metric exports.

func SetStartSpan

func SetStartSpan(f startSpanFunc)

SetStartSpan sets the function used to start tracing spans. This may be used to add runtime support for different tracing implementation.

func StartSpan

func StartSpan(ctx context.Context, name string) (context.Context, func())

StartSpan starts a new tracing span using the given message. The returned context should be used for all child calls within the span, and the returned func should be called to close the span.

The default implementation of this method is a no-op; insert a real tracing span implementation by setting this global variable to the relevant function at start of day.

Types

type Counter

type Counter interface {
	Inc(labelVals ...string)
	Add(val float64, labelVals ...string)
	Value(labelVals ...string) float64
}

Counter is a metric class for numeric values that increase.

type Gauge

type Gauge interface {
	Inc(labelVals ...string)
	Dec(labelVals ...string)
	Add(val float64, labelVals ...string)
	Set(val float64, labelVals ...string)
	// Value retrieves the value for a particular set of labels.
	// This is only really useful for testing implementations.
	Value(labelVals ...string) float64
}

Gauge is a metric class for numeric values that can go up and down.

type Histogram

type Histogram interface {
	Observe(val float64, labelVals ...string)
	// Info retrieves the count and sum of observations for a particular set of labels.
	// This is only really useful for testing implementations.
	Info(labelVals ...string) (uint64, float64)
}

Histogram is a metric class that tracks the distribution of a collection of observations.

type InertDistribution

type InertDistribution struct {
	// contains filtered or unexported fields
}

InertDistribution is an internal-only implementation of the Distribution interface.

func (*InertDistribution) Info

func (m *InertDistribution) Info(labelVals ...string) (uint64, float64)

Info returns count, sum for the distribution.

func (*InertDistribution) Observe

func (m *InertDistribution) Observe(val float64, labelVals ...string)

Observe adds a single observation to the distribution.

type InertFloat

type InertFloat struct {
	// contains filtered or unexported fields
}

InertFloat is an internal-only implementation of both the Counter and Gauge interfaces.

func (*InertFloat) Add

func (m *InertFloat) Add(val float64, labelVals ...string)

Add adds the given amount to the value.

func (*InertFloat) Dec

func (m *InertFloat) Dec(labelVals ...string)

Dec subtracts 1 from the value.

func (*InertFloat) Inc

func (m *InertFloat) Inc(labelVals ...string)

Inc adds 1 to the value.

func (*InertFloat) Set

func (m *InertFloat) Set(val float64, labelVals ...string)

Set sets the value.

func (*InertFloat) Value

func (m *InertFloat) Value(labelVals ...string) float64

Value returns the current value.

type InertMetricFactory

type InertMetricFactory struct{}

InertMetricFactory creates inert metrics for testing.

func (InertMetricFactory) NewCounter

func (imf InertMetricFactory) NewCounter(name, help string, labelNames ...string) Counter

NewCounter creates a new inert Counter.

func (InertMetricFactory) NewGauge

func (imf InertMetricFactory) NewGauge(name, help string, labelNames ...string) Gauge

NewGauge creates a new inert Gauge.

func (InertMetricFactory) NewHistogram

func (imf InertMetricFactory) NewHistogram(name, help string, labelNames ...string) Histogram

NewHistogram creates a new inert Histogram.

func (InertMetricFactory) NewHistogramWithBuckets

func (imf InertMetricFactory) NewHistogramWithBuckets(name, help string, _ []float64, labelNames ...string) Histogram

NewHistogramWithBuckets creates a new inert Histogram with supplied buckets. The buckets are not actually used.

type MetricFactory

type MetricFactory interface {
	NewCounter(name, help string, labelNames ...string) Counter
	NewGauge(name, help string, labelNames ...string) Gauge
	NewHistogram(name, help string, labelNames ...string) Histogram
	NewHistogramWithBuckets(name, help string, buckets []float64, labelNames ...string) Histogram
}

MetricFactory allows the creation of different types of metric.

type RPCStatsInterceptor

type RPCStatsInterceptor struct {
	ReqCount          Counter
	ReqSuccessCount   Counter
	ReqSuccessLatency Histogram
	ReqErrorCount     Counter
	ReqErrorLatency   Histogram
	// contains filtered or unexported fields
}

RPCStatsInterceptor provides a gRPC interceptor that records statistics about the RPCs passing through it.

func NewRPCStatsInterceptor

func NewRPCStatsInterceptor(timeSource clock.TimeSource, prefix string, mf MetricFactory) *RPCStatsInterceptor

NewRPCStatsInterceptor creates a new RPCStatsInterceptor for the given application/component, with a specified time source.

func (*RPCStatsInterceptor) Interceptor

func (r *RPCStatsInterceptor) Interceptor() grpc.UnaryServerInterceptor

Interceptor returns a UnaryServerInterceptor that can be registered with an RPC server and will record request counts / errors and latencies for that servers handlers