metric

package
v0.0.0-...-5fdedb5 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Dec 17, 2024 License: Apache-2.0, MIT Imports: 24 Imported by: 34

Documentation

Overview

Package metric provides primitives for collecting metrics.

Index

Constants

View Source 
const (

	// MetricsPrefix is prepended before every metrics line.
	MetricsPrefix = "GVISOR_METRICS\t"
	// MetricsHashIndicator is prepended before the hash of the metrics
	// data at the end of the metrics stream.
	MetricsHashIndicator = "ADLER32\t"
	// TimeColumn is the column header for the time column.
	TimeColumn = "Time (ns)"
	// MetricsMetaIndicator is prepended before every metrics metadata line
	// after metricsPrefix.
	MetricsMetaIndicator = "META\t"
	// MetricsStartTimeIndicator is prepended before the start time of the
	// metrics collection.
	MetricsStartTimeIndicator = "START_TIME\t"
	// MetricsStatsIndicator is prepended before the stats of the metrics
	// collection process.
	MetricsStatsIndicator = "STATS\t"
)

Variables

View Source 
var (
	// ErrNameInUse indicates that another metric is already defined for
	// the given name.
	ErrNameInUse = errors.New("metric name already in use")

	// ErrInitializationDone indicates that the caller tried to create a
	// new metric after initialization.
	ErrInitializationDone = errors.New("metric cannot be created after initialization is complete")

	// ErrFieldValueContainsIllegalChar indicates that the value of a metric
	// field had an invalid character in it.
	ErrFieldValueContainsIllegalChar = errors.New("metric field value contains illegal character")

	// ErrFieldHasNoAllowedValues indicates that the field needs to define some
	// allowed values to be a valid and useful field.
	ErrFieldHasNoAllowedValues = errors.New("metric field does not define any allowed values")

	// ErrTooManyFieldCombinations indicates that the number of unique
	// combinations of fields is too large to support.
	ErrTooManyFieldCombinations = errors.New("metric has too many combinations of allowed field values")
)
View Source 
var (
	WeirdnessTypeTimeFallback         = FieldValue{"time_fallback"}
	WeirdnessTypePartialResult        = FieldValue{"partial_result"}
	WeirdnessTypeVsyscallCount        = FieldValue{"vsyscall_count"}
	WeirdnessTypeWatchdogStuckStartup = FieldValue{"watchdog_stuck_startup"}
	WeirdnessTypeWatchdogStuckTasks   = FieldValue{"watchdog_stuck_tasks"}
)

Weirdness metric type constants.

View Source 
var (
	// WeirdnessMetric is a metric with fields created to track the number
	// of weird occurrences such as time fallback, partial_result, vsyscall
	// count, watchdog startup timeouts and stuck tasks.
	WeirdnessMetric = MustCreateNewUint64Metric(
		"/weirdness",
		Uint64Metadata{
			Cumulative:  true,
			Sync:        true,
			Description: "Increment for weird occurrences of problems such as time fallback, partial result, vsyscalls invoked in the sandbox, watchdog startup timeouts and stuck tasks.",
			Fields: []Field{
				NewField("weirdness_type",
					&WeirdnessTypeTimeFallback,
					&WeirdnessTypePartialResult,
					&WeirdnessTypeVsyscallCount,
					&WeirdnessTypeWatchdogStuckStartup,
					&WeirdnessTypeWatchdogStuckTasks,
				),
			},
		})

	// SuspiciousOperationsMetric is a metric with fields created to detect
	// operations such as opening an executable file to write from a gofer.
	SuspiciousOperationsMetric = MustCreateNewUint64Metric(
		"/suspicious_operations",
		Uint64Metadata{
			Cumulative:  true,
			Sync:        true,
			Description: "Increment for suspicious operations such as opening an executable file to write from a gofer.",
			Fields: []Field{
				NewField("operation_type",
					&SuspiciousOperationsTypeOpenedWriteExecuteFile,
				),
			},
		})
)

List of global metrics that are used in multiple places.

View Source 
var ErrNotYetInitialized = errors.New("metrics are not yet initialized")

ErrNotYetInitialized is returned by GetMetricRegistration if metrics are not yet initialized.

View Source 
var SentryProfiling = FakeMetricBuilder{}

SentryProfiling is a builder that produces conditionally compiled metrics. Metrics made from this are compiled and active at runtime when the "sentry_profiling" go-tag is specified at compilation.

View Source 
var (
	SuspiciousOperationsTypeOpenedWriteExecuteFile = FieldValue{"opened_write_execute_file"}
)

Suspicious operations metric type constants.

Functions

func CheapNowNano 

func CheapNowNano() int64

CheapNowNano returns the a timestamp in nanoseconds. It is *NOT* measured from the Unix epoch. It is monotonic.

func Disable 

func Disable() error

Disable sends an empty metric registration event over the event channel, disabling metric collection.

Precondition:

  • All metrics are registered.
  • Initialize/Disable has not been called.

func EmitMetricUpdate 

func EmitMetricUpdate()

EmitMetricUpdate emits a MetricUpdate over the event channel.

Only metrics that have changed since the last call are emitted.

EmitMetricUpdate is thread-safe.

Preconditions:

  • Initialize has been called.

func GetMetricRegistration 

func GetMetricRegistration() (*pb.MetricRegistration, error)

GetMetricRegistration returns the metric registration data for all registered metrics. Must be called after Initialize(). Returns ErrNotYetInitialized if metrics are not yet initialized.

func GetSnapshot 

func GetSnapshot(options SnapshotOptions) (*prometheus.Snapshot, error)

GetSnapshot returns a Prometheus snapshot of the metric data. Returns ErrNotYetInitialized if metrics have not yet been initialized.

func Initialize 

func Initialize() error

Initialize sends a metric registration event over the event channel.

Precondition:

  • All metrics are registered.
  • Initialize/Disable has not been called.

func MustRegisterCustomUint64Metric 

func MustRegisterCustomUint64Metric(name string, metadata Uint64Metadata, value func(...*FieldValue) uint64)

MustRegisterCustomUint64Metric calls RegisterCustomUint64Metric for metrics without fields and panics if it returns an error.

func RegisterCustomUint64Metric 

func RegisterCustomUint64Metric(name string, metadata Uint64Metadata, value func(...*FieldValue) uint64) error

RegisterCustomUint64Metric registers a metric with the given name.

Register must only be called at init and will return and error if called after Initialized.

Preconditions:

  • name must be globally unique.
  • Initialize/Disable have not been called.
  • value is expected to accept exactly len(fields) arguments.

func StartProfilingMetrics 

func StartProfilingMetrics[T ProfilingMetricsWriter](opts ProfilingMetricsOptions[T]) error

StartProfilingMetrics checks the ProfilingMetrics runsc flags and creates goroutines responsible for outputting the profiling metric data.

Preconditions:

  • All metrics are registered.
  • Initialize/Disable has been called.

func StartStage 

func StartStage(stage InitStage) func()

StartStage should be called when an initialization stage is started. It returns a function that must be called to indicate that the stage ended. Alternatively, future calls to StartStage will implicitly indicate that the previous stage ended. Stage information will be emitted in the next call to EmitMetricUpdate after a stage has ended.

This function may (and is expected to) be called prior to final initialization of this metric library, as it has to capture early stages of Sentry initialization.

func StopProfilingMetrics 

func StopProfilingMetrics()

StopProfilingMetrics stops the profiling metrics goroutines. Call to make sure all metric data has been flushed. Note that calling this function prior to StartProfilingMetrics has no effect.

Types

type Bucketer 

type Bucketer interface {
	// NumFiniteBuckets is the number of finite buckets in the distribution.
	// This is only called once and never expected to return a different value.
	NumFiniteBuckets() int

	// LowerBound takes the index of a bucket (within [0, NumBuckets()]) and
	// returns the inclusive lower bound of that bucket.
	// In other words, the lowest value of `x` for which `BucketIndex(x) == i`
	// should be `x = LowerBound(i)`.
	// The upper bound of a bucket is the lower bound of the next bucket.
	// The last bucket (with `bucketIndex == NumFiniteBuckets()`) is infinite,
	// i.e. it has no upper bound (but it still has a lower bound).
	LowerBound(bucketIndex int) int64

	// BucketIndex takes a sample and returns the index of the bucket that the
	// sample should fall into.
	// Must return either:
	//   - A value within [0, NumBuckets() -1] if the sample falls within a
	//     finite bucket
	//   - NumBuckets() if the sample falls within the last (infinite) bucket
	//   - '-1' if the sample is lower than what any bucket can represent, i.e.
	//     the sample should be in the implicit "underflow" bucket.
	// This function must be go:nosplit-compatible and have no escapes.
	// +checkescape:all
	BucketIndex(sample int64) int
}

Bucketer is an interface to bucket values into finite, distinct buckets.

func NewDurationBucketer 

func NewDurationBucketer(numFiniteBuckets int, minDuration, maxDuration time.Duration) Bucketer

NewDurationBucketer returns a Bucketer well-suited for measuring durations in nanoseconds. Useful for NewTimerMetric. minDuration and maxDuration are conservative estimates of the minimum and maximum durations expected to be accurately measured by the Bucketer.

type CollectionStats 

type CollectionStats struct {

	// CollectionRate is the rate at which the metrics are meant to be
	// collected.
	CollectionRateNanos uint64 `json:"collection_rate_nanos"`

	// CheapStartNanos is the time at which the collector started in nanoseconds,
	// as returned by CheapNowNano.
	CheapStartNanos uint64 `json:"cheap_start_nanos"`

	// CheapLastCollectionNanos is the time at which the last collection was
	// meant to be performed, in nanoseconds as returned by CheapNowNano.
	CheapLastCollectionNanos uint64 `json:"cheap_last_collection_nanos"`

	// TotalSnapshots is the total number of snapshots successfully taken.
	TotalSnapshots uint64 `json:"total_snapshots"`

	// TotalSleepTimingError is the running sum of absolute difference in timing
	// between when the collector was meant to start collecting a metric snapshot
	// vs when it actually collected it. It should be divided by TotalSnapshots
	// to get the average sleep timing error.
	TotalSleepTimingErrorNanos uint64 `json:"total_sleep_timing_error_nanos"`

	// TotalCollectionTimingError is the running sum of time spent doing actual
	// metric collection, i.e. retrieving numerical values from metrics.
	// The larger this duration is, the more time gap there is between the first
	// metric being collected and the last within a single metric collection
	// cycle. This can cause the metric data to be less accurate because all of
	// these points will be recorded as having the same timestamp despite this
	// not actually being the case.
	// It should be divided by TotalSnapshots to get the average
	// per-collection-cycle collection timing error.
	TotalCollectionTimingErrorNanos uint64 `json:"total_collection_timing_error_nanos"`

	// NumBackoffSleeps is the number of times the collector had to back off
	// because the writer was too slow.
	NumBackoffSleeps uint64 `json:"num_backoff_sleeps"`

	// TotalBackoffSleep is the running sum of time the collector had to back
	// off because the writer was too slow.
	TotalBackoffSleepNanos uint64 `json:"total_backoff_sleep_nanos"`
	// contains filtered or unexported fields
}

CollectionStats contains statistics about the profiling metrics collection process itself.

func ParseCollectionStats 

func ParseCollectionStats(line string) (*CollectionStats, error)

ParseCollectionStats parses the profiling metrics collection stats from the given line.

func (*CollectionStats) Log 

func (s *CollectionStats) Log(infoFn func(format string, val ...any), warningFn func(format string, val ...any))

Log logs some statistics about the profiling metrics collection process.

func (*CollectionStats) WriteTo 

func (s *CollectionStats) WriteTo(w io.Writer) (int64, error)

WriteTo writes the statistics about the profiling metrics collection process to the given io.Writer.

type DistributionMetric 

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

DistributionMetric represents a distribution of values in finite buckets. It also separately keeps track of min/max in order to ascertain whether the buckets can faithfully represent the range of values encountered in the distribution.

func MustCreateNewDistributionMetric 

func MustCreateNewDistributionMetric(name string, sync bool, bucketer Bucketer, unit pb.MetricMetadata_Units, description string, fields ...Field) *DistributionMetric

MustCreateNewDistributionMetric creates and registers a distribution metric. If an error occurs, it panics.

func NewDistributionMetric 

func NewDistributionMetric(name string, sync bool, bucketer Bucketer, unit pb.MetricMetadata_Units, description string, fields ...Field) (*DistributionMetric, error)

NewDistributionMetric creates and registers a new distribution metric.

func (*DistributionMetric) AddSample 

func (d *DistributionMetric) AddSample(sample int64, fields ...*FieldValue)

AddSample adds a sample to the distribution. This *must* be called with the correct number of fields, or it will panic. +checkescape:all

type ExponentialBucketer 

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

ExponentialBucketer implements Bucketer, with the first bucket starting with 0 as lowest bound with `Width` width, and each subsequent bucket being wider by a scaled exponentially-growing series, until `NumFiniteBuckets` buckets exist.

func NewExponentialBucketer 

func NewExponentialBucketer(numFiniteBuckets int, width uint64, scale, growth float64) *ExponentialBucketer

NewExponentialBucketer returns a new Bucketer with exponential buckets.

func (*ExponentialBucketer) BucketIndex 

func (b *ExponentialBucketer) BucketIndex(sample int64) int

BucketIndex implements Bucketer.BucketIndex. +checkescape:all

func (*ExponentialBucketer) LowerBound 

func (b *ExponentialBucketer) LowerBound(bucketIndex int) int64

LowerBound implements Bucketer.LowerBound.

func (*ExponentialBucketer) NumFiniteBuckets 

func (b *ExponentialBucketer) NumFiniteBuckets() int

NumFiniteBuckets implements Bucketer.NumFiniteBuckets.

type FakeDistributionMetric 

type FakeDistributionMetric struct{}

FakeDistributionMetric is a type that implements all the methods of a DistributionMetric as a no-op.

func (*FakeDistributionMetric) AddSample 

func (d *FakeDistributionMetric) AddSample(sample int64, fields ...*FieldValue)

AddSample on a FakeUint64Metric does nothing.

type FakeMetricBuilder 

type FakeMetricBuilder struct{}

FakeMetricBuilder is a type used to produce conditionally compiled metrics. Methods of this struct produce fake, inactive metrics.

func (*FakeMetricBuilder) MustCreateNewDistributionMetric 

func (b *FakeMetricBuilder) MustCreateNewDistributionMetric(name string, sync bool, bucketer Bucketer, unit pb.MetricMetadata_Units, description string, fields ...Field) *FakeDistributionMetric

MustCreateNewDistributionMetric creates a fake distribution metric.

func (*FakeMetricBuilder) MustCreateNewTimerMetric 

func (b *FakeMetricBuilder) MustCreateNewTimerMetric(name string, nanoBucketer Bucketer, description string, fields ...Field) *FakeTimerMetric

MustCreateNewTimerMetric creates a fake timer metric.

func (*FakeMetricBuilder) MustCreateNewUint64Metric 

func (b *FakeMetricBuilder) MustCreateNewUint64Metric(name string, metadata Uint64Metadata) *FakeUint64Metric

MustCreateNewUint64Metric creates a fake Uint64 metric.

func (*FakeMetricBuilder) NewDistributionMetric 

func (b *FakeMetricBuilder) NewDistributionMetric(name string, sync bool, bucketer Bucketer, unit pb.MetricMetadata_Units, description string, fields ...Field) (*FakeDistributionMetric, error)

NewDistributionMetric creates a fake distribution metric.

func (*FakeMetricBuilder) NewTimerMetric 

func (b *FakeMetricBuilder) NewTimerMetric(name string, nanoBucketer Bucketer, description string, fields ...Field) (*FakeTimerMetric, error)

NewTimerMetric creates a fake timer metric.

func (*FakeMetricBuilder) NewUint64Metric 

func (b *FakeMetricBuilder) NewUint64Metric(name string, metadata Uint64Metadata) (*FakeUint64Metric, error)

NewUint64Metric creates a fake Uint64 metric.

type FakeTimedOperation 

type FakeTimedOperation struct{}

FakeTimedOperation is a type that implements all the methods of a TimedOperation as a no-op.

func (FakeTimedOperation) Finish 

func (o FakeTimedOperation) Finish(extraFields ...*FieldValue)

Finish on a FakeTimedOperation does nothing.

type FakeTimerMetric 

type FakeTimerMetric struct{}

FakeTimerMetric is a type that implements all the methods of a TimerMetric as a no-op.

func (*FakeTimerMetric) Start 

func (t *FakeTimerMetric) Start(fields ...*FieldValue) FakeTimedOperation

Start on a FakeUint64Metric returns a FakeTimedOperation struct, which does nothing and does not keep the time.

type FakeUint64Metric 

type FakeUint64Metric struct{}

FakeUint64Metric is a type that implements all the methods of a Uint64Metric as a no-op.

func (*FakeUint64Metric) Decrement 

func (m *FakeUint64Metric) Decrement(fieldValues ...*FieldValue)

Decrement on a FakeUint64Metric does nothing.

func (*FakeUint64Metric) Increment 

func (m *FakeUint64Metric) Increment(fieldValues ...*FieldValue)

Increment on a FakeUint64Metric does nothing.

func (*FakeUint64Metric) IncrementBy 

func (m *FakeUint64Metric) IncrementBy(v uint64, fieldValues ...*FieldValue)

IncrementBy on a FakeUint64Metric does nothing.

func (*FakeUint64Metric) Set 

func (m *FakeUint64Metric) Set(v uint64, fieldValues ...*FieldValue)

Set on a FakeUint64Metric does nothing.

func (*FakeUint64Metric) Value 

func (m *FakeUint64Metric) Value(fieldValues ...*FieldValue) uint64

Value from a FakeUint64Metric always returns a meaningless value.

type Field 

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

Field contains the field name and allowed values for the metric which is used in registration of the metric.

func NewField 

func NewField(name string, allowedValues ...*FieldValue) Field

NewField defines a new Field that can be used to break down a metric. The set of allowedValues must be unique strings wrapped with `FieldValue`. The *same* `FieldValue` pointers must be used during metric modifications. In practice, in most cases, this means you should declare these `FieldValue`s as package-level `var`s, and always use the address of these package-level `var`s during metric modifications.

type FieldValue 

type FieldValue struct {
	Value string
}

FieldValue is a string that can be used as a value for a Field. It must be referred to by address when the Field is created and when its metric value is modified. This ensures that the same FieldValue reference is used, which in turn enables the metric code to use the address of a FieldValue as comparison operator, rather than doing string comparisons.

type InitStage 

type InitStage string

InitStage is the name of a Sentry initialization stage. 

var (
	InitRestoreConfig InitStage = "restore_config"
	InitExecConfig    InitStage = "exec_config"
	InitRestore       InitStage = "restore"
	InitCreateProcess InitStage = "create_process"
	InitTaskStart     InitStage = "task_start"
)

List of all Sentry initialization stages.

type ProfilingMetricsOptions 

type ProfilingMetricsOptions[T ProfilingMetricsWriter] struct {
	// Sink is the sink to write the profiling metrics data to.
	Sink T

	// Lossy specifies whether the sink is lossy, i.e. data may be dropped from
	// too large logging volume. In this case, data integrity is desirable at the
	// expense of extra CPU cost at data-writing time. The data will be prefixed
	// with `MetricsPrefix` and the hash of the data will be appended at the end.
	Lossy bool

	// Metrics is the comma-separated list of metrics to profile.
	Metrics string

	// Rate is the rate at which the metrics are collected.
	Rate time.Duration
}

ProfilingMetricsOptions is the set of options to profile metrics.

type ProfilingMetricsWriter 

type ProfilingMetricsWriter interface {
	// Write from the io.Writer interface.
	io.Writer

	// WriteString from the io.StringWriter interface.
	io.StringWriter

	// Truncate truncates the underlying writer, if possible.
	Truncate(size int64) error

	// Close closes the writer.
	Close() error
}

ProfilingMetricsWriter is the interface for profiling metrics sinks.

type RealMetricBuilder 

type RealMetricBuilder struct{}

RealMetricBuilder is a type used to produce conditionally compiled metrics. Methods of this struct produce real active metrics.

func (*RealMetricBuilder) MustCreateNewDistributionMetric 

func (b *RealMetricBuilder) MustCreateNewDistributionMetric(name string, sync bool, bucketer Bucketer, unit pb.MetricMetadata_Units, description string, fields ...Field) *DistributionMetric

MustCreateNewDistributionMetric creates a real distribution metric or panics if unable to do so.

func (*RealMetricBuilder) MustCreateNewTimerMetric 

func (b *RealMetricBuilder) MustCreateNewTimerMetric(name string, nanoBucketer Bucketer, description string, fields ...Field) *TimerMetric

MustCreateNewTimerMetric creates a real timer metric or panics if unable to do so.

func (*RealMetricBuilder) MustCreateNewUint64Metric 

func (b *RealMetricBuilder) MustCreateNewUint64Metric(name string, metadata Uint64Metadata) *Uint64Metric

MustCreateNewUint64Metric creates a real Uint64 metric or panics if unable to do so.

func (*RealMetricBuilder) NewDistributionMetric 

func (b *RealMetricBuilder) NewDistributionMetric(name string, sync bool, bucketer Bucketer, unit pb.MetricMetadata_Units, description string, fields ...Field) (*DistributionMetric, error)

NewDistributionMetric calls the generic metric.NewDistributionMetric to produce a real distribution metric.

func (*RealMetricBuilder) NewTimerMetric 

func (b *RealMetricBuilder) NewTimerMetric(name string, nanoBucketer Bucketer, description string, fields ...Field) (*TimerMetric, error)

NewTimerMetric calls the generic metric.NewTimerMetric to produce a real timer metric.

func (*RealMetricBuilder) NewUint64Metric 

func (b *RealMetricBuilder) NewUint64Metric(name string, metadata Uint64Metadata) (*Uint64Metric, error)

NewUint64Metric calls the generic metric.NewUint64Metric to produce a real Uint64 metric.

type SnapshotOptions 

type SnapshotOptions struct {
	// Filter, if set, should return true for metrics that should be written to
	// the snapshot. If unset, all metrics are written to the snapshot.
	Filter func(*prometheus.Metric) bool
}

SnapshotOptions controls how snapshots are exported in GetSnapshot.

type TimedOperation 

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

TimedOperation is used by TimerMetric to keep track of the time elapsed between an operation starting and stopping.

func (TimedOperation) Finish 

func (o TimedOperation) Finish(extraFields ...*FieldValue)

Finish marks an operation as finished and records its duration. `extraFields` is the rest of the fields appended to the fields passed to `TimerMetric.Start`. The concatenation of these two must be the exact number of fields that the underlying metric has. +checkescape:all

type TimerMetric 

type TimerMetric struct {
	DistributionMetric
}

TimerMetric wraps a distribution metric with convenience functions for latency measurements, which is a popular specialization of distribution metrics.

func MustCreateNewTimerMetric 

func MustCreateNewTimerMetric(name string, nanoBucketer Bucketer, description string, fields ...Field) *TimerMetric

MustCreateNewTimerMetric creates and registers a timer metric. If an error occurs, it panics.

func NewTimerMetric 

func NewTimerMetric(name string, nanoBucketer Bucketer, description string, fields ...Field) (*TimerMetric, error)

NewTimerMetric provides a convenient way to measure latencies. The arguments are the same as `NewDistributionMetric`, except:

  • `nanoBucketer`: Same as `NewDistribution`'s `bucketer`, expected to hold durations in nanoseconds. Adjust parameters accordingly. NewDurationBucketer may be helpful here.

func (*TimerMetric) Start 

func (t *TimerMetric) Start(fields ...*FieldValue) TimedOperation

Start starts a timer measurement for the given combination of fields. It returns a TimedOperation which can be passed around as necessary to measure the duration of the operation. Once the operation is finished, call Finish on the TimedOperation. The fields passed to Start may be partially specified; if so, the remaining fields must be passed to TimedOperation.Finish. This is useful for cases where which path an operation took is only known after it happens. This path can be part of the fields passed to Finish. +checkescape:all

type Uint64Metadata 

type Uint64Metadata struct {
	Cumulative  bool
	Sync        bool
	Unit        pb.MetricMetadata_Units
	Description string
	Fields      []Field
}

Uint64Metadata is the metadata for a uint64 metric.

type Uint64Metric 

type Uint64Metric struct {
	// contains filtered or unexported fields
}

Uint64Metric encapsulates a uint64 that represents some kind of metric to be monitored.

Metrics are not saved across save/restore and thus reset to zero on restore.

func MustCreateNewUint64Metric 

func MustCreateNewUint64Metric(name string, metadata Uint64Metadata) *Uint64Metric

MustCreateNewUint64Metric calls NewUint64Metric and panics if it returns an error.

func NewUint64Metric 

func NewUint64Metric(name string, metadata Uint64Metadata) (*Uint64Metric, error)

NewUint64Metric creates and registers a new cumulative metric with the given name.

Metrics must be statically defined (i.e., at init).

func (*Uint64Metric) Decrement 

func (m *Uint64Metric) Decrement(fieldValues ...*FieldValue)

Decrement decrements the metric by 1. This must be called with the correct number of field values or it will panic.

func (*Uint64Metric) Increment 

func (m *Uint64Metric) Increment(fieldValues ...*FieldValue)

Increment increments the metric by 1. This must be called with the correct number of field values or it will panic.

func (*Uint64Metric) IncrementBy 

func (m *Uint64Metric) IncrementBy(v uint64, fieldValues ...*FieldValue)

IncrementBy increments the metric by v. It is also possible to use this function to decrement the metric by using a two's-complement int64 representation of the negative number to add. This must be called with the correct number of field values or it will panic.

func (*Uint64Metric) Set 

func (m *Uint64Metric) Set(v uint64, fieldValues ...*FieldValue)

Set sets the metric to v. This must be called with the correct number of field values or it will panic.

func (*Uint64Metric) Value 

func (m *Uint64Metric) Value(fieldValues ...*FieldValue) uint64

Value returns the current value of the metric for the given set of fields. This must be called with the correct number of field values or it will panic.

Source Files

View all Source files

Directories

Path Synopsis
buckettool prints buckets for distribution metrics.
 buckettool prints buckets for distribution metrics.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.