Skip to content

Commit

Permalink
Implement go.schedule.duration (#5991)
Browse files Browse the repository at this point in the history
Fixes
#5974.

This uses the same method for estimating the sum as the prometheus
client-go:
https://github.com/prometheus/client_golang/blob/46f77a97fa1e83e7285634745bd5c92c11bf20c7/prometheus/go_collector_latest.go#L543,
which is to use the lower-bound of each bucket as the value of each
observation.

It implements this using a metric producer. It is designed to be used
together with existing runtime instrumentation.

---------

Co-authored-by: Sam Xie <sam@samxie.me>
  • Loading branch information
dashpole and XSAM authored Sep 4, 2024
1 parent 06ace3e commit 8f5330c
Show file tree
Hide file tree
Showing 17 changed files with 423 additions and 365 deletions.
4 changes: 4 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,10 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm

## [Unreleased]

### Added

- Add `NewProducer` to `go.opentelemetry.io/contrib/instrumentation/runtime`, which allows collecting the `go.schedule.duration` histogram metric from the Go runtime. (#5991)

### Removed

- Drop support for [Go 1.21]. (#6046, #6047)
Expand Down
5 changes: 0 additions & 5 deletions instrumentation/runtime/example/doc.go

This file was deleted.

22 changes: 0 additions & 22 deletions instrumentation/runtime/example/go.mod

This file was deleted.

31 changes: 0 additions & 31 deletions instrumentation/runtime/example/go.sum

This file was deleted.

56 changes: 0 additions & 56 deletions instrumentation/runtime/example/main.go

This file was deleted.

38 changes: 38 additions & 0 deletions instrumentation/runtime/example_test.go
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
// Copyright The OpenTelemetry Authors
// SPDX-License-Identifier: Apache-2.0

package runtime_test

import (
"context"
"log"
"time"

"go.opentelemetry.io/contrib/instrumentation/runtime"
"go.opentelemetry.io/otel"
"go.opentelemetry.io/otel/sdk/metric"
)

func Example() {
// This reader is used as a stand-in for a reader that will actually export
// data. See https://pkg.go.dev/go.opentelemetry.io/otel/exporters for
// exporters that can be used as or with readers.
reader := metric.NewManualReader(
// Add the runtime producer to get histograms from the Go runtime.
metric.WithProducer(runtime.NewProducer()),
)
provider := metric.NewMeterProvider(metric.WithReader(reader))
defer func() {
err := provider.Shutdown(context.Background())
if err != nil {
log.Fatal(err)
}
}()
otel.SetMeterProvider(provider)

// Start go runtime metric collection.
err := runtime.Start(runtime.WithMinimumReadMemStatsInterval(time.Second))
if err != nil {
log.Fatal(err)
}
}
4 changes: 4 additions & 0 deletions instrumentation/runtime/go.mod
Original file line number Diff line number Diff line change
Expand Up @@ -6,13 +6,17 @@ require (
github.com/stretchr/testify v1.9.0
go.opentelemetry.io/otel v1.29.0
go.opentelemetry.io/otel/metric v1.29.0
go.opentelemetry.io/otel/sdk v1.29.0
go.opentelemetry.io/otel/sdk/metric v1.29.0
)

require (
github.com/davecgh/go-spew v1.1.1 // indirect
github.com/go-logr/logr v1.4.2 // indirect
github.com/go-logr/stdr v1.2.2 // indirect
github.com/google/uuid v1.6.0 // indirect
github.com/pmezard/go-difflib v1.0.0 // indirect
go.opentelemetry.io/otel/trace v1.29.0 // indirect
golang.org/x/sys v0.24.0 // indirect
gopkg.in/yaml.v3 v3.0.1 // indirect
)
8 changes: 8 additions & 0 deletions instrumentation/runtime/go.sum
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,8 @@ github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
github.com/stretchr/testify v1.9.0 h1:HtqpIVDClZ4nwg75+f6Lvsy/wHu+3BoSGCbBAcpTsTg=
Expand All @@ -15,8 +17,14 @@ go.opentelemetry.io/otel v1.29.0 h1:PdomN/Al4q/lN6iBJEN3AwPvUiHPMlt93c8bqTG5Llw=
go.opentelemetry.io/otel v1.29.0/go.mod h1:N/WtXPs1CNCUEx+Agz5uouwCba+i+bJGFicT8SR4NP8=
go.opentelemetry.io/otel/metric v1.29.0 h1:vPf/HFWTNkPu1aYeIsc98l4ktOQaL6LeSoeV2g+8YLc=
go.opentelemetry.io/otel/metric v1.29.0/go.mod h1:auu/QWieFVWx+DmQOUMgj0F8LHWdgalxXqvp7BII/W8=
go.opentelemetry.io/otel/sdk v1.29.0 h1:vkqKjk7gwhS8VaWb0POZKmIEDimRCMsopNYnriHyryo=
go.opentelemetry.io/otel/sdk v1.29.0/go.mod h1:pM8Dx5WKnvxLCb+8lG1PRNIDxu9g9b9g59Qr7hfAAok=
go.opentelemetry.io/otel/sdk/metric v1.29.0 h1:K2CfmJohnRgvZ9UAj2/FhIf/okdWcNdBwe1m8xFXiSY=
go.opentelemetry.io/otel/sdk/metric v1.29.0/go.mod h1:6zZLdCl2fkauYoZIOn/soQIDSWFmNSRcICarHfuhNJQ=
go.opentelemetry.io/otel/trace v1.29.0 h1:J/8ZNK4XgR7a21DZUAsbF8pZ5Jcw1VhACmnYt39JTi4=
go.opentelemetry.io/otel/trace v1.29.0/go.mod h1:eHl3w0sp3paPkYstJOmAimxhiFXPg+MMTlEh3nsQgWQ=
golang.org/x/sys v0.24.0 h1:Twjiwq9dn6R1fQcyiK+wQyHWfaz/BJB+YIpzU/Cv3Xg=
golang.org/x/sys v0.24.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
Expand Down
27 changes: 25 additions & 2 deletions instrumentation/runtime/options.go
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,13 @@ type Option interface {
apply(*config)
}

// ProducerOption supports configuring optional settings for runtime metrics using a
// metric producer in addition to standard instrumentation.
type ProducerOption interface {
Option
applyProducer(*config)
}

// DefaultMinimumReadMemStatsInterval is the default minimum interval
// between calls to runtime.ReadMemStats(). Use the
// WithMinimumReadMemStatsInterval() option to modify this setting in
Expand All @@ -48,6 +55,8 @@ func (o minimumReadMemStatsIntervalOption) apply(c *config) {
}
}

func (o minimumReadMemStatsIntervalOption) applyProducer(c *config) { o.apply(c) }

// WithMeterProvider sets the Metric implementation to use for
// reporting. If this option is not used, the global metric.MeterProvider
// will be used. `provider` must be non-nil.
Expand All @@ -66,11 +75,25 @@ func (o metricProviderOption) apply(c *config) {
// newConfig computes a config from the supplied Options.
func newConfig(opts ...Option) config {
c := config{
MeterProvider: otel.GetMeterProvider(),
MinimumReadMemStatsInterval: DefaultMinimumReadMemStatsInterval,
MeterProvider: otel.GetMeterProvider(),
}
for _, opt := range opts {
opt.apply(&c)
}
if c.MinimumReadMemStatsInterval <= 0 {
c.MinimumReadMemStatsInterval = DefaultMinimumReadMemStatsInterval
}
return c
}

// newConfig computes a config from the supplied ProducerOptions.
func newProducerConfig(opts ...ProducerOption) config {
c := config{}
for _, opt := range opts {
opt.applyProducer(&c)
}
if c.MinimumReadMemStatsInterval <= 0 {
c.MinimumReadMemStatsInterval = DefaultMinimumReadMemStatsInterval
}
return c
}
120 changes: 120 additions & 0 deletions instrumentation/runtime/producer.go
Original file line number Diff line number Diff line change
@@ -0,0 +1,120 @@
// Copyright The OpenTelemetry Authors
// SPDX-License-Identifier: Apache-2.0

package runtime // import "go.opentelemetry.io/contrib/instrumentation/runtime"

import (
"context"
"fmt"
"math"
"runtime/metrics"
"sync"
"time"

"go.opentelemetry.io/otel/attribute"
"go.opentelemetry.io/otel/sdk/instrumentation"
"go.opentelemetry.io/otel/sdk/metric"
"go.opentelemetry.io/otel/sdk/metric/metricdata"
)

var startTime time.Time

func init() {
startTime = time.Now()
}

var histogramMetrics = []string{goSchedLatencies}

// Producer is a metric.Producer, which provides precomputed histogram metrics from the go runtime.
type Producer struct {
lock sync.Mutex
collector *goCollector
}

var _ metric.Producer = (*Producer)(nil)

// NewProducer creates a Producer which provides precomputed histogram metrics from the go runtime.
func NewProducer(opts ...ProducerOption) *Producer {
c := newProducerConfig(opts...)
return &Producer{
collector: newCollector(c.MinimumReadMemStatsInterval, histogramMetrics),
}
}

// Produce returns precomputed histogram metrics from the go runtime, or an error if unsuccessful.
func (p *Producer) Produce(context.Context) ([]metricdata.ScopeMetrics, error) {
p.lock.Lock()
p.collector.refresh()
schedHist := p.collector.getHistogram(goSchedLatencies)
p.lock.Unlock()
// Use the last collection time (which may or may not be now) for the timestamp.
histDp := convertRuntimeHistogram(schedHist, p.collector.lastCollect)
if len(histDp) == 0 {
return nil, fmt.Errorf("unable to obtain go.schedule.duration metric from the runtime")
}
return []metricdata.ScopeMetrics{
{
Scope: instrumentation.Scope{
Name: ScopeName,
Version: Version(),
},
Metrics: []metricdata.Metrics{
{
Name: "go.schedule.duration",
Description: "The time goroutines have spent in the scheduler in a runnable state before actually running.",
Unit: "s",
Data: metricdata.Histogram[float64]{
Temporality: metricdata.CumulativeTemporality,
DataPoints: histDp,
},
},
},
},
}, nil
}

var emptySet = attribute.EmptySet()

func convertRuntimeHistogram(runtimeHist *metrics.Float64Histogram, ts time.Time) []metricdata.HistogramDataPoint[float64] {
if runtimeHist == nil {
return nil
}
bounds := runtimeHist.Buckets
counts := runtimeHist.Counts
if len(bounds) < 2 {
// runtime histograms are guaranteed to have at least two bucket boundaries.
return nil
}
// trim the first bucket since it is a lower bound. OTel histogram boundaries only have an upper bound.
bounds = bounds[1:]
if bounds[len(bounds)-1] == math.Inf(1) {
// trim the last bucket if it is +Inf, since the +Inf boundary is implicit in OTel.
bounds = bounds[:len(bounds)-1]
} else {
// if the last bucket is not +Inf, append an extra zero count since
// the implicit +Inf bucket won't have any observations.
counts = append(counts, 0)
}
count := uint64(0)
sum := float64(0)
for i, c := range counts {
count += c
// This computed sum is an underestimate, since it assumes each
// observation happens at the bucket's lower bound.
if i > 0 && count != 0 {
sum += bounds[i-1] * float64(count)
}
}

return []metricdata.HistogramDataPoint[float64]{
{
StartTime: startTime,
Count: count,
Sum: sum,
Time: ts,
Bounds: bounds,
BucketCounts: counts,
Attributes: *emptySet,
},
}
}
Loading

0 comments on commit 8f5330c

Please sign in to comment.