You've already forked opentelemetry-go
mirror of
https://github.com/open-telemetry/opentelemetry-go.git
synced 2025-11-27 22:49:15 +02:00
Fixes #7004 This PR adds support for experimental otel.sdk.processor.span.processed metric in simple span processor. Definition of metric at: https://github.com/open-telemetry/semantic-conventions/blob/v1.36.0/docs/otel/sdk-metrics.md Experimental metrics are behind a feature flag: `OTEL_GO_X_OBSERVABILITY` <details> <summary>Observability Implementation Checklist</summary> ## Observability Implementation Checklist Based on the [project Observability guidelines](e4ab314112/CONTRIBUTING.md (observability)), ensure the following are completed: ### Environment Variable Activation * [x] Observability features are disabled by default * [x] Features are activated through the `OTEL_GO_X_OBSERVABILITY` environment variable * [x] Use consistent pattern with `x.Observability.Enabled()` check [^1] * [x] Follow established experimental feature pattern [^2][^3] [^1]:e4ab314112/exporters/stdout/stdouttrace/internal/observ/instrumentation.go (L101-L103)[^2]:e4ab314112/exporters/stdout/stdouttrace/internal/x/x.go[^3]:e4ab314112/sdk/internal/x/x.go### Encapsulation * [x] Instrumentation is encapsulated within a dedicated `struct` (e.g., [`Instrumentation`](e4ab314112/exporters/stdout/stdouttrace/internal/observ/instrumentation.go (L86-L94))) * [x] Instrumentation is not mixed into the instrumented component * [x] Instrumentation code is in its own file or package if complex/reused * [x] Instrumentation setup doesn't bloat the main component code ### Initialization * [x] Initialization is only done when observability is enabled * [x] Setup is explicit and side-effect free * [x] Return errors from initialization when appropriate * [x] Use the global Meter provider (e.g., `otel.GetMeterProvider()`) * [x] Include proper meter configuration with: * [x] Instrumentation package name is used for the Meter * [x] Instrumentation version (e.g. [`Version`](e4ab314112/exporters/stdout/stdouttrace/internal/observ/instrumentation.go (L40-L43))) * [x] Schema URL (e.g. [`SchemaURL`](e4ab314112/exporters/stdout/stdouttrace/internal/observ/instrumentation.go (L36-L38))) ### Performance * [x] Little to no overhead when observability is disabled * [x] Expensive operations are only executed when observability is enabled * [x] When enabled, instrumentation code paths are optimized to reduce allocation/computation overhead #### Attribute and Option Allocation Management * [x] Use `sync.Pool` for attribute slices and options with dynamic attributes * [x] Pool objects are properly reset before returning to pool * [x] Pools are scoped for maximum efficiency while ensuring correctness #### Caching * [x] Static attribute sets known at compile time are pre-computed and cached * [x] Common attribute combinations use lookup tables/maps #### Benchmarking * [x] Benchmarks provided for all instrumentation code * [ ] Benchmark scenarios include both enabled and disabled observability * [x] Benchmark results show impact on allocs/op, B/op, and ns/op (use `b.ReportAllocs()` in benchmarks) ### Error Handling and Robustness * [x] Errors are reported back to caller when possible * [x] Partial failures are handled gracefully * [x] Use partially initialized components when available * [x] Return errors to caller instead of only using `otel.Handle()` * [x] Use `otel.Handle()` only when component cannot report error to user ### Context Propagation * [x] Observability measurements receive the context from the function being measured (don't break context propagation by using `context.Background()`) ### Semantic Conventions Compliance * [x] All metrics follow [OpenTelemetry Semantic Conventions](5ee549b1ce/docs/otel/sdk-metrics.md) * [x] Use the [`otelconv`](https://pkg.go.dev/go.opentelemetry.io/otel@v1.38.0/semconv/v1.37.0/otelconv) convenience package for metric semantic conventions * [x] Component names follow semantic conventions * [x] Use package path scope type as stable identifier for non-standard components * [x] Component names are stable unique identifiers * [x] Use global counter for uniqueness if necessary * [x] Component ID counter is resettable for deterministic testing ### Testing * [x] Use deterministic testing with isolated state * [x] Restore previous state after tests (`t.Cleanup()`) * [x] Isolate meter provider for testing * [x] Use `t.Setenv()` for environment variable testing * [x] Reset component ID counter for deterministic component names * [x] Test order doesn't affect results </details> ### Benchmarks ```console > benchstat bmark.result goos: darwin goarch: arm64 pkg: go.opentelemetry.io/otel/sdk/trace/internal/observ cpu: Apple M1 Pro │ bmark.result │ │ sec/op │ SSP/SpanProcessed-8 146.7n ± 15% SSP/SpanProcessedWithError-8 205.1n ± 3% geomean 173.5n │ bmark.result │ │ B/op │ SSP/SpanProcessed-8 280.0 ± 0% SSP/SpanProcessedWithError-8 408.0 ± 0% geomean 338.0 │ bmark.result │ │ allocs/op │ SSP/SpanProcessed-8 3.000 ± 0% SSP/SpanProcessedWithError-8 3.000 ± 0% geomean 3.000 ``` --------- Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com>
331 lines
8.8 KiB
Go
331 lines
8.8 KiB
Go
// Copyright The OpenTelemetry Authors
|
|
// SPDX-License-Identifier: Apache-2.0
|
|
|
|
package trace
|
|
|
|
import (
|
|
"context"
|
|
"errors"
|
|
"strconv"
|
|
"sync"
|
|
"testing"
|
|
"time"
|
|
|
|
"github.com/stretchr/testify/assert"
|
|
"github.com/stretchr/testify/require"
|
|
|
|
"go.opentelemetry.io/otel"
|
|
"go.opentelemetry.io/otel/attribute"
|
|
"go.opentelemetry.io/otel/sdk"
|
|
"go.opentelemetry.io/otel/sdk/instrumentation"
|
|
"go.opentelemetry.io/otel/sdk/metric"
|
|
"go.opentelemetry.io/otel/sdk/metric/metricdata"
|
|
"go.opentelemetry.io/otel/sdk/metric/metricdata/metricdatatest"
|
|
semconv "go.opentelemetry.io/otel/semconv/v1.37.0"
|
|
"go.opentelemetry.io/otel/semconv/v1.37.0/otelconv"
|
|
)
|
|
|
|
type simpleTestExporter struct {
|
|
spans []ReadOnlySpan
|
|
shutdown bool
|
|
}
|
|
|
|
func (t *simpleTestExporter) ExportSpans(_ context.Context, spans []ReadOnlySpan) error {
|
|
t.spans = append(t.spans, spans...)
|
|
return nil
|
|
}
|
|
|
|
func (t *simpleTestExporter) Shutdown(ctx context.Context) error {
|
|
t.shutdown = true
|
|
select {
|
|
case <-ctx.Done():
|
|
// Ensure context deadline tests receive the expected error.
|
|
return ctx.Err()
|
|
default:
|
|
return nil
|
|
}
|
|
}
|
|
|
|
var _ SpanExporter = (*failingTestExporter)(nil)
|
|
|
|
type failingTestExporter struct {
|
|
simpleTestExporter
|
|
}
|
|
|
|
func (f *failingTestExporter) ExportSpans(ctx context.Context, spans []ReadOnlySpan) error {
|
|
_ = f.simpleTestExporter.ExportSpans(ctx, spans)
|
|
return errors.New("failed to export spans")
|
|
}
|
|
|
|
var _ SpanExporter = (*simpleTestExporter)(nil)
|
|
|
|
func TestNewSimpleSpanProcessor(t *testing.T) {
|
|
if ssp := NewSimpleSpanProcessor(&simpleTestExporter{}); ssp == nil {
|
|
t.Error("failed to create new SimpleSpanProcessor")
|
|
}
|
|
}
|
|
|
|
func TestNewSimpleSpanProcessorWithNilExporter(t *testing.T) {
|
|
if ssp := NewSimpleSpanProcessor(nil); ssp == nil {
|
|
t.Error("failed to create new SimpleSpanProcessor with nil exporter")
|
|
}
|
|
}
|
|
|
|
func TestSimpleSpanProcessorOnEnd(t *testing.T) {
|
|
tp := basicTracerProvider(t)
|
|
te := simpleTestExporter{}
|
|
ssp := NewSimpleSpanProcessor(&te)
|
|
|
|
tp.RegisterSpanProcessor(ssp)
|
|
startSpan(tp, "TestSimpleSpanProcessorOnEnd").End()
|
|
|
|
wantTraceID := tid
|
|
gotTraceID := te.spans[0].SpanContext().TraceID()
|
|
if wantTraceID != gotTraceID {
|
|
t.Errorf("SimplerSpanProcessor OnEnd() check: got %+v, want %+v\n", gotTraceID, wantTraceID)
|
|
}
|
|
}
|
|
|
|
func TestSimpleSpanProcessorShutdown(t *testing.T) {
|
|
exporter := &simpleTestExporter{}
|
|
ssp := NewSimpleSpanProcessor(exporter)
|
|
|
|
// Ensure we can export a span before we test we cannot after shutdown.
|
|
tp := basicTracerProvider(t)
|
|
tp.RegisterSpanProcessor(ssp)
|
|
startSpan(tp, "TestSimpleSpanProcessorShutdown").End()
|
|
nExported := len(exporter.spans)
|
|
if nExported != 1 {
|
|
t.Error("failed to verify span export")
|
|
}
|
|
|
|
if err := ssp.Shutdown(t.Context()); err != nil {
|
|
t.Errorf("shutting the SimpleSpanProcessor down: %v", err)
|
|
}
|
|
if !exporter.shutdown {
|
|
t.Error("SimpleSpanProcessor.Shutdown did not shut down exporter")
|
|
}
|
|
|
|
startSpan(tp, "TestSimpleSpanProcessorShutdown").End()
|
|
if len(exporter.spans) > nExported {
|
|
t.Error("exported span to shutdown exporter")
|
|
}
|
|
}
|
|
|
|
func TestSimpleSpanProcessorShutdownOnEndConcurrentSafe(t *testing.T) {
|
|
exporter := &simpleTestExporter{}
|
|
ssp := NewSimpleSpanProcessor(exporter)
|
|
tp := basicTracerProvider(t)
|
|
tp.RegisterSpanProcessor(ssp)
|
|
|
|
stop := make(chan struct{})
|
|
done := make(chan struct{})
|
|
go func() {
|
|
defer func() {
|
|
done <- struct{}{}
|
|
}()
|
|
for {
|
|
select {
|
|
case <-stop:
|
|
return
|
|
default:
|
|
startSpan(tp, "TestSimpleSpanProcessorShutdownOnEndConcurrentSafe").End()
|
|
}
|
|
}
|
|
}()
|
|
|
|
if err := ssp.Shutdown(t.Context()); err != nil {
|
|
t.Errorf("shutting the SimpleSpanProcessor down: %v", err)
|
|
}
|
|
if !exporter.shutdown {
|
|
t.Error("SimpleSpanProcessor.Shutdown did not shut down exporter")
|
|
}
|
|
|
|
stop <- struct{}{}
|
|
<-done
|
|
}
|
|
|
|
func TestSimpleSpanProcessorShutdownOnEndConcurrentSafe2(t *testing.T) {
|
|
exporter := &simpleTestExporter{}
|
|
ssp := NewSimpleSpanProcessor(exporter)
|
|
tp := basicTracerProvider(t)
|
|
tp.RegisterSpanProcessor(ssp)
|
|
|
|
var wg sync.WaitGroup
|
|
wg.Add(2)
|
|
|
|
span := func(spanName string) {
|
|
assert.NotPanics(t, func() {
|
|
defer wg.Done()
|
|
_, span := tp.Tracer("test").Start(t.Context(), spanName)
|
|
span.End()
|
|
})
|
|
}
|
|
|
|
go span("test-span-1")
|
|
go span("test-span-2")
|
|
|
|
wg.Wait()
|
|
|
|
assert.NoError(t, ssp.Shutdown(t.Context()))
|
|
assert.True(t, exporter.shutdown, "exporter shutdown")
|
|
}
|
|
|
|
func TestSimpleSpanProcessorShutdownHonorsContextDeadline(t *testing.T) {
|
|
ctx, cancel := context.WithTimeout(t.Context(), time.Nanosecond)
|
|
defer cancel()
|
|
<-ctx.Done()
|
|
|
|
ssp := NewSimpleSpanProcessor(&simpleTestExporter{})
|
|
if got, want := ssp.Shutdown(ctx), context.DeadlineExceeded; !errors.Is(got, want) {
|
|
t.Errorf("SimpleSpanProcessor.Shutdown did not return %v, got %v", want, got)
|
|
}
|
|
}
|
|
|
|
func TestSimpleSpanProcessorShutdownHonorsContextCancel(t *testing.T) {
|
|
ctx, cancel := context.WithCancel(t.Context())
|
|
cancel()
|
|
|
|
ssp := NewSimpleSpanProcessor(&simpleTestExporter{})
|
|
if got, want := ssp.Shutdown(ctx), context.Canceled; !errors.Is(got, want) {
|
|
t.Errorf("SimpleSpanProcessor.Shutdown did not return %v, got %v", want, got)
|
|
}
|
|
}
|
|
|
|
func TestSimpleSpanProcessorObservability(t *testing.T) {
|
|
tests := []struct {
|
|
name string
|
|
enabled bool
|
|
exporter SpanExporter
|
|
assertMetrics func(t *testing.T, rm metricdata.ResourceMetrics)
|
|
}{
|
|
{
|
|
name: "Disabled",
|
|
enabled: false,
|
|
exporter: &simpleTestExporter{},
|
|
assertMetrics: func(t *testing.T, rm metricdata.ResourceMetrics) {
|
|
assert.Empty(t, rm.ScopeMetrics)
|
|
},
|
|
},
|
|
{
|
|
name: "Enabled",
|
|
enabled: true,
|
|
exporter: &simpleTestExporter{},
|
|
assertMetrics: func(t *testing.T, rm metricdata.ResourceMetrics) {
|
|
assert.Len(t, rm.ScopeMetrics, 1)
|
|
sm := rm.ScopeMetrics[0]
|
|
|
|
want := metricdata.ScopeMetrics{
|
|
Scope: instrumentation.Scope{
|
|
Name: "go.opentelemetry.io/otel/sdk/trace/internal/observ",
|
|
Version: sdk.Version(),
|
|
SchemaURL: semconv.SchemaURL,
|
|
},
|
|
Metrics: []metricdata.Metrics{
|
|
{
|
|
Name: otelconv.SDKProcessorSpanProcessed{}.Name(),
|
|
Description: otelconv.SDKProcessorSpanProcessed{}.Description(),
|
|
Unit: otelconv.SDKProcessorSpanProcessed{}.Unit(),
|
|
Data: metricdata.Sum[int64]{
|
|
DataPoints: []metricdata.DataPoint[int64]{
|
|
{
|
|
Value: 1,
|
|
Attributes: attribute.NewSet(
|
|
semconv.OTelComponentName("simple_span_processor/0"),
|
|
semconv.OTelComponentTypeKey.String("simple_span_processor"),
|
|
),
|
|
},
|
|
},
|
|
Temporality: metricdata.CumulativeTemporality,
|
|
IsMonotonic: true,
|
|
},
|
|
},
|
|
},
|
|
}
|
|
|
|
metricdatatest.AssertEqual(
|
|
t,
|
|
want,
|
|
sm,
|
|
metricdatatest.IgnoreTimestamp(),
|
|
metricdatatest.IgnoreExemplars(),
|
|
)
|
|
},
|
|
},
|
|
{
|
|
name: "Enabled, Exporter error",
|
|
enabled: true,
|
|
exporter: &failingTestExporter{
|
|
simpleTestExporter: simpleTestExporter{},
|
|
},
|
|
assertMetrics: func(t *testing.T, rm metricdata.ResourceMetrics) {
|
|
assert.Len(t, rm.ScopeMetrics, 1)
|
|
sm := rm.ScopeMetrics[0]
|
|
|
|
want := metricdata.ScopeMetrics{
|
|
Scope: instrumentation.Scope{
|
|
Name: "go.opentelemetry.io/otel/sdk/trace/internal/observ",
|
|
Version: sdk.Version(),
|
|
SchemaURL: semconv.SchemaURL,
|
|
},
|
|
Metrics: []metricdata.Metrics{
|
|
{
|
|
Name: otelconv.SDKProcessorSpanProcessed{}.Name(),
|
|
Description: otelconv.SDKProcessorSpanProcessed{}.Description(),
|
|
Unit: otelconv.SDKProcessorSpanProcessed{}.Unit(),
|
|
Data: metricdata.Sum[int64]{
|
|
DataPoints: []metricdata.DataPoint[int64]{
|
|
{
|
|
Value: 1,
|
|
Attributes: attribute.NewSet(
|
|
semconv.OTelComponentName("simple_span_processor/0"),
|
|
semconv.OTelComponentTypeKey.String("simple_span_processor"),
|
|
semconv.ErrorTypeKey.String("*errors.errorString"),
|
|
),
|
|
},
|
|
},
|
|
Temporality: metricdata.CumulativeTemporality,
|
|
IsMonotonic: true,
|
|
},
|
|
},
|
|
},
|
|
}
|
|
|
|
metricdatatest.AssertEqual(
|
|
t,
|
|
want,
|
|
sm,
|
|
metricdatatest.IgnoreTimestamp(),
|
|
metricdatatest.IgnoreExemplars(),
|
|
)
|
|
},
|
|
},
|
|
}
|
|
|
|
for _, test := range tests {
|
|
t.Run(test.name, func(t *testing.T) {
|
|
t.Setenv("OTEL_GO_X_OBSERVABILITY", strconv.FormatBool(test.enabled))
|
|
|
|
original := otel.GetMeterProvider()
|
|
t.Cleanup(func() { otel.SetMeterProvider(original) })
|
|
|
|
r := metric.NewManualReader()
|
|
mp := metric.NewMeterProvider(
|
|
metric.WithReader(r),
|
|
metric.WithView(dropSpanMetricsView),
|
|
)
|
|
otel.SetMeterProvider(mp)
|
|
|
|
ssp := NewSimpleSpanProcessor(test.exporter)
|
|
tp := basicTracerProvider(t)
|
|
tp.RegisterSpanProcessor(ssp)
|
|
startSpan(tp, test.name).End()
|
|
|
|
var rm metricdata.ResourceMetrics
|
|
require.NoError(t, r.Collect(t.Context(), &rm))
|
|
test.assertMetrics(t, rm)
|
|
simpleProcessorIDCounter.Store(0) // reset simpleProcessorIDCounter
|
|
})
|
|
}
|
|
}
|