Use the cause of the context error in OTLP retry

CHANGELOG.md

@@ -34,6 +34,7 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 - The semantic conventions have been upgraded from `v1.26.0` to `v1.34.0` in `go.opentelemetry.io/otel/sdk/trace`. (#6835)
 - The semantic conventions have been upgraded from `v1.26.0` to `v1.34.0` in `go.opentelemetry.io/otel/trace`. (#6836)
 - `Record.Resource` now returns `*resource.Resource` instead of `resource.Resource` in `go.opentelemetry.io/otel/sdk/log`. (#6864)
+- Retry now shows error cause for context timeout in `go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc`, `go.opentelemetry.io/otel/exporters/otlp/otlpmetric/otlpmetricgrpc`, `go.opentelemetry.io/otel/exporters/otlp/otlplog/otlploggrpc`, `go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp`, `go.opentelemetry.io/otel/exporters/otlp/otlpmetric/otlpmetrichttp`, `go.opentelemetry.io/otel/exporters/otlp/otlplog/otlploghttp`. (#6898)
 
 ### Fixed
 

exporters/otlp/otlplog/otlploggrpc/client.go

@@ -5,6 +5,7 @@ package otlploggrpc // import "go.opentelemetry.io/otel/exporters/otlp/otlplog/o
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"time"
 
@@ -192,7 +193,7 @@ func (c *client) exportContext(parent context.Context) (context.Context, context
 	)
 
 	if c.exportTimeout > 0 {
-		ctx, cancel = context.WithTimeout(parent, c.exportTimeout)
+		ctx, cancel = context.WithTimeoutCause(parent, c.exportTimeout, errors.New("exporter export timeout"))
 	} else {
 		ctx, cancel = context.WithCancel(parent)
 	}
@@ -228,6 +229,8 @@ func retryable(err error) (bool, time.Duration) {
 
 func retryableGRPCStatus(s *status.Status) (bool, time.Duration) {
 	switch s.Code() {
+	// Follows the retryable error codes defined in
+	// https://opentelemetry.io/docs/specs/otlp/#failures
 	case codes.Canceled,
 		codes.DeadlineExceeded,
 		codes.Aborted,

exporters/otlp/otlpmetric/otlpmetricgrpc/client.go

@@ -5,6 +5,7 @@ package otlpmetricgrpc // import "go.opentelemetry.io/otel/exporters/otlp/otlpme
 
 import (
 	"context"
+	"errors"
 	"time"
 
 	"google.golang.org/genproto/googleapis/rpc/errdetails"
@@ -149,7 +150,7 @@ func (c *client) exportContext(parent context.Context) (context.Context, context
 	)
 
 	if c.exportTimeout > 0 {
-		ctx, cancel = context.WithTimeout(parent, c.exportTimeout)
+		ctx, cancel = context.WithTimeoutCause(parent, c.exportTimeout, errors.New("exporter export timeout"))
 	} else {
 		ctx, cancel = context.WithCancel(parent)
 	}

exporters/otlp/otlptrace/otlptracegrpc/client.go

@@ -223,7 +223,7 @@ func (c *client) exportContext(parent context.Context) (context.Context, context
 	)
 
 	if c.exportTimeout > 0 {
-		ctx, cancel = context.WithTimeout(parent, c.exportTimeout)
+		ctx, cancel = context.WithTimeoutCause(parent, c.exportTimeout, errors.New("exporter export timeout"))
 	} else {
 		ctx, cancel = context.WithCancel(parent)
 	}

internal/shared/otlp/retry/retry.go.tmpl

@@ -132,7 +132,7 @@ func wait(ctx context.Context, delay time.Duration) error {
 		select {
 		case <-timer.C:
 		default:
-			return ctx.Err()
+			return context.Cause(ctx)
 		}
 	case <-timer.C:
 	}

sdk/log/exporter.go

@@ -119,7 +119,9 @@ func newTimeoutExporter(exp Exporter, timeout time.Duration) Exporter {
 
 // Export sets the timeout of ctx before calling the Exporter e wraps.
 func (e *timeoutExporter) Export(ctx context.Context, records []Record) error {
-	ctx, cancel := context.WithTimeout(ctx, e.timeout)
+	// This only used by the batch processor, and it takes processor timeout config.
+	// Thus, the error message points to the processor. So users know they should adjust the processor timeout.
+	ctx, cancel := context.WithTimeoutCause(ctx, e.timeout, errors.New("processor export timeout"))
 	defer cancel()
 	return e.Exporter.Export(ctx, records)
 }

sdk/metric/periodic_reader.go

@@ -202,7 +202,7 @@ func (r *PeriodicReader) aggregation(
 // collectAndExport gather all metric data related to the periodicReader r from
 // the SDK and exports it with r's exporter.
 func (r *PeriodicReader) collectAndExport(ctx context.Context) error {
-	ctx, cancel := context.WithTimeout(ctx, r.timeout)
+	ctx, cancel := context.WithTimeoutCause(ctx, r.timeout, errors.New("reader collect and export timeout"))
 	defer cancel()
 
 	// TODO (#3047): Use a sync.Pool or persistent pointer instead of allocating rm every Collect.
@@ -278,7 +278,7 @@ func (r *PeriodicReader) ForceFlush(ctx context.Context) error {
 	// Prioritize the ctx timeout if it is set.
 	if _, ok := ctx.Deadline(); !ok {
 		var cancel context.CancelFunc
-		ctx, cancel = context.WithTimeout(ctx, r.timeout)
+		ctx, cancel = context.WithTimeoutCause(ctx, r.timeout, errors.New("reader force flush timeout"))
 		defer cancel()
 	}
 
@@ -311,7 +311,7 @@ func (r *PeriodicReader) Shutdown(ctx context.Context) error {
 		// Prioritize the ctx timeout if it is set.
 		if _, ok := ctx.Deadline(); !ok {
 			var cancel context.CancelFunc
-			ctx, cancel = context.WithTimeout(ctx, r.timeout)
+			ctx, cancel = context.WithTimeoutCause(ctx, r.timeout, errors.New("reader shutdown timeout"))
 			defer cancel()
 		}
 

sdk/trace/batch_span_processor.go

@@ -5,6 +5,7 @@ package trace // import "go.opentelemetry.io/otel/sdk/trace"
 
 import (
 	"context"
+	"errors"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -267,7 +268,7 @@ func (bsp *batchSpanProcessor) exportSpans(ctx context.Context) error {
 
 	if bsp.o.ExportTimeout > 0 {
 		var cancel context.CancelFunc
-		ctx, cancel = context.WithTimeout(ctx, bsp.o.ExportTimeout)
+		ctx, cancel = context.WithTimeoutCause(ctx, bsp.o.ExportTimeout, errors.New("processor export timeout"))
 		defer cancel()
 	}