[receiver/kafkareceiver] move receiver.kafkareceiver.UseFranzGo feature gate to Beta

.chloggen/receiver_kafka_franz_go_beta.yaml

@@ -0,0 +1,27 @@
+# Use this changelog template to create an entry for release notes.
+
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: "enhancement"
+
+# The name of the component, or a single word describing the area of concern, (e.g. filelogreceiver)
+component: receiver/kafkareceiver
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: "Use franz-go client for Kafka receiver as default, promoting the receiver.kafkareceiver.UseFranzGo feature gate to Beta."
+
+# Mandatory: One or more tracking issues related to the change. You can use the PR number here if no issue exists.
+issues: [42155]
+
+# (Optional) One or more lines of additional information to render under the primary note.
+# These lines will be padded with 2 spaces and then inserted directly into the document.
+# Use pipe (|) for multiline entries.
+subtext:
+
+# If your change doesn't affect end users or the exported elements of any package,
+# you should instead start your pull request title with [chore] or use the "Skip Changelog" label.
+# Optional: The change log or logs in which this entry should be included.
+# e.g. '[user]' or '[user, api]'
+# Include 'user' if the change is relevant to end users.
+# Include 'api' if there is a change to a library API.
+# Default: '[user]'
+change_logs: [user]

receiver/kafkareceiver/README.md

@@ -23,7 +23,7 @@ If used in conjunction with the `kafkaexporter` configured with `include_metadat
 ## Getting Started
 
 > [!NOTE]
-> You can opt-in to use [`franz-go`](https://github.com/twmb/franz-go) client by enabling the feature gate
+> You can opt out of using the [`franz-go`](https://github.com/twmb/franz-go) client by disabling the feature gate
 > `receiver.kafkareceiver.UseFranzGo` when you run the OpenTelemetry Collector. See the following page
 > for more details: [Feature Gates](https://github.com/open-telemetry/opentelemetry-collector/tree/main/featuregate#controlling-gates)
 >

receiver/kafkareceiver/consumer_franz.go

@@ -14,6 +14,7 @@ import (
 	"github.com/cenkalti/backoff/v4"
 	"github.com/twmb/franz-go/pkg/kgo"
 	"go.opentelemetry.io/collector/component"
+	"go.opentelemetry.io/collector/component/componentstatus"
 	"go.opentelemetry.io/collector/consumer/consumererror"
 	"go.opentelemetry.io/collector/featuregate"
 	"go.opentelemetry.io/collector/receiver"
@@ -34,7 +35,7 @@ const franzGoConsumerFeatureGateName = "receiver.kafkareceiver.UseFranzGo"
 // the Kafka receiver will use the franz-go client, which is more performant and has
 // better support for modern Kafka features.
 var franzGoConsumerFeatureGate = featuregate.GlobalRegistry().MustRegister(
-	franzGoConsumerFeatureGateName, featuregate.StageAlpha,
+	franzGoConsumerFeatureGateName, featuregate.StageBeta,
 	featuregate.WithRegisterDescription("When enabled, the Kafka receiver will use the franz-go client to consume messages."),
 	featuregate.WithRegisterFromVersion("v0.129.0"),
 )
@@ -63,6 +64,11 @@ type franzConsumer struct {
 	client      *kgo.Client
 	obsrecv     *receiverhelper.ObsReport
 	assignments map[topicPartition]*pc
+
+	// ---- status reporting (parity with Sarama) ----
+	host         component.Host
+	stoppingOnce sync.Once
+	stoppedOnce  sync.Once
 }
 
 // pc represents the partition consumer shared information.
@@ -137,6 +143,22 @@ func newFranzKafkaConsumer(
 	}, nil
 }
 
+// reportStatus emits a component status event if we have a host.
+func (c *franzConsumer) reportStatus(s componentstatus.Status) {
+	if c.host == nil {
+		return
+	}
+	componentstatus.ReportStatus(c.host, componentstatus.NewEvent(s))
+}
+
+// reportRecoverable reports a recoverable error status event.
+func (c *franzConsumer) reportRecoverable(err error) {
+	if c.host == nil || err == nil {
+		return
+	}
+	componentstatus.ReportStatus(c.host, componentstatus.NewRecoverableErrorEvent(err))
+}
+
 func (c *franzConsumer) Start(ctx context.Context, host component.Host) error {
 	c.mu.Lock()
 	defer c.mu.Unlock()
@@ -149,6 +171,10 @@ func (c *franzConsumer) Start(ctx context.Context, host component.Host) error {
 		close(c.started)
 	}
 
+	// Parity with Sarama: report "Starting" as soon as Start() is called.
+	c.host = host
+	c.reportStatus(componentstatus.StatusStarting)
+
 	obsrecv, err := receiverhelper.NewObsReport(receiverhelper.ObsReportSettings{
 		ReceiverID:             c.settings.ID,
 		Transport:              transport,
@@ -205,7 +231,11 @@ func (c *franzConsumer) Start(ctx context.Context, host component.Host) error {
 }
 
 func (c *franzConsumer) consumeLoop(ctx context.Context) {
-	defer close(c.consumerClosed)
+	// Parity with Sarama: when the loop exits, report Stopped.
+	defer func() {
+		c.stoppedOnce.Do(func() { c.reportStatus(componentstatus.StatusStopped) })
+		close(c.consumerClosed)
+	}()
 
 	for {
 		// Consume messages until the ctx is cancelled (the client is closed).
@@ -237,6 +267,8 @@ func (c *franzConsumer) consume(ctx context.Context, size int) bool {
 			zap.String("topic", topic),
 			zap.Int64("partition", int64(partition)),
 		)
+		// Parity with Sarama: report recoverable error while consuming.
+		c.reportRecoverable(err)
 		if !hasError {
 			hasError = true
 		}
@@ -354,21 +386,31 @@ func (c *franzConsumer) consume(ctx context.Context, size int) bool {
 	if !c.config.AutoCommit.Enable {
 		if err := c.client.CommitMarkedOffsets(ctx); err != nil {
 			c.settings.Logger.Error("failed to commit offsets", zap.Error(err))
+			// Surface as recoverable error (parity with Sarama’s loop error reporting).
+			c.reportRecoverable(err)
 		}
 	}
 	return true
 }
 
 func (c *franzConsumer) Shutdown(ctx context.Context) error {
+	// Parity with Sarama: report Stopping at shutdown start.
+	c.stoppingOnce.Do(func() { c.reportStatus(componentstatus.StatusStopping) })
+
 	if !c.triggerShutdown() {
-		return errors.New("kafka consumer: consumer isn't running")
+		// Idempotent: never fail if not started.
+		// We still want to ensure Stopped is eventually emitted (consumeLoop defer handles it).
+		// However, if the loop was never started, emit Stopped here too.
+		c.stoppedOnce.Do(func() { c.reportStatus(componentstatus.StatusStopped) })
+		return nil
 	}
 
 	select {
 	case <-ctx.Done():
 		return context.Cause(ctx)
 	case <-c.consumerClosed:
 	}
+
 	return nil
 }
 
@@ -387,18 +429,24 @@ func (c *franzConsumer) triggerShutdown() bool {
 		return true
 	default:
 		close(c.closing)
+		client := c.client
 		c.mu.Unlock()
 		// Close the client without holding the write mutex, otherwise, the
 		// Shutdown will deadlock when `franzConsumer` inevitably calls the
 		// lost/assigned callback.
-		c.client.Close()
+		if client != nil {
+			client.Close()
+		}
 	}
 	return true
 }
 
 // assigned must be set as kgo.OnPartitionsAssigned callback. Ensuring all
 // assigned partitions to this consumer process received records.
 func (c *franzConsumer) assigned(ctx context.Context, _ *kgo.Client, assigned map[string][]int32) {
+	// Report OK on each successful assignment so we can recover status after transient errors.
+	c.reportStatus(componentstatus.StatusOK)
+
 	c.mu.Lock()
 	defer c.mu.Unlock()
 	for topic, partitions := range assigned {
@@ -468,6 +516,8 @@ func (c *franzConsumer) lost(ctx context.Context, _ *kgo.Client,
 	// away from this consumer.
 	if err := c.client.CommitMarkedOffsets(ctx); err != nil {
 		c.settings.Logger.Error("failed to commit marked offsets", zap.Error(err))
+		// Parity with Sarama: report recoverable error on commit errors.
+		c.reportRecoverable(err)
 	}
 }
 

receiver/kafkareceiver/consumer_franz_test.go

@@ -199,6 +199,13 @@ func TestConsumerShutdownConsuming(t *testing.T) {
 			consumer, e := newFranzKafkaConsumer(cfg, settings, []string{topic}, consumeFn)
 			require.NoError(tb, e)
 			require.NoError(tb, consumer.Start(ctx, componenttest.NewNopHost()))
+			// Wait until the group has assigned at least one partition.
+			require.Eventually(tb, func() bool {
+				consumer.mu.RLock()
+				n := len(consumer.assignments)
+				consumer.mu.RUnlock()
+				return n > 0
+			}, 10*time.Second, 10*time.Millisecond)
 			require.NoError(tb, kafkaClient.ProduceSync(ctx, rs...).FirstErr())
 
 			select {
@@ -239,9 +246,12 @@ func TestConsumerShutdownNotStarted(t *testing.T) {
 	require.NoError(t, err)
 
 	for range 2 {
-		require.EqualError(t, c.Shutdown(t.Context()),
-			"kafka consumer: consumer isn't running")
+		require.NoError(t, c.Shutdown(t.Context()))
 	}
+
+	// Verify internal signal that there's nothing to shut down.
+	// (Same package, so we can call the unexported helper.)
+	require.False(t, c.triggerShutdown(), "triggerShutdown should indicate no-op when never started")
 }
 
 // TestRaceLostVsConsume verifies no data race occurs between concurrent