[processor/tailsampling] Gate OTTL condition error mode default

processor/tailsamplingprocessor/documentation.md

@@ -169,5 +169,6 @@ This component has the following feature gates:
 | `processor.tailsamplingprocessor.metricstatcountspanssampled` | alpha | When enabled, a new metric stat_count_spans_sampled will be available in the tail sampling processor. Differently from stat_count_traces_sampled, this metric will count the number of spans sampled or not per sampling policy, where the original counts traces. | v0.95.0 | N/A | [Link](https://github.com/open-telemetry/opentelemetry-collector-contrib/issues/30482) |
 | `processor.tailsamplingprocessor.recordpolicy` | alpha | When enabled, attaches the name of the policy (and if applicable, composite policy) responsible for sampling a trace in the 'tailsampling.policy', 'tailsampling.composite_policy', and `tailsampling.cached_decision` attributes. | v0.120.0 | N/A | [Link](https://github.com/open-telemetry/opentelemetry-collector-contrib/issues/35180) |
 | `processor.tailsamplingprocessor.tailstorageextension` | alpha | When enabled, allows configuring tail_storage to use a tail storage extension implementation. | v0.150.0 | N/A | [Link](https://github.com/open-telemetry/opentelemetry-collector-contrib/pull/47331) |
+| `processor.tailsamplingprocessor.defaultottlconditionerrormodeignore` | alpha | When enabled, OTTL condition policies without an explicit error_mode use ignore instead of propagate. | v0.152.0 | N/A | [Link](https://github.com/open-telemetry/opentelemetry-collector-contrib/issues/48420) |
 
 For more information about feature gates, see the [Feature Gates](https://github.com/open-telemetry/opentelemetry-collector/blob/main/featuregate/README.md) documentation.

processor/tailsamplingprocessor/metadata.yaml

@@ -37,6 +37,12 @@ feature_gates:
     from_version: v0.150.0
     reference_url: https://github.com/open-telemetry/opentelemetry-collector-contrib/pull/47331
 
+  - id: "processor.tailsamplingprocessor.defaultottlconditionerrormodeignore"
+    description: "When enabled, OTTL condition policies without an explicit error_mode use ignore instead of propagate."
+    stage: alpha
+    from_version: v0.152.0
+    reference_url: https://github.com/open-telemetry/opentelemetry-collector-contrib/issues/48420
+
 tests:
   config:
 

processor/tailsamplingprocessor/processor.go

@@ -24,6 +24,7 @@ import (
 	"go.opentelemetry.io/otel/trace"
 	"go.uber.org/zap"
 
+	"github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/processor/tailsamplingprocessor/cache"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/processor/tailsamplingprocessor/internal/idbatcher"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/processor/tailsamplingprocessor/internal/metadata"
@@ -450,7 +451,7 @@ func getSharedPolicyEvaluator(settings component.TelemetrySettings, cfg *sharedP
 		return sampling.NewBooleanAttributeFilter(settings, bafCfg.Key, bafCfg.Value, bafCfg.InvertMatch), nil
 	case OTTLCondition:
 		ottlfCfg := cfg.OTTLConditionCfg
-		return sampling.NewOTTLConditionFilter(settings, ottlfCfg.SpanConditions, ottlfCfg.SpanEventConditions, ottlfCfg.ErrorMode)
+		return sampling.NewOTTLConditionFilter(settings, ottlfCfg.SpanConditions, ottlfCfg.SpanEventConditions, defaultOTTLConditionErrorMode(ottlfCfg.ErrorMode))
 	case TraceFlags:
 		return sampling.NewTraceFlags(settings), nil
 	default:
@@ -468,6 +469,16 @@ func getSharedPolicyEvaluator(settings component.TelemetrySettings, cfg *sharedP
 	}
 }
 
+func defaultOTTLConditionErrorMode(configured ottl.ErrorMode) ottl.ErrorMode {
+	if configured != "" {
+		return configured
+	}
+	if metadata.ProcessorTailsamplingprocessorDefaultottlconditionerrormodeignoreFeatureGate.IsEnabled() {
+		return ottl.IgnoreError
+	}
+	return ottl.PropagateError
+}
+
 type policyDecisionMetrics struct {
 	tracesSampled int
 	spansSampled  int64

processor/tailsamplingprocessor/processor_test.go

@@ -30,6 +30,7 @@ import (
 	"go.uber.org/zap"
 	"go.uber.org/zap/zaptest/observer"
 
+	"github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/processor/tailsamplingprocessor/internal/idbatcher"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/processor/tailsamplingprocessor/internal/metadata"
 	"github.com/open-telemetry/opentelemetry-collector-contrib/processor/tailsamplingprocessor/internal/tailstorageextension"
@@ -71,6 +72,23 @@ func (t *TestPolicyEvaluator) IsStateful() bool {
 	return t.pe.IsStateful()
 }
 
+func TestDefaultOTTLConditionErrorMode(t *testing.T) {
+	gate := metadata.ProcessorTailsamplingprocessorDefaultottlconditionerrormodeignoreFeatureGate
+	originalValue := gate.IsEnabled()
+	require.NoError(t, featuregate.GlobalRegistry().Set(gate.ID(), false))
+	t.Cleanup(func() {
+		require.NoError(t, featuregate.GlobalRegistry().Set(gate.ID(), originalValue))
+	})
+
+	require.Equal(t, ottl.PropagateError, defaultOTTLConditionErrorMode(""))
+	require.Equal(t, ottl.IgnoreError, defaultOTTLConditionErrorMode(ottl.IgnoreError))
+	require.Equal(t, ottl.PropagateError, defaultOTTLConditionErrorMode(ottl.PropagateError))
+
+	require.NoError(t, featuregate.GlobalRegistry().Set(gate.ID(), true))
+	require.Equal(t, ottl.IgnoreError, defaultOTTLConditionErrorMode(""))
+	require.Equal(t, ottl.PropagateError, defaultOTTLConditionErrorMode(ottl.PropagateError))
+}
+
 // testTSPController is a set of mechanisms to make the TSP do predictable
 // things in tests.
 type testTSPController struct {