Merge branch 'main' into nv/4189-3

* fix: show warning for non-existent cost meter metrics

* chore: lint fix by removing unused list

* chore: py fmt add new line

* chore: missing newline between tests

* fix: no warnings or errors for internal metrics

* fix: pylint fix by adding new line

* fix: lint fix in test

pkg/querier/config.go

@@ -54,7 +54,7 @@ func (c Config) Validate() error {
 	if c.MaxConcurrentQueries <= 0 {
 		return errors.NewInvalidInputf(errors.CodeInvalidInput, "max_concurrent_queries must be positive, got %v", c.MaxConcurrentQueries)
 	}
-	if c.SkipResourceFingerprint.Enabled && c.SkipResourceFingerprint.Threshold <= 0 {
+	if c.SkipResourceFingerprint.Enabled && c.SkipResourceFingerprint.Threshold == 0 {
 		return errors.NewInvalidInputf(errors.CodeInvalidInput, "skip_resource_fingerprint.threshold must be > 0 when enabled")
 	}
 	return nil

pkg/querier/querier.go

@@ -217,7 +217,7 @@ func (q *querier) QueryRange(ctx context.Context, orgID valuer.UUID, req *qbtype
 		}
 	}
 	preseededResults := make(map[string]any)
-	for _, name := range missingMetricQueries { // at this point missing metrics will not have any non existent metrics, only normal ones
+	for _, name := range missingMetricQueries {
 		switch req.RequestType {
 		case qbtypes.RequestTypeTimeSeries:
 			preseededResults[name] = &qbtypes.TimeSeriesData{QueryName: name}
@@ -375,11 +375,24 @@ func (q *querier) resolveMetricMetadata(ctx context.Context, queries []qbtypes.Q
 		return missingMetricQueries, "", nil
 	}
 
+	isInternalMetric := func(n string) bool { return strings.HasPrefix(n, "signoz.") || strings.HasPrefix(n, "signoz_") }
+	externalMissingMetrics := make([]string, 0, len(missingMetrics))
+	for _, m := range missingMetrics {
+		if !isInternalMetric(m) {
+			externalMissingMetrics = append(externalMissingMetrics, m)
+		}
+	}
+	if len(externalMissingMetrics) == 0 {
+		// this means all missing metrics are internal, and since internal metrics
+		// aren't user-controlled, skip errors/warnings for them since users can't act on them
+		return missingMetricQueries, "", nil
+	}
+
 	// Classify each missing metric: never-seen → NotFound error; seen-but-no-
 	// data-in-window → dormant warning.
-	lastSeenInfo, _ := q.metadataStore.FetchLastSeenInfoMulti(ctx, missingMetrics...)
+	lastSeenInfo, _ := q.metadataStore.FetchLastSeenInfoMulti(ctx, externalMissingMetrics...)
 	nonExistentMetrics := []string{}
-	for _, name := range missingMetrics {
+	for _, name := range externalMissingMetrics {
 		if ts, ok := lastSeenInfo[name]; ok && ts > 0 {
 			continue
 		}
@@ -400,11 +413,11 @@ func (q *querier) resolveMetricMetadata(ctx context.Context, queries []qbtypes.Q
 		}
 		return name
 	}
-	if len(missingMetrics) == 1 {
+	if len(externalMissingMetrics) == 1 {
 		dormantWarning = fmt.Sprintf("no data found for the metric %s in the query time range", lastSeenStr(missingMetrics[0]))
 	} else {
-		parts := make([]string, len(missingMetrics))
-		for i, m := range missingMetrics {
+		parts := make([]string, len(externalMissingMetrics))
+		for i, m := range externalMissingMetrics {
 			parts[i] = lastSeenStr(m)
 		}
 		dormantWarning = fmt.Sprintf("no data found for the following metrics in the query time range: %s", strings.Join(parts, ", "))

pkg/telemetrylogs/statement_builder.go

@@ -716,22 +716,6 @@ func aggOrderBy(k qbtypes.OrderBy, q qbtypes.QueryBuilderQuery[qbtypes.LogAggreg
 	return 0, false
 }
 
-// maybeAttachResourceFilter decides whether to pre-filter on the resource table.
-//
-// The resource table maps resource attributes (e.g. service.name) to fingerprints.
-// When it helps, we look up the matching fingerprints there first and feed them to the
-// main query as a CTE, so the main table only scans those fingerprints.
-//
-// There are three outcomes:
-//
-//  1. Skip it — the filter has nothing we can pre-resolve. This happens when the filter
-//     is empty,no resource filter, or when its resource conditions sit under an OR (e.g.
-//     `name='GET' OR service.name='abc'`), because then we can't reduce to a fixed set
-//     of fingerprints. The main query filters on resource attributes inline instead.
-//  2. Skip it — too many fingerprints match (over the configured threshold), so the CTE
-//     would not be selective enough to be worth it. Again, filter inline on the main table.
-//  3. Use it — attach the matching fingerprints as the __resource_filter CTE and join
-//     the main table on resource_fingerprint.
 func (b *logQueryStatementBuilder) maybeAttachResourceFilter(
 	ctx context.Context,
 	sb *sqlbuilder.SelectBuilder,

pkg/telemetrylogs/stmt_builder_test.go

@@ -1230,7 +1230,7 @@ func TestSkipResourceFingerprintLogs(t *testing.T) {
 		mockStore := telemetrystoretest.New(telemetrystore.Config{}, &regexQueryMatcher{})
 		mock := mockStore.Mock()
 
-		mock.ExpectQueryRow(`SELECT uniq\(fingerprint\) FROM signoz_logs\.distributed_logs_v2_resource`).
+		mock.ExpectQueryRow(`SELECT count\(\) FROM \(SELECT fingerprint FROM signoz_logs\.distributed_logs_v2_resource`).
 			WillReturnRow(cmock.NewRow([]cmock.ColumnType{
 				{Name: "count", Type: "UInt64"},
 			}, []any{uint64(2)}))
@@ -1250,7 +1250,7 @@ func TestSkipResourceFingerprintLogs(t *testing.T) {
 		mockStore := telemetrystoretest.New(telemetrystore.Config{}, &regexQueryMatcher{})
 		mock := mockStore.Mock()
 
-		mock.ExpectQueryRow(`SELECT uniq\(fingerprint\) FROM signoz_logs\.distributed_logs_v2_resource`).
+		mock.ExpectQueryRow(`SELECT count\(\) FROM \(SELECT fingerprint FROM signoz_logs\.distributed_logs_v2_resource`).
 			WillReturnRow(cmock.NewRow([]cmock.ColumnType{
 				{Name: "count", Type: "UInt64"},
 			}, []any{threshold}))

pkg/telemetrymetrics/statement_builder.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"log/slog"
 
+	"github.com/SigNoz/signoz/pkg/errors"
 	"github.com/SigNoz/signoz/pkg/factory"
 	"github.com/SigNoz/signoz/pkg/flagger"
 	"github.com/SigNoz/signoz/pkg/querybuilder"
@@ -118,6 +119,16 @@ func (b *MetricQueryStatementBuilder) buildPipelineStatement(
 		cteArgs      [][]any
 	)
 
+	if query.Aggregations[0].SpaceAggregation.IsPercentile() && !query.Aggregations[0].Type.IsPercentileSpaceAggregationAllowed() {
+		return nil, errors.Newf(
+			errors.TypeInvalidInput,
+			errors.CodeInvalidInput,
+			"invalid space aggregation `%s` for metric type `%s`, percentile space aggregations are only supported for [`histogram`, `exponentialhistogram`, `summary`] metric types",
+			query.Aggregations[0].SpaceAggregation.StringValue(),
+			query.Aggregations[0].Type.StringValue(),
+		)
+	}
+
 	origSpaceAgg := query.Aggregations[0].SpaceAggregation
 	origTimeAgg := query.Aggregations[0].TimeAggregation
 	origGroupBy := slices.Clone(query.GroupBy)

pkg/telemetrymetrics/stmt_builder_test.go

@@ -5,6 +5,7 @@ import (
 	"testing"
 	"time"
 
+	"github.com/SigNoz/signoz/pkg/errors"
 	"github.com/SigNoz/signoz/pkg/flagger"
 	"github.com/SigNoz/signoz/pkg/instrumentation/instrumentationtest"
 	"github.com/SigNoz/signoz/pkg/types/metrictypes"
@@ -217,6 +218,25 @@ func TestStatementBuilder(t *testing.T) {
 			},
 			expectedErr: nil,
 		},
+		{
+			name:        "test_percentile_space_aggregation_on_non_histogram_type",
+			requestType: qbtypes.RequestTypeTimeSeries,
+			query: qbtypes.QueryBuilderQuery[qbtypes.MetricAggregation]{
+				Signal:       telemetrytypes.SignalMetrics,
+				StepInterval: qbtypes.Step{Duration: 30 * time.Second},
+				Aggregations: []qbtypes.MetricAggregation{
+					{
+						MetricName:       "signoz_calls_total",
+						Type:             metrictypes.SumType,
+						Temporality:      metrictypes.Cumulative,
+						TimeAggregation:  metrictypes.TimeAggregationRate,
+						SpaceAggregation: metrictypes.SpaceAggregationPercentile95,
+					},
+				},
+				Limit: 10,
+			},
+			expectedErr: errors.New(errors.TypeInvalidInput, errors.CodeInvalidInput, "invalid space aggregation `p95` for metric type `sum`, percentile space aggregations are only supported for [`histogram`, `exponentialhistogram`, `summary`] metric types"),
+		},
 	}
 
 	fm := NewFieldMapper()

pkg/telemetryresourcefilter/statement_builder.go

@@ -98,7 +98,17 @@ func (b *resourceFilterStatementBuilder[T]) Build(
 	query qbtypes.QueryBuilderQuery[T],
 	variables map[string]qbtypes.VariableItem,
 ) (*qbtypes.Statement, error) {
-	q, isNoOp, err := b.buildQuery(ctx, start, end, "fingerprint", query, variables)
+	q := sqlbuilder.NewSelectBuilder()
+	q.Select("fingerprint")
+	q.From(fmt.Sprintf("%s.%s", b.dbName, b.tableName))
+
+	keySelectors := b.getKeySelectors(query)
+	keys, _, err := b.metadataStore.GetKeysMulti(ctx, keySelectors)
+	if err != nil {
+		return nil, err
+	}
+
+	isNoOp, err := b.addConditions(ctx, q, start, end, query, keys, variables)
 	if err != nil {
 		return nil, err
 	}
@@ -117,37 +127,8 @@ func (b *resourceFilterStatementBuilder[T]) Build(
 	}, nil
 }
 
-// buildQuery selects selectExpr from the resource table and applies the filter and
-// time conditions. isNoOp is true when the filter resolves to no resource conditions.
-func (b *resourceFilterStatementBuilder[T]) buildQuery(
-	ctx context.Context,
-	start, end uint64,
-	selectExpr string,
-	query qbtypes.QueryBuilderQuery[T],
-	variables map[string]qbtypes.VariableItem,
-) (*sqlbuilder.SelectBuilder, bool, error) {
-	q := sqlbuilder.NewSelectBuilder()
-	q.Select(selectExpr)
-	q.From(fmt.Sprintf("%s.%s", b.dbName, b.tableName))
-
-	keySelectors := b.getKeySelectors(query)
-	keys, _, err := b.metadataStore.GetKeysMulti(ctx, keySelectors)
-	if err != nil {
-		return nil, false, err
-	}
-
-	isNoOp, err := b.addConditions(ctx, q, start, end, query, keys, variables)
-	if err != nil {
-		return nil, false, err
-	}
-	return q, isNoOp, nil
-}
-
 // BuildCount returns a statement that counts the distinct fingerprints matching
 // the resource filter. Returns (nil, nil) when the filter is a no-op.
-//
-// It uses uniq() rather than count() over a GROUP BY: uniq's approximation is well
-// within tolerance for a threshold check and is ~2x faster with far less memory.
 func (b *resourceFilterStatementBuilder[T]) BuildCount(
 	ctx context.Context,
 	start uint64,
@@ -155,18 +136,13 @@ func (b *resourceFilterStatementBuilder[T]) BuildCount(
 	query qbtypes.QueryBuilderQuery[T],
 	variables map[string]qbtypes.VariableItem,
 ) (*qbtypes.Statement, error) {
-	q, isNoOp, err := b.buildQuery(ctx, start, end, "uniq(fingerprint)", query, variables)
-	if err != nil {
+	inner, err := b.Build(ctx, start, end, qbtypes.RequestTypeRaw, query, variables)
+	if err != nil || inner == nil {
 		return nil, err
 	}
-	if isNoOp {
-		return nil, nil //nolint:nilnil
-	}
-
-	stmt, args := q.BuildWithFlavor(sqlbuilder.ClickHouse)
 	return &qbtypes.Statement{
-		Query: stmt,
-		Args:  args,
+		Query: fmt.Sprintf("SELECT count() FROM (%s)", inner.Query),
+		Args:  inner.Args,
 	}, nil
 }
 

pkg/telemetrytraces/statement_builder.go

@@ -818,22 +818,6 @@ func aggOrderBy(k qbtypes.OrderBy, q qbtypes.QueryBuilderQuery[qbtypes.TraceAggr
 	return 0, false
 }
 
-// maybeAttachResourceFilter decides whether to pre-filter on the resource table.
-//
-// The resource table maps resource attributes (e.g. service.name) to fingerprints.
-// When it helps, we look up the matching fingerprints there first and feed them to the
-// main query as a CTE, so the main table only scans those fingerprints.
-//
-// There are three outcomes:
-//
-//  1. Skip it — the filter has nothing we can pre-resolve. This happens when the filter
-//     is empty,no resource filter, or when its resource conditions sit under an OR (e.g.
-//     `name='GET' OR service.name='abc'`), because then we can't reduce to a fixed set
-//     of fingerprints. The main query filters on resource attributes inline instead.
-//  2. Skip it — too many fingerprints match (over the configured threshold), so the CTE
-//     would not be selective enough to be worth it. Again, filter inline on the main table.
-//  3. Use it — attach the matching fingerprints as the __resource_filter CTE and join
-//     the main table on resource_fingerprint.
 func (b *traceQueryStatementBuilder) maybeAttachResourceFilter(
 	ctx context.Context,
 	sb *sqlbuilder.SelectBuilder,

pkg/telemetrytraces/stmt_builder_test.go

@@ -1629,7 +1629,7 @@ func TestSkipResourceFingerprint(t *testing.T) {
 
 		// Only the count query runs against the telemetry store; the CTE
 		// itself is embedded as SQL in the main query (no extra round trip).
-		mock.ExpectQueryRow(`SELECT uniq\(fingerprint\) FROM signoz_traces\.distributed_traces_v3_resource`).
+		mock.ExpectQueryRow(`SELECT count\(\) FROM \(SELECT fingerprint FROM signoz_traces\.distributed_traces_v3_resource`).
 			WillReturnRow(cmock.NewRow([]cmock.ColumnType{
 				{Name: "count", Type: "UInt64"},
 			}, []any{uint64(2)}))
@@ -1649,7 +1649,7 @@ func TestSkipResourceFingerprint(t *testing.T) {
 		mockStore := telemetrystoretest.New(telemetrystore.Config{}, &regexQueryMatcher{})
 		mock := mockStore.Mock()
 
-		mock.ExpectQueryRow(`SELECT uniq\(fingerprint\) FROM signoz_traces\.distributed_traces_v3_resource`).
+		mock.ExpectQueryRow(`SELECT count\(\) FROM \(SELECT fingerprint FROM signoz_traces\.distributed_traces_v3_resource`).
 			WillReturnRow(cmock.NewRow([]cmock.ColumnType{
 				{Name: "count", Type: "UInt64"},
 			}, []any{threshold}))

tests/integration/tests/querier/03_metrics.py

@@ -640,6 +640,32 @@ def test_non_existent_metrics_returns_404(
     assert get_error_message(response.json()) == "could not find the metric whatevergoennnsgoeshere"
 
 
+def test_non_existent_internal_metrics_returns_no_warning(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+) -> None:
+
+    now = datetime.now(tz=UTC).replace(second=0, microsecond=0)
+    metric_name = "signoz_calls_total"
+
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    query = build_builder_query(
+        "A",
+        metric_name,
+        "doesnotreallymatter",
+        "sum",
+    )
+
+    end_ms = int(now.timestamp() * 1000)
+
+    start_2h = int((now - timedelta(hours=2)).timestamp() * 1000)
+    response = make_query_request(signoz, token, start_2h, end_ms, [query])
+    assert response.status_code == HTTPStatus.OK
+    data = response.json()
+    assert get_all_warnings(data) == []
+
+
 # Verify /api/v1/fields/values filters label values by metricNamespace prefix.
 # Inserts metrics under ns.a and ns.b, then asserts a specific prefix returns
 # only matching values while a common prefix returns both.

tests/integration/tests/querier_skip_resource_fingerprint/01_skip_resource_fingerprint.py

@@ -1,21 +1,11 @@
 """
 Transparency check for the skip_resource_fingerprint optimization (traces and logs).
 
-The optimization changes how a query's resource conditions are resolved depending on
-how selective they are, but must change only ClickHouse performance, never the rows.
-Each test runs the same query against the primary instance (optimization on,
-threshold=3) and `signoz_fingerprint` (optimization off) and asserts the responses
-are identical, covering all three resolver outcomes:
-
-  - CTE: a filter matching fewer fingerprints than the threshold resolves through the
-    fingerprint CTE.
-  - Fallback: a filter matching at or above the threshold pushes resource conditions
-    onto the main spans/logs table instead.
-  - No-op: a filter with no resource conditions to pre-resolve (no resource field, or
-    resource fields only under an OR) filters inline on the main table.
-
-A fingerprint is the hash of the entire resource set, so two rows sharing the filtered
-attribute but differing in any other resource attribute are distinct fingerprints.
+At or above the configured fingerprint threshold the optimization pushes resource
+conditions onto the main spans/logs table instead of the fingerprint CTE. That
+rewrite must change ClickHouse performance, never the rows: each test runs the same
+query against the primary instance (optimization on, threshold=2) and
+`signoz_fingerprint` (optimization off) and asserts the responses are identical.
 """
 
 from collections.abc import Callable
@@ -42,10 +32,10 @@ def test_skip_resource_fingerprint_traces_fallback_matches_fingerprint(
     get_token: Callable[[str, str], str],
     insert_traces: Callable[[list[Traces]], None],
 ) -> None:
-    """A filter matching >= threshold fingerprints drives the fallback path; rows must match the fingerprint baseline."""
+    """A >= 2-fingerprint filter drives the fallback path; rows must match the fingerprint baseline."""
     now = datetime.now(tz=UTC).replace(second=0, microsecond=0)
 
-    # 3 distinct services share one env (3 fingerprints >= threshold 3 -> fallback);
+    # 3 distinct services share one env (3 fingerprints > threshold 2 -> fallback);
     # the 4th has a different env and must be excluded.
     env = {"deployment.environment": "skip-fallback"}
     insert_traces(
@@ -75,124 +65,6 @@ def test_skip_resource_fingerprint_traces_fallback_matches_fingerprint(
     assert_identical_query_response(optimized, fingerprint)
 
 
-def test_skip_resource_fingerprint_traces_cte_matches_fingerprint(
-    signoz: types.SigNoz,
-    signoz_fingerprint: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_traces: Callable[[list[Traces]], None],
-) -> None:
-    """A filter matching < threshold fingerprints resolves through the fingerprint CTE; rows must match the baseline."""
-    now = datetime.now(tz=UTC).replace(second=0, microsecond=0)
-
-    # 2 distinct services share one env -> 2 fingerprints, below threshold 3 -> CTE.
-    # svc-a carries two spans to also exercise CTE dedup. A service in a different env
-    # is a third fingerprint but excluded by the filter.
-    env = {"deployment.environment": "skip-cte"}
-    insert_traces(
-        [
-            Traces(timestamp=now - timedelta(seconds=10), resources={"service.name": "skip-cte-svc-a", **env}),
-            Traces(timestamp=now - timedelta(seconds=9), resources={"service.name": "skip-cte-svc-a", **env}),
-            Traces(timestamp=now - timedelta(seconds=8), resources={"service.name": "skip-cte-svc-b", **env}),
-            Traces(timestamp=now - timedelta(seconds=7), resources={"service.name": "skip-cte-other", "deployment.environment": "skip-cte-other-env"}),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-    query = BuilderQuery(
-        signal="traces",
-        limit=50,
-        order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
-        filter_expression="deployment.environment = 'skip-cte'",
-        select_fields=[TelemetryFieldKey("service.name", "string", "resource")],
-    ).to_dict()
-
-    start_ms = int((datetime.now(tz=UTC) - timedelta(minutes=5)).timestamp() * 1000)
-    end_ms = int(datetime.now(tz=UTC).timestamp() * 1000)
-    optimized = make_query_request(signoz, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-    fingerprint = make_query_request(signoz_fingerprint, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-
-    assert len(get_rows(optimized)) == 3
-    assert_identical_query_response(optimized, fingerprint)
-
-
-def test_skip_resource_fingerprint_traces_or_filter_matches_fingerprint(
-    signoz: types.SigNoz,
-    signoz_fingerprint: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_traces: Callable[[list[Traces]], None],
-) -> None:
-    """A resource condition under an OR has no fixed fingerprint set, so it filters inline; rows must match the baseline."""
-    now = datetime.now(tz=UTC).replace(second=0, microsecond=0)
-
-    # `name = ... OR service.name = ...` can't reduce to a fingerprint set (no-op path).
-    # span-1 matches on name, span-2 on service.name, span-3 matches neither.
-    env = {"deployment.environment": "skip-or"}
-    insert_traces(
-        [
-            Traces(timestamp=now - timedelta(seconds=10), name="tr-or-name", resources={"service.name": "tr-or-svc-x", **env}),
-            Traces(timestamp=now - timedelta(seconds=9), name="tr-or-other", resources={"service.name": "tr-or-svc-a", **env}),
-            Traces(timestamp=now - timedelta(seconds=8), name="tr-or-other", resources={"service.name": "tr-or-svc-b", **env}),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-    query = BuilderQuery(
-        signal="traces",
-        limit=50,
-        order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
-        filter_expression="name = 'tr-or-name' OR service.name = 'tr-or-svc-a'",
-        select_fields=[TelemetryFieldKey("service.name", "string", "resource")],
-    ).to_dict()
-
-    start_ms = int((datetime.now(tz=UTC) - timedelta(minutes=5)).timestamp() * 1000)
-    end_ms = int(datetime.now(tz=UTC).timestamp() * 1000)
-    optimized = make_query_request(signoz, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-    fingerprint = make_query_request(signoz_fingerprint, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-
-    assert len(get_rows(optimized)) == 2
-    assert_identical_query_response(optimized, fingerprint)
-
-
-def test_skip_resource_fingerprint_traces_no_resource_filter_matches_fingerprint(
-    signoz: types.SigNoz,
-    signoz_fingerprint: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_traces: Callable[[list[Traces]], None],
-) -> None:
-    """A filter with no resource field has nothing to pre-resolve, so it filters inline; rows must match the baseline."""
-    now = datetime.now(tz=UTC).replace(second=0, microsecond=0)
-
-    # Filtering only on the span name (an intrinsic, non-resource field) is a no-op for
-    # the resolver; span-1 matches, span-2 does not.
-    env = {"deployment.environment": "skip-nr"}
-    insert_traces(
-        [
-            Traces(timestamp=now - timedelta(seconds=10), name="tr-nr-name", resources={"service.name": "tr-nr-svc-a", **env}),
-            Traces(timestamp=now - timedelta(seconds=9), name="tr-nr-other", resources={"service.name": "tr-nr-svc-b", **env}),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-    query = BuilderQuery(
-        signal="traces",
-        limit=50,
-        order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
-        filter_expression="name = 'tr-nr-name'",
-        select_fields=[TelemetryFieldKey("service.name", "string", "resource")],
-    ).to_dict()
-
-    start_ms = int((datetime.now(tz=UTC) - timedelta(minutes=5)).timestamp() * 1000)
-    end_ms = int(datetime.now(tz=UTC).timestamp() * 1000)
-    optimized = make_query_request(signoz, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-    fingerprint = make_query_request(signoz_fingerprint, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-
-    assert len(get_rows(optimized)) == 1
-    assert_identical_query_response(optimized, fingerprint)
-
-
 def test_skip_resource_fingerprint_logs_fallback_matches_fingerprint(
     signoz: types.SigNoz,
     signoz_fingerprint: types.SigNoz,
@@ -200,10 +72,10 @@ def test_skip_resource_fingerprint_logs_fallback_matches_fingerprint(
     get_token: Callable[[str, str], str],
     insert_logs: Callable[[list[Logs]], None],
 ) -> None:
-    """A filter matching >= threshold fingerprints drives the fallback path; rows must match the fingerprint baseline."""
+    """A >= 2-fingerprint filter drives the fallback path; rows must match the fingerprint baseline."""
     now = datetime.now(tz=UTC)
 
-    # 3 distinct services share one env (3 fingerprints >= threshold 3 -> fallback);
+    # 3 distinct services share one env (3 fingerprints > threshold 2 -> fallback);
     # the 4th has a different env and must be excluded.
     env = {"deployment.environment": "skip-logs-fallback"}
     insert_logs(
@@ -231,121 +103,3 @@ def test_skip_resource_fingerprint_logs_fallback_matches_fingerprint(
 
     assert len(get_rows(optimized)) == 3
     assert_identical_query_response(optimized, fingerprint)
-
-
-def test_skip_resource_fingerprint_logs_cte_matches_fingerprint(
-    signoz: types.SigNoz,
-    signoz_fingerprint: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_logs: Callable[[list[Logs]], None],
-) -> None:
-    """A filter matching < threshold fingerprints resolves through the fingerprint CTE; rows must match the baseline."""
-    now = datetime.now(tz=UTC)
-
-    # 2 distinct services share one env -> 2 fingerprints, below threshold 3 -> CTE.
-    # svc-a carries two logs to also exercise CTE dedup. A service in a different env
-    # is a third fingerprint but excluded by the filter.
-    env = {"deployment.environment": "skip-logs-cte"}
-    insert_logs(
-        [
-            Logs(timestamp=now - timedelta(seconds=10), resources={"service.name": "skip-logs-cte-svc-a", **env}, body="a"),
-            Logs(timestamp=now - timedelta(seconds=9), resources={"service.name": "skip-logs-cte-svc-a", **env}, body="b"),
-            Logs(timestamp=now - timedelta(seconds=8), resources={"service.name": "skip-logs-cte-svc-b", **env}, body="c"),
-            Logs(timestamp=now - timedelta(seconds=7), resources={"service.name": "skip-logs-cte-other", "deployment.environment": "skip-logs-cte-other-env"}, body="noise"),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-    query = BuilderQuery(
-        signal="logs",
-        limit=50,
-        order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
-        filter_expression="deployment.environment = 'skip-logs-cte'",
-        select_fields=[TelemetryFieldKey("service.name", "string", "resource"), TelemetryFieldKey("body")],
-    ).to_dict()
-
-    start_ms = int((datetime.now(tz=UTC) - timedelta(minutes=5)).timestamp() * 1000)
-    end_ms = int(datetime.now(tz=UTC).timestamp() * 1000)
-    optimized = make_query_request(signoz, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-    fingerprint = make_query_request(signoz_fingerprint, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-
-    assert len(get_rows(optimized)) == 3
-    assert_identical_query_response(optimized, fingerprint)
-
-
-def test_skip_resource_fingerprint_logs_or_filter_matches_fingerprint(
-    signoz: types.SigNoz,
-    signoz_fingerprint: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_logs: Callable[[list[Logs]], None],
-) -> None:
-    """A resource condition under an OR has no fixed fingerprint set, so it filters inline; rows must match the baseline."""
-    now = datetime.now(tz=UTC)
-
-    # `test.marker = ... OR service.name = ...` can't reduce to a fingerprint set (no-op path).
-    # log-1 matches on the attribute, log-2 on service.name, log-3 matches neither.
-    env = {"deployment.environment": "skip-logs-or"}
-    insert_logs(
-        [
-            Logs(timestamp=now - timedelta(seconds=10), resources={"service.name": "logs-or-svc-x", **env}, attributes={"test.marker": "logs-or-hit"}, body="a"),
-            Logs(timestamp=now - timedelta(seconds=9), resources={"service.name": "logs-or-svc-a", **env}, body="b"),
-            Logs(timestamp=now - timedelta(seconds=8), resources={"service.name": "logs-or-svc-b", **env}, body="noise"),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-    query = BuilderQuery(
-        signal="logs",
-        limit=50,
-        order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
-        filter_expression="test.marker = 'logs-or-hit' OR service.name = 'logs-or-svc-a'",
-        select_fields=[TelemetryFieldKey("service.name", "string", "resource"), TelemetryFieldKey("body")],
-    ).to_dict()
-
-    start_ms = int((datetime.now(tz=UTC) - timedelta(minutes=5)).timestamp() * 1000)
-    end_ms = int(datetime.now(tz=UTC).timestamp() * 1000)
-    optimized = make_query_request(signoz, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-    fingerprint = make_query_request(signoz_fingerprint, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-
-    assert len(get_rows(optimized)) == 2
-    assert_identical_query_response(optimized, fingerprint)
-
-
-def test_skip_resource_fingerprint_logs_no_resource_filter_matches_fingerprint(
-    signoz: types.SigNoz,
-    signoz_fingerprint: types.SigNoz,
-    create_user_admin: None,  # pylint: disable=unused-argument
-    get_token: Callable[[str, str], str],
-    insert_logs: Callable[[list[Logs]], None],
-) -> None:
-    """A filter with no resource field has nothing to pre-resolve, so it filters inline; rows must match the baseline."""
-    now = datetime.now(tz=UTC)
-
-    # Filtering only on an attribute (a non-resource field) is a no-op for the resolver;
-    # log-1 matches, log-2 does not.
-    env = {"deployment.environment": "skip-logs-nr"}
-    insert_logs(
-        [
-            Logs(timestamp=now - timedelta(seconds=10), resources={"service.name": "logs-nr-svc-a", **env}, attributes={"test.marker": "logs-nr-hit"}, body="a"),
-            Logs(timestamp=now - timedelta(seconds=9), resources={"service.name": "logs-nr-svc-b", **env}, body="b"),
-        ]
-    )
-
-    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
-    query = BuilderQuery(
-        signal="logs",
-        limit=50,
-        order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
-        filter_expression="test.marker = 'logs-nr-hit'",
-        select_fields=[TelemetryFieldKey("service.name", "string", "resource"), TelemetryFieldKey("body")],
-    ).to_dict()
-
-    start_ms = int((datetime.now(tz=UTC) - timedelta(minutes=5)).timestamp() * 1000)
-    end_ms = int(datetime.now(tz=UTC).timestamp() * 1000)
-    optimized = make_query_request(signoz, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-    fingerprint = make_query_request(signoz_fingerprint, token, start_ms=start_ms, end_ms=end_ms, request_type="raw", queries=[query])
-
-    assert len(get_rows(optimized)) == 1
-    assert_identical_query_response(optimized, fingerprint)

tests/integration/tests/querier_skip_resource_fingerprint/conftest.py

@@ -18,9 +18,8 @@ def signoz_skip_resource_fingerprint(
 ) -> types.SigNoz:
     """
     Package-scoped SigNoz instance with the skip_resource_fingerprint
-    optimization enabled and a low threshold (3) so both resolver paths are
-    exercised: filters matching < 3 fingerprints resolve through the fingerprint
-    CTE, and filters matching >= 3 fingerprints fall back to the main table.
+    optimization enabled and a low threshold so the fallback resolver path is
+    exercised (filters matching >= 2 fingerprints skip the fingerprint CTE).
     """
     return create_signoz(
         network=network,
@@ -33,7 +32,7 @@ def signoz_skip_resource_fingerprint(
         cache_key="signoz-skip-resource-fingerprint",
         env_overrides={
             "SIGNOZ_QUERIER_SKIP__RESOURCE__FINGERPRINT_ENABLED": True,
-            "SIGNOZ_QUERIER_SKIP__RESOURCE__FINGERPRINT_THRESHOLD": 3,
+            "SIGNOZ_QUERIER_SKIP__RESOURCE__FINGERPRINT_THRESHOLD": 2,
         },
     )