refactor: move symbols logic from segment to head & change needsSymbolizion by hasNativeProfiles

Author: marcsanmi

pkg/experiment/block/metadata/metadata_labels.go

@@ -16,9 +16,9 @@ import (
 // TODO(kolesnikovae): LabelBuilder pool.
 
 const (
-	LabelNameTenantDataset      = "__tenant_dataset__"
-	LabelValueDatasetTSDBIndex  = "dataset_tsdb_index"
-	LabelNameNeedsSymbolization = "__needs_symbolization__"
+	LabelNameTenantDataset     = "__tenant_dataset__"
+	LabelValueDatasetTSDBIndex = "dataset_tsdb_index"
+	LabelNameHasNativeProfiles = "__has_native_profiles__"
 )
 
 type LabelBuilder struct {

pkg/experiment/ingester/memdb/head.go

@@ -21,10 +21,11 @@ import (
 )
 
 type FlushedHead struct {
-	Index    []byte
-	Profiles []byte
-	Symbols  []byte
-	Meta     struct {
+	Index             []byte
+	Profiles          []byte
+	Symbols           []byte
+	HasNativeProfiles bool
+	Meta              struct {
 		ProfileTypeNames []string
 		MinTimeNanos     int64
 		MaxTimeNanos     int64
@@ -153,6 +154,18 @@ func (h *Head) flush(ctx context.Context) (*FlushedHead, error) {
 		return res, nil
 	}
 
+	var hasNativeProfiles bool
+	mappings := h.symbols.Symbols().Mappings
+	for i := range mappings {
+		hasNativeProfiles = hasNativeProfiles || !mappings[i].HasFunctions
+	}
+	locations := h.symbols.Symbols().Locations
+	for i := range locations {
+		hasNativeProfiles = hasNativeProfiles || len(locations[i].Line) == 0
+	}
+
+	res.HasNativeProfiles = hasNativeProfiles
+
 	symbolsBuffer := bytes.NewBuffer(nil)
 	if err := symdb.WritePartition(h.symbols, symbolsBuffer); err != nil {
 		return nil, err

pkg/experiment/ingester/segment.go

@@ -12,7 +12,6 @@ import (
 	"slices"
 	"strings"
 	"sync"
-	"sync/atomic"
 	"time"
 
 	"github.com/go-kit/log"
@@ -338,10 +337,8 @@ func concatSegmentHead(f *headFlush, w *writerOffset, s *metadata.StringTable) (
 		lb.WithLabelSet(model.LabelNameServiceName, f.head.key.service, model.LabelNameProfileType, profileType)
 	}
 
-	if f.head.needsSymbolization {
-		lb.WithLabelSet(metadata.LabelNameNeedsSymbolization, "true")
-	if f.head.needsSymbolization.Load() {
-		lb.WithLabelSet(model.LabelNameServiceName, f.head.key.service, metadata.LabelNameNeedsSymbolization, "true")
+	if f.flushed.HasNativeProfiles {
+		lb.WithLabelSet(model.LabelNameServiceName, f.head.key.service, metadata.LabelNameHasNativeProfiles, "true")
 	}
 
 	// Other optional labels:
@@ -442,9 +439,8 @@ func (k datasetKey) compare(x datasetKey) int {
 }
 
 type dataset struct {
-	key                datasetKey
-	head               *memdb.Head
-	needsSymbolization atomic.Bool
+	key  datasetKey
+	head *memdb.Head
 }
 
 type headFlush struct {
@@ -508,10 +504,6 @@ func (s *segment) ingest(tenantID string, p *profilev1.Profile, id uuid.UUID, la
 		service: model.Labels(labels).Get(model.LabelNameServiceName),
 	}
 	ds := s.datasetForIngest(k)
-	if !ds.needsSymbolization.Load() && !hasSymbols(p) {
-		ds.needsSymbolization.Store(true)
-	}
-
 	size := p.SizeVT()
 	rules := s.sw.limits.IngestionRelabelingRules(tenantID)
 	usage := s.sw.limits.DistributorUsageGroups(tenantID).GetUsageGroups(tenantID, labels)

pkg/frontend/read_path/query_frontend/query_frontend.go

@@ -95,11 +95,11 @@ func (q *QueryFrontend) Query(
 	// TODO(kolesnikovae): Should be dynamic.
 	p := queryplan.Build(blocks, 4, 20)
 
-	needsSymbolization := false
+	hasNativeProfiles := false
 	if q.symbolizer != nil {
 		for _, block := range blocks {
-			if q.symbolizationNeeded(block) {
-				needsSymbolization = true
+			if q.hasNativeProfiles(block) {
+				hasNativeProfiles = true
 				break
 			}
 		}
@@ -111,7 +111,7 @@ func (q *QueryFrontend) Query(
 		modifiedQueries[i] = proto.Clone(originalQuery).(*queryv1.Query)
 
 		// If we need symbolization and this is a TREE query, convert it to PPROF
-		if needsSymbolization && originalQuery.QueryType == queryv1.QueryType_QUERY_TREE {
+		if hasNativeProfiles && originalQuery.QueryType == queryv1.QueryType_QUERY_TREE {
 			modifiedQueries[i].QueryType = queryv1.QueryType_QUERY_PPROF
 			modifiedQueries[i].Pprof = &queryv1.PprofQuery{
 				MaxNodes: 0,
@@ -133,17 +133,17 @@ func (q *QueryFrontend) Query(
 		return nil, err
 	}
 
-	if needsSymbolization && q.symbolizer != nil {
+	if hasNativeProfiles && q.symbolizer != nil {
 		for i, r := range resp.Reports {
 			if r.Pprof != nil && r.Pprof.Pprof != nil {
 				var prof profilev1.Profile
 				if err := pprof.Unmarshal(r.Pprof.Pprof, &prof); err != nil {
-					level.Error(q.logger).Log("msg", "Unmarshal needsSymbolization", "error", err)
+					level.Error(q.logger).Log("msg", "unmarshal pprof", "err", err)
 					continue
 				}
 
 				if err := q.symbolizer.SymbolizePprof(ctx, &prof); err != nil {
-					level.Error(q.logger).Log("msg", "SymbolizePprof needsSymbolization", "error", err)
+					level.Error(q.logger).Log("msg", "symbolize pprof", "err", err)
 				}
 
 				// Convert back to tree if originally a tree
@@ -191,7 +191,7 @@ func (q *QueryFrontend) QueryMetadata(
 		TenantId:  tenants,
 		StartTime: req.StartTime,
 		EndTime:   req.EndTime,
-		Labels:    []string{metadata.LabelNameNeedsSymbolization},
+		Labels:    []string{metadata.LabelNameHasNativeProfiles},
 	}
 
 	// Delete all matchers but service_name with strict match. If no matchers
@@ -221,21 +221,20 @@ func (q *QueryFrontend) QueryMetadata(
 	return md.Blocks, nil
 }
 
-// symbolizationNeeded checks if a block needs symbolization
-func (q *QueryFrontend) symbolizationNeeded(block *metastorev1.BlockMeta) bool {
-	matcher, err := labels.NewMatcher(labels.MatchEqual, metadata.LabelNameNeedsSymbolization, "true")
+// hasNativeProfiles checks if a block has native profiles
+func (q *QueryFrontend) hasNativeProfiles(block *metastorev1.BlockMeta) bool {
+	matcher, err := labels.NewMatcher(labels.MatchEqual, metadata.LabelNameHasNativeProfiles, "true")
 	if err != nil {
 		return false
 	}
 
 	datasetFinder := metadata.FindDatasets(block, matcher)
 
-	// Check if any dataset matches
-	needsSymbolization := false
+	hasNativeProfiles := false
 	datasetFinder(func(ds *metastorev1.Dataset) bool {
-		needsSymbolization = true
+		hasNativeProfiles = true
 		return false
 	})
 
-	return needsSymbolization
+	return hasNativeProfiles
 }