Cherry pick Aleksandar readpath patches (#4404)

* Switch to swiss map for profile merges

* Add support for disabling profile sanitization in the read path

* fmt

---------

Co-authored-by: Aleksandar Petrov <[email protected]>
Co-authored-by: Tolya Korniltsev <[email protected]>

Author: 3 people

api/gen/proto/go/query/v1/query.pb.go

@@ -1328,7 +1328,12 @@
     },
     "v1InvokeOptions": {
       "type": "object",
-      "description": "Query workers might not have access to the tenant\n overrides, therefore all the necessary options should\n be listed in the request explicitly."
+      "properties": {
+        "sanitizeOnMerge": {
+          "type": "boolean",
+          "description": "Query workers might not have access to the tenant\noverrides, therefore all the necessary options should\nbe listed in the request explicitly."
+        }
+      }
     },
     "v1InvokeResponse": {
       "type": "object",

api/gen/proto/go/query/v1/query_vtproto.pb.go

@@ -40,6 +40,7 @@ message InvokeOptions {
   // Query workers might not have access to the tenant
   // overrides, therefore all the necessary options should
   // be listed in the request explicitly.
+  bool sanitize_on_merge = 1;
 }
 
 message InvokeRequest {

api/openapiv2/gen/phlare.swagger.json

@@ -591,6 +591,8 @@ Usage of ./pyroscope:
     	Whether the series portion of query analysis is enabled. If disabled, no series data (e.g., series count) will be calculated by the /AnalyzeQuery endpoint.
   -querier.query-store-after duration
     	The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'. (default 4h0m0s)
+  -querier.sanitize-on-merge
+    	Whether profiles should be sanitized when merging. (default true)
   -querier.split-queries-by-interval duration
     	Split queries by a time interval and execute in parallel. The value 0 disables splitting by time
   -query-frontend.grpc-client-config.backoff-max-period duration

api/query/v1/query.proto

@@ -171,6 +171,8 @@ Usage of ./pyroscope:
     	Whether query analysis is enabled in the query frontend. If disabled, the /AnalyzeQuery endpoint will return an empty response. (default true)
   -querier.query-analysis-series-enabled
     	Whether the series portion of query analysis is enabled. If disabled, no series data (e.g., series count) will be calculated by the /AnalyzeQuery endpoint.
+  -querier.sanitize-on-merge
+    	Whether profiles should be sanitized when merging. (default true)
   -querier.split-queries-by-interval duration
     	Split queries by a time interval and execute in parallel. The value 0 disables splitting by time
   -query-scheduler.max-outstanding-requests-per-tenant int

cmd/pyroscope/help-all.txt.tmpl

@@ -2312,6 +2312,10 @@ distributor_usage_groups:
 # CLI flag: -querier.split-queries-by-interval
 [split_queries_by_interval: <duration> | default = 0s]
 
+# Whether profiles should be sanitized when merging.
+# CLI flag: -querier.sanitize-on-merge
+[query_sanitize_on_merge: <boolean> | default = true]
+
 # Delete blocks containing samples older than the specified retention period. 0
 # to disable.
 # CLI flag: -compactor.blocks-retention-period

cmd/pyroscope/help.txt.tmpl

@@ -748,7 +748,7 @@ func mergeProfile(profile *profilev1.Profile) aggregator.AggregateFn[*pprof.Prof
 		if m == nil {
 			m = new(pprof.ProfileMerge)
 		}
-		if err := m.Merge(profile); err != nil {
+		if err := m.Merge(profile, true); err != nil {
 			return nil, connect.NewError(connect.CodeInvalidArgument, err)
 		}
 		return m, nil

docs/sources/configure-server/reference-configuration-parameters/index.md

@@ -113,6 +113,7 @@ type Limits interface {
 	MaxQueryLookback(tenantID string) time.Duration
 	QueryAnalysisEnabled(string) bool
 	SymbolizerEnabled(string) bool
+	QuerySanitizeOnMerge(string) bool
 	validation.FlameGraphLimits
 }
 

pkg/distributor/distributor.go

@@ -23,6 +23,10 @@ func (m *mockLimits) QuerySplitDuration(_ string) time.Duration {
 	return time.Hour
 }
 
+func (m *mockLimits) QuerySanitizeOnMerge(_ string) bool {
+	return true
+}
+
 func (m *mockLimits) MaxQueryParallelism(_ string) int {
 	return 100
 }