PMM-13374 Improve fingerprinter generation.

percona · Oct 28, 2024 · 33434f4 · 33434f4
1 parent 0971d2e
commit 33434f4
Show file tree

Hide file tree

Showing 2 changed files with 43 additions and 22 deletions.
diff --git a/agent/agents/mongodb/internal/profiler/fingerprinter/fingerprinter.go b/agent/agents/mongodb/internal/profiler/fingerprinter/fingerprinter.go
@@ -57,7 +57,7 @@ func (pf *ProfilerFingerprinter) Fingerprint(doc proto.SystemProfile) (fingerpri
 
 // Helper for find operations with optional parameters.
 func (pf *ProfilerFingerprinter) fingerprintFind(fp fingerprinter.Fingerprint, doc proto.SystemProfile) (fingerprinter.Fingerprint, error) {
-	filter := "{}"
+	filter := ""
 	command := doc.Command.Map()
 	if f, ok := command["filter"]; ok {
 		values := maskValues(f, make(map[string]maskOption))
@@ -141,23 +141,28 @@ func (pf *ProfilerFingerprinter) fingerprintDelete(fp fingerprinter.Fingerprint,
 // Helper for general command operations, including support for "aggregate" commands
 func (pf *ProfilerFingerprinter) fingerprintCommand(fp fingerprinter.Fingerprint, doc proto.SystemProfile) (fingerprinter.Fingerprint, error) {
 	// Unmarshal the command into a map for easy access and manipulation
-	commandMap := doc.Command.Map()
-
-	maskOptions := make(map[string]maskOption)
-	maskOptions["$db"] = maskOption{remove: true}
-	maskOptions["$readPreference"] = maskOption{remove: true}
-	maskOptions["lsid"] = maskOption{remove: true}
-	maskOptions["findAndModify"] = maskOption{skipMask: true}
-	maskOptions["remove"] = maskOption{skipMask: true}
-	if cmd, exists := commandMap["aggregate"]; exists {
+	command := doc.Command.Map()
+
+	maskOptions := map[string]maskOption{
+		"$db":                      {remove: true},
+		"$readPreference":          {remove: true},
+		"$readConcern":             {remove: true},
+		"$writeConcern":            {remove: true},
+		"$clusterTime":             {remove: true},
+		"$oplogQueryData":          {remove: true},
+		"$replData":                {remove: true},
+		"lastKnownCommittedOpTime": {remove: true},
+		"lsid":                     {remove: true},
+		"findAndModify":            {skipMask: true},
+		"remove":                   {skipMask: true},
+	}
+	if _, exists := command["aggregate"]; exists {
 		// Set collection and initialize aggregation structure
-		collectionName, _ := cmd.(string)
-		fp.Collection = collectionName
-		fp.Fingerprint = fmt.Sprintf(`db.%s.aggregate([`, collectionName)
+		fp.Fingerprint = fmt.Sprintf(`db.%s.aggregate([`, fp.Collection)
 		stageStrings := []string{}
 
 		// Process pipeline stages, replacing all values with "?"
-		if pipeline, exists := commandMap["pipeline"]; exists {
+		if pipeline, exists := command["pipeline"]; exists {
 			pipelineStages, _ := pipeline.(bson.A)
 
 			for _, stage := range pipelineStages {
@@ -166,9 +171,7 @@ func (pf *ProfilerFingerprinter) fingerprintCommand(fp fingerprinter.Fingerprint
 				switch {
 				case stageMap["$match"] != nil:
 					stageJSON, _ = json.Marshal(maskValues(stageMap, maskOptions))
-				case stageMap["$group"] != nil:
-					stageJSON, _ = bson.MarshalExtJSON(stageMap, false, false)
-				case stageMap["$sort"] != nil:
+				default:
 					stageJSON, _ = bson.MarshalExtJSON(stageMap, false, false)
 				}
 
@@ -178,7 +181,7 @@ func (pf *ProfilerFingerprinter) fingerprintCommand(fp fingerprinter.Fingerprint
 			fp.Fingerprint += strings.Join(stageStrings, ", ")
 		}
 		fp.Fingerprint += "])"
-		if collation, exists := commandMap["collation"]; exists {
+		if collation, exists := command["collation"]; exists {
 			collationMasked, _ := json.Marshal(maskValues(collation, maskOptions))
 			fp.Fingerprint += fmt.Sprintf(`, collation: %s`, collationMasked)
 		}

diff --git a/agent/agents/mongodb/internal/profiler/fingerprinter/fingerprinter_test.go b/agent/agents/mongodb/internal/profiler/fingerprinter/fingerprinter_test.go
@@ -107,18 +107,35 @@ func TestProfilerFingerprinter(t *testing.T) {
 		database.Collection("test").FindOneAndDelete(ctx, bson.M{"id": 1})
 		database.Collection("secondcollection").Find(ctx, bson.M{"name": "sec"}, options.Find().SetLimit(1).SetSort(bson.M{"id": -1}))
 		database.Collection("test").Aggregate(ctx,
-			bson.A{
-				bson.M{
+			[]bson.M{
+				{
 					"$match": bson.M{"id": 0, "time": bson.M{"$gt": time.Now().Add(-time.Hour)}},
 				},
-				bson.M{
+				{
 					"$group": bson.M{"_id": "$id", "count": bson.M{"$sum": 1}},
 				},
-				bson.M{
+				{
 					"$sort": bson.M{"_id": 1},
 				},
 			},
 		)
+		database.Collection("secondcollection").Aggregate(ctx, mongo.Pipeline{
+			bson.D{
+				{
+					Key: "$collStats", Value: bson.M{
+					// TODO: PMM-9568 : Add support to handle histogram metrics
+					"latencyStats": bson.M{"histograms": false},
+					"storageStats": bson.M{"scale": 1},
+				},
+				},
+			}, bson.D{
+				{
+					Key: "$project", Value: bson.M{
+					"storageStats.wiredTiger":   0,
+					"storageStats.indexDetails": 0,
+				},
+				},
+			}})
 		database.Collection("secondcollection").DeleteOne(ctx, bson.M{"id": 0})
 		database.Collection("test").DeleteMany(ctx, bson.M{"name": "test"})
 		profilerCollection := database.Collection("system.profile")
@@ -160,6 +177,7 @@ func TestProfilerFingerprinter(t *testing.T) {
 			`db.runCommand({"findAndModify":"test","query":{"id":"?"},"remove":true})`,
 			`db.secondcollection.find({"name":"?"}).sort({"id":-1}).limit(?)`,
 			`db.test.aggregate([{"$match":{"id":"?","time":{"$gt":"?"}}}, {"$group":{"_id":"$id","count":{"$sum":1}}}, {"$sort":{"_id":1}}])`,
+			`db.secondcollection.aggregate([{"$collStats":{"latencyStats":{"histograms":false},"storageStats":{"scale":1}}}, {"$project":{"storageStats.wiredTiger":0,"storageStats.indexDetails":0}}])`,
 			`db.secondcollection.deleteOne({"id":"?"})`,
 			`db.test.deleteMany({"name":"?"})`,
 		}