metricbeat/module/mongodb: Improve logic

metricbeat/module/mongodb/replstatus/info.go

@@ -43,18 +43,29 @@ type CollSize struct {
 
 const oplogCol = "oplog.rs"
 
+// getReplicationInfo returns oplog info from local.oplog.rs
 func getReplicationInfo(client *mongo.Client) (*oplogInfo, error) {
-	// get oplog.rs collection
+	// Get oplog collection info from local.oplog.rs (<db>.<collection>)
 	db := client.Database("local")
 
-	// Get oplog size using collStats - this is lightweight
-	var oplogSize CollSize
+	// NOTE(shmsr):
+	// https://www.mongodb.com/docs/manual/reference/command/collStats/#syntax
+	// "scale" field is ommitted here as it is by default 1, i.e., it return sizes in bytes.
+	//
+	// Also, note that collStats is deprecated since v6.2 but as we support older
+	// versions i.e., >= 5.0, let's keep it for now as this still works.
+	// TODO(shmsr): For newers versions, we can use db.collection.stats()
+	// https://www.mongodb.com/docs/manual/reference/method/db.collection.stats/#mongodb-method-db.collection.stats
+	// or use this: https://github.com/percona/mongodb_exporter/blob/95d1865e34940d0d610bb1fbff9745bc66ddbc73/exporter/collstats_collector.go#L100
 	res := db.RunCommand(context.Background(), bson.D{
 		{Key: "collStats", Value: oplogCol},
 	})
 	if err := res.Err(); err != nil {
 		return nil, fmt.Errorf("collStats command failed: %w", err)
 	}
+
+	// Get MaxSize and Size from collStats by using db.runCommand
+	var oplogSize CollSize
 	if err := res.Decode(&oplogSize); err != nil {
 		return nil, fmt.Errorf("decode error: %w", err)
 	}
@@ -76,8 +87,32 @@ func getReplicationInfo(client *mongo.Client) (*oplogInfo, error) {
 	return info, nil
 }
 
+// getOpTimestamp returns the first and last timestamp of the oplog collection.
 func getOpTimestamp(collection *mongo.Collection) (uint32, uint32, error) {
-	// Use natural order for efficiency
+	// NOTE(shmsr):
+	//
+	// When you do db.getReplicationInfo() in monogo shell (mongosh), you can see
+	// 	{
+	//		...
+	// 		tFirst: 'Tue Jan 07 2025 22:33:28 GMT+0530 (India Standard Time)',
+	// 		tLast: 'Wed Jan 08 2025 11:45:07 GMT+0530 (India Standard Time)',
+	// 		now: 'Wed Jan 08 2025 11:45:14 GMT+0530 (India Standard Time)'
+	// 	}
+	// i.e., we get tFirst and tLast from oplog.rs which is the first and last
+	// timestamp of the oplog.
+	// Source from the same is here:
+	// 	https://github.com/mongodb/mongo/blob/20cbee37a0ee4d40b35d08b6a34ade81252f86a8/src/mongo/shell/db.js#L863
+	// This is how they calculate tFirst and tLast:
+	// 	https://github.com/mongodb/mongo/blob/20cbee37a0ee4d40b35d08b6a34ade81252f86a8/src/mongo/shell/db.js#L889
+	// So ideally, we will replicate the same logic here:
+	// 	var firstc = ol.find().sort({$natural: 1}).limit(1);
+	// 	var lastc = ol.find().sort({$natural: -1}).limit(1);
+	//
+	// oplog.rs is designed to scanned in natural ($natural) order. So, when
+	// querying without any sort, it will return the first entry in natural order.
+	// When we sort in reverse natural order (i.e., $natural: -1), it will return
+	// the last entry in natural order.
+
 	var firstDoc struct {
 		Timestamp time.Time `bson:"ts"`
 	}