Merge pull request #1006 from SiaFoundation/chris/aggregate-contract-…

…speedup

Improve `findAggregateContractMetric` performance and introduce 30 second granularity for contract metrics

stores/metrics.go

@@ -14,6 +14,10 @@ import (
 	"gorm.io/gorm/clause"
 )
 
+const (
+	contractMetricGranularity = 5 * time.Minute
+)
+
 type (
 	// dbContractMetric tracks information about a contract's funds.  It is
 	// supposed to be reported by a worker every time a contract is revised.
@@ -246,6 +250,21 @@ func (s *SQLStore) RecordContractMetric(ctx context.Context, metrics ...api.Cont
 		}
 	}
 	return s.dbMetrics.Transaction(func(tx *gorm.DB) error {
+		// delete any existing metric for the same contract that has happened
+		// within the same 5' window by diving the timestamp by 5' and use integer division.
+		for _, metric := range metrics {
+			intervalStart := metric.Timestamp.Std().Truncate(contractMetricGranularity)
+			intervalEnd := intervalStart.Add(contractMetricGranularity)
+			err := tx.
+				Where("timestamp >= ?", unixTimeMS(intervalStart)).
+				Where("timestamp < ?", unixTimeMS(intervalEnd)).
+				Where("fcid", fileContractID(metric.ContractID)).
+				Delete(&dbContractMetric{}).
+				Error
+			if err != nil {
+				return err
+			}
+		}
 		return tx.Create(&dbMetrics).Error
 	})
 }
@@ -522,43 +541,43 @@ func (s *SQLStore) findAggregatedContractPeriods(start time.Time, n uint64, inte
 		return nil, api.ErrMaxIntervalsExceeded
 	}
 	end := start.Add(time.Duration(n) * interval)
-	var metricsWithPeriod []struct {
+
+	type metricWithPeriod struct {
 		Metric dbContractMetric `gorm:"embedded"`
 		Period int64
 	}
-	err := s.dbMetrics.Raw(`
-		WITH RECURSIVE periods AS (
-			SELECT ? AS period_start
-			UNION ALL
-			SELECT period_start + ?
-			FROM periods
-			WHERE period_start < ? - ?
-		)
-		SELECT contracts.*, i.Period FROM contracts
-		INNER JOIN (
-		SELECT
-			p.period_start as Period,
-			MIN(c.id) AS id
-		FROM
-			periods p
-		INNER JOIN
-			contracts c ON c.timestamp >= p.period_start AND c.timestamp < p.period_start + ?
-		GROUP BY
-			p.period_start, c.fcid
-		ORDER BY
-			p.period_start ASC
-		) i ON contracts.id = i.id
-	`, unixTimeMS(start),
-		interval.Milliseconds(),
-		unixTimeMS(end),
-		interval.Milliseconds(),
-		interval.Milliseconds(),
-	).
-		Scan(&metricsWithPeriod).
-		Error
+	var metricsWithPeriod []metricWithPeriod
+
+	err := s.dbMetrics.Transaction(func(tx *gorm.DB) error {
+		var fcids []fileContractID
+		if err := tx.Raw("SELECT DISTINCT fcid FROM contracts WHERE contracts.timestamp >= ? AND contracts.timestamp < ?", unixTimeMS(start), unixTimeMS(end)).
+			Scan(&fcids).Error; err != nil {
+			return fmt.Errorf("failed to fetch distinct contract ids: %w", err)
+		}
+
+		for intervalStart := start; intervalStart.Before(end); intervalStart = intervalStart.Add(interval) {
+			intervalEnd := intervalStart.Add(interval)
+			for _, fcid := range fcids {
+				var metrics []dbContractMetric
+				err := tx.Raw("SELECT * FROM contracts WHERE contracts.timestamp >= ? AND contracts.timestamp < ? AND contracts.fcid = ? LIMIT 1", unixTimeMS(intervalStart), unixTimeMS(intervalEnd), fileContractID(fcid)).
+					Scan(&metrics).Error
+				if err != nil {
+					return fmt.Errorf("failed to fetch contract metrics: %w", err)
+				} else if len(metrics) == 0 {
+					continue
+				}
+				metricsWithPeriod = append(metricsWithPeriod, metricWithPeriod{
+					Metric: metrics[0],
+					Period: intervalStart.UnixMilli(),
+				})
+			}
+		}
+		return nil
+	})
 	if err != nil {
-		return nil, fmt.Errorf("failed to fetch aggregate metrics: %w", err)
+		return nil, err
 	}
+
 	currentPeriod := int64(math.MinInt64)
 	var metrics []dbContractMetric
 	for _, m := range metricsWithPeriod {

stores/metrics_test.go

@@ -488,6 +488,30 @@ func TestContractMetrics(t *testing.T) {
 	} else if len(metrics) != 1 {
 		t.Fatalf("expected 1 metric, got %v", len(metrics))
 	}
+
+	// Drop all metrics.
+	if err := ss.dbMetrics.Where("TRUE").Delete(&dbContractMetric{}).Error; err != nil {
+		t.Fatal(err)
+	}
+
+	// Record multiple metrics for the same contract - one per second over 10 minutes
+	for i := int64(0); i < 600; i++ {
+		err := ss.RecordContractMetric(context.Background(), api.ContractMetric{
+			ContractID: types.FileContractID{1},
+			Timestamp:  api.TimeRFC3339(time.Unix(i, 0)),
+		})
+		if err != nil {
+			t.Fatal(err)
+		}
+	}
+
+	// Check how many metrics were recorded.
+	var n int64
+	if err := ss.dbMetrics.Model(&dbContractMetric{}).Count(&n).Error; err != nil {
+		t.Fatal(err)
+	} else if n != 2 {
+		t.Fatalf("expected 2 metrics, got %v", n)
+	}
 }
 
 func TestWalletMetrics(t *testing.T) {

stores/migrations/mysql/metrics/migration_00001_idx_contracts_fcid_timestamp.sql

@@ -0,0 +1 @@
+CREATE INDEX `idx_contracts_fcid_timestamp` ON `contracts`(`fcid`,`timestamp`);

stores/migrations/mysql/metrics/schema.sql

@@ -82,7 +82,8 @@ CREATE TABLE `contracts` (
   KEY `idx_contracts_timestamp` (`timestamp`),
   KEY `idx_remaining_funds` (`remaining_funds_lo`,`remaining_funds_hi`),
   KEY `idx_delete_spending` (`delete_spending_lo`,`delete_spending_hi`),
-  KEY `idx_list_spending` (`list_spending_lo`,`list_spending_hi`)
+  KEY `idx_list_spending` (`list_spending_lo`,`list_spending_hi`),
+  KEY `idx_contracts_fcid_timestamp` (`fcid`,`timestamp`)
 ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci;
 
 -- dbPerformanceMetric

stores/migrations/sqlite/metrics/migration_00001_idx_contracts_fcid_timestamp.sql

@@ -0,0 +1 @@
+CREATE INDEX `idx_contracts_fcid_timestamp` ON `contracts`(`fcid`,`timestamp`);

stores/migrations/sqlite/metrics/schema.sql

@@ -11,6 +11,7 @@ CREATE INDEX `idx_download_spending` ON `contracts`(`download_spending_lo`,`down
 CREATE INDEX `idx_upload_spending` ON `contracts`(`upload_spending_lo`,`upload_spending_hi`);
 CREATE INDEX `idx_contracts_revision_number` ON `contracts`(`revision_number`);
 CREATE INDEX `idx_remaining_funds` ON `contracts`(`remaining_funds_lo`,`remaining_funds_hi`);
+CREATE INDEX `idx_contracts_fcid_timestamp` ON `contracts`(`fcid`,`timestamp`);
 
 -- dbContractPruneMetric
 CREATE TABLE `contract_prunes` (`id` integer PRIMARY KEY AUTOINCREMENT,`created_at` datetime,`timestamp` BIGINT NOT NULL,`fcid` blob NOT NULL,`host` blob NOT NULL,`host_version` text,`pruned` BIGINT NOT NULL,`remaining` BIGINT NOT NULL,`duration` integer NOT NULL);

stores/migrations_metrics.go

@@ -8,20 +8,26 @@ import (
 	"gorm.io/gorm"
 )
 
-func performMetricsMigrations(tx *gorm.DB, logger *zap.SugaredLogger) error {
+func performMetricsMigrations(db *gorm.DB, logger *zap.SugaredLogger) error {
 	dbIdentifier := "metrics"
 	migrations := []*gormigrate.Migration{
 		{
 			ID:      "00001_init",
 			Migrate: func(tx *gorm.DB) error { return errRunV072 },
 		},
+		{
+			ID: "00001_idx_contracts_fcid_timestamp",
+			Migrate: func(tx *gorm.DB) error {
+				return performMigration(tx, dbIdentifier, "00001_idx_contracts_fcid_timestamp", logger)
+			},
+		},
 	}
 
 	// Create migrator.
-	m := gormigrate.New(tx, gormigrate.DefaultOptions, migrations)
+	m := gormigrate.New(db, gormigrate.DefaultOptions, migrations)
 
 	// Set init function.
-	m.InitSchema(initSchema(tx, dbIdentifier, logger))
+	m.InitSchema(initSchema(db, dbIdentifier, logger))
 
 	// Perform migrations.
 	if err := m.Migrate(); err != nil {