[pull] master from postgres-ai:master

PROMETHEUS.md

@@ -74,6 +74,17 @@ The endpoint is publicly accessible (no authentication required) and returns met
 | `dblab_datasets_total` | Gauge | `pool` | Total number of datasets (slots) in the pool |
 | `dblab_datasets_available` | Gauge | `pool` | Number of available (non-busy) dataset slots for reuse |
 
+### Sync Instance Metrics (Physical Mode)
+
+These metrics are only available when DBLab is running in physical mode with a sync instance enabled. They track the WAL replay status of the sync instance.
+
+| Metric Name | Type | Labels | Description |
+|-------------|------|--------|-------------|
+| `dblab_sync_status` | Gauge | `status` | Status of the sync instance (1=active for status code) |
+| `dblab_sync_wal_lag_seconds` | Gauge | - | WAL replay lag in seconds for the sync instance |
+| `dblab_sync_uptime_seconds` | Gauge | - | Uptime of the sync instance in seconds |
+| `dblab_sync_last_replayed_timestamp` | Gauge | - | Unix timestamp of the last replayed transaction |
+
 ### Observability Metrics
 
 These metrics help monitor the health of the metrics collection system itself.
@@ -146,6 +157,18 @@ dblab_clones_by_status
 time() - dblab_scrape_success_timestamp
 ```
 
+### WAL Replay Lag (Physical Mode)
+
+```promql
+dblab_sync_wal_lag_seconds
+```
+
+### Time Since Last Replayed Transaction
+
+```promql
+time() - dblab_sync_last_replayed_timestamp
+```
+
 ## Alerting Examples
 
 ### Low Disk Space Alert
@@ -200,6 +223,32 @@ time() - dblab_scrape_success_timestamp
     description: "DBLab metrics have not been updated for more than 5 minutes"
 ```
 
+### High WAL Replay Lag Alert (Physical Mode)
+
+```yaml
+- alert: DBLabHighWALLag
+  expr: dblab_sync_wal_lag_seconds > 3600
+  for: 10m
+  labels:
+    severity: warning
+  annotations:
+    summary: "DBLab sync instance has high WAL lag"
+    description: "DBLab sync instance WAL replay is {{ $value | humanizeDuration }} behind"
+```
+
+### Sync Instance Down Alert (Physical Mode)
+
+```yaml
+- alert: DBLabSyncDown
+  expr: dblab_sync_status{status="down"} == 1 or dblab_sync_status{status="error"} == 1
+  for: 5m
+  labels:
+    severity: critical
+  annotations:
+    summary: "DBLab sync instance is down"
+    description: "DBLab sync instance is not healthy"
+```
+
 ## OpenTelemetry Integration
 
 DBLab metrics can be exported to OpenTelemetry-compatible backends using the OpenTelemetry Collector. This allows you to send metrics to Grafana Cloud, Datadog, New Relic, and other observability platforms.

engine/internal/srv/metrics/collector.go

@@ -27,6 +27,9 @@ const (
 	dockerStatsTimeout   = 5 * time.Second
 	dockerStatsWorkers   = 10
 	cpuNoData            = -1.0
+
+	// postgresTimestampFormat is the default text representation of PostgreSQL's timestamp with time zone.
+	postgresTimestampFormat = "2006-01-02 15:04:05.999999-07"
 )
 
 // CloningService defines the interface for clone and snapshot operations needed by metrics.
@@ -39,6 +42,7 @@ type CloningService interface {
 type RetrievalService interface {
 	GetRetrievalMode() models.RetrievalMode
 	GetRetrievalStatus() models.RetrievalStatus
+	ReportSyncStatus(ctx context.Context) (*models.Sync, error)
 }
 
 // PoolService defines the interface for pool operations needed by metrics.
@@ -152,6 +156,7 @@ func (c *Collector) collectAll(ctx context.Context) {
 	c.collectCloneMetrics(ctx)
 	c.collectSnapshotMetrics()
 	c.collectBranchMetrics()
+	c.collectSyncMetrics(ctx)
 
 	c.metrics.ScrapeDurationSeconds.Set(time.Since(start).Seconds())
 	c.metrics.ScrapeSuccessTimestamp.Set(float64(time.Now().Unix()))
@@ -564,3 +569,57 @@ func (c *Collector) collectBranchMetrics() {
 
 	c.metrics.BranchesTotal.Set(float64(totalBranches))
 }
+
+func (c *Collector) collectSyncMetrics(ctx context.Context) {
+	if c.retrieval.GetRetrievalMode() != models.Physical {
+		return
+	}
+
+	syncState, err := c.retrieval.ReportSyncStatus(ctx)
+	if err != nil {
+		log.Dbg("failed to get sync status for metrics:", err)
+		c.metrics.ScrapeErrorsTotal.Inc()
+		c.setSyncStatusNotAvailable()
+
+		return
+	}
+
+	if syncState == nil {
+		c.setSyncStatusNotAvailable()
+
+		return
+	}
+
+	c.metrics.SyncStatus.Reset()
+	c.metrics.SyncStatus.WithLabelValues(string(syncState.Status.Code)).Set(1)
+	c.metrics.SyncWALLagSeconds.Set(float64(syncState.ReplicationLag))
+	c.metrics.SyncUptimeSeconds.Set(float64(syncState.ReplicationUptime))
+
+	if ts := parseTimestamp(syncState.LastReplayedLsnAt); ts != nil {
+		c.metrics.SyncLastReplayedAt.Set(float64(ts.Unix()))
+	}
+}
+
+func (c *Collector) setSyncStatusNotAvailable() {
+	c.metrics.SyncStatus.Reset()
+	c.metrics.SyncStatus.WithLabelValues(string(models.SyncStatusNotAvailable)).Set(1)
+	c.metrics.SyncUptimeSeconds.Set(0)
+}
+
+func parseTimestamp(value string) *time.Time {
+	if value == "" {
+		return nil
+	}
+
+	formats := []string{time.RFC3339Nano, postgresTimestampFormat}
+
+	for _, format := range formats {
+		if ts, err := time.Parse(format, value); err == nil {
+			return &ts
+		}
+	}
+
+	log.Dbg("failed to parse timestamp:", value)
+
+	return nil
+}

engine/internal/srv/metrics/collector_test.go

@@ -16,6 +16,7 @@ import (
 	"github.com/docker/docker/api/types/container"
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/client_golang/prometheus/promhttp"
+	dto "github.com/prometheus/client_model/go"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 
@@ -36,6 +37,10 @@ func (m *mockRetrievalService) GetRetrievalMode() models.RetrievalMode { return
 
 func (m *mockRetrievalService) GetRetrievalStatus() models.RetrievalStatus { return models.Inactive }
 
+func (m *mockRetrievalService) ReportSyncStatus(_ context.Context) (*models.Sync, error) {
+	return nil, nil
+}
+
 type mockPoolService struct{}
 
 func (m *mockPoolService) GetFSManagerList() []pool.FSManager { return nil }
@@ -446,3 +451,175 @@ func TestCalculateCPUPercent_Concurrent(t *testing.T) {
 
 	assert.Len(t, c.prevCPUStats, cloneCount)
 }
+
+type configMockRetrievalService struct {
+	mode       models.RetrievalMode
+	syncStatus *models.Sync
+	syncErr    error
+}
+
+func (m *configMockRetrievalService) GetRetrievalMode() models.RetrievalMode { return m.mode }
+
+func (m *configMockRetrievalService) GetRetrievalStatus() models.RetrievalStatus {
+	return models.Inactive
+}
+func (m *configMockRetrievalService) ReportSyncStatus(_ context.Context) (*models.Sync, error) {
+	return m.syncStatus, m.syncErr
+}
+
+func TestCollectSyncMetrics(t *testing.T) {
+	t.Run("non-physical mode skips collection", func(t *testing.T) {
+		m := NewMetrics()
+		retrieval := &configMockRetrievalService{mode: models.Logical}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncWALLagSeconds))
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncUptimeSeconds))
+	})
+
+	t.Run("error from ReportSyncStatus sets not available status", func(t *testing.T) {
+		m := NewMetrics()
+		retrieval := &configMockRetrievalService{mode: models.Physical, syncErr: fmt.Errorf("connection failed")}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		initialErrors := getCounterValue(m.ScrapeErrorsTotal)
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, initialErrors+1, getCounterValue(m.ScrapeErrorsTotal))
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncUptimeSeconds))
+		assert.Equal(t, float64(1), getGaugeVecValue(m.SyncStatus, string(models.SyncStatusNotAvailable)))
+	})
+
+	t.Run("nil sync status sets not available status and resets uptime", func(t *testing.T) {
+		m := NewMetrics()
+		retrieval := &configMockRetrievalService{mode: models.Physical, syncStatus: nil}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncWALLagSeconds))
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncUptimeSeconds))
+		assert.Equal(t, float64(1), getGaugeVecValue(m.SyncStatus, string(models.SyncStatusNotAvailable)))
+	})
+
+	t.Run("successful collection sets all metrics", func(t *testing.T) {
+		m := NewMetrics()
+		syncStatus := &models.Sync{
+			Status:            models.Status{Code: models.StatusOK},
+			ReplicationLag:    120,
+			ReplicationUptime: 3600,
+			LastReplayedLsnAt: "2025-01-15T10:30:00.123456789Z",
+		}
+		retrieval := &configMockRetrievalService{mode: models.Physical, syncStatus: syncStatus}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, float64(120), getGaugeValue(m.SyncWALLagSeconds))
+		assert.Equal(t, float64(3600), getGaugeValue(m.SyncUptimeSeconds))
+		assert.Equal(t, float64(1), getGaugeVecValue(m.SyncStatus, "OK"))
+		assert.Greater(t, getGaugeValue(m.SyncLastReplayedAt), float64(0))
+	})
+
+	t.Run("postgres timestamp format is parsed correctly", func(t *testing.T) {
+		m := NewMetrics()
+		syncStatus := &models.Sync{
+			Status:            models.Status{Code: models.StatusOK},
+			ReplicationLag:    60,
+			ReplicationUptime: 1800,
+			LastReplayedLsnAt: "2025-01-15 10:30:00.123456+00",
+		}
+		retrieval := &configMockRetrievalService{mode: models.Physical, syncStatus: syncStatus}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, float64(60), getGaugeValue(m.SyncWALLagSeconds))
+		assert.Equal(t, float64(1800), getGaugeValue(m.SyncUptimeSeconds))
+		assert.Greater(t, getGaugeValue(m.SyncLastReplayedAt), float64(0))
+	})
+
+	t.Run("empty timestamp leaves metric unchanged", func(t *testing.T) {
+		m := NewMetrics()
+		syncStatus := &models.Sync{
+			Status:            models.Status{Code: models.StatusOK},
+			ReplicationLag:    30,
+			ReplicationUptime: 900,
+			LastReplayedLsnAt: "",
+		}
+		retrieval := &configMockRetrievalService{mode: models.Physical, syncStatus: syncStatus}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, float64(30), getGaugeValue(m.SyncWALLagSeconds))
+		assert.Equal(t, float64(900), getGaugeValue(m.SyncUptimeSeconds))
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncLastReplayedAt))
+	})
+
+	t.Run("unparseable timestamp leaves metric unchanged", func(t *testing.T) {
+		m := NewMetrics()
+		syncStatus := &models.Sync{
+			Status:            models.Status{Code: models.StatusOK},
+			ReplicationLag:    45,
+			ReplicationUptime: 1200,
+			LastReplayedLsnAt: "invalid-timestamp",
+		}
+		retrieval := &configMockRetrievalService{mode: models.Physical, syncStatus: syncStatus}
+		c := &Collector{metrics: m, retrieval: retrieval, prevCPUStats: make(map[string]containerCPUState)}
+
+		c.collectSyncMetrics(context.Background())
+
+		assert.Equal(t, float64(45), getGaugeValue(m.SyncWALLagSeconds))
+		assert.Equal(t, float64(1200), getGaugeValue(m.SyncUptimeSeconds))
+		assert.Equal(t, float64(0), getGaugeValue(m.SyncLastReplayedAt))
+	})
+}
+
+func getGaugeValue(g prometheus.Gauge) float64 {
+	ch := make(chan prometheus.Metric, 1)
+	g.Collect(ch)
+	select {
+	case m := <-ch:
+		var metric dto.Metric
+		_ = m.Write(&metric)
+		if metric.Gauge != nil {
+			return *metric.Gauge.Value
+		}
+	default:
+	}
+	return 0
+}
+
+func getGaugeVecValue(g *prometheus.GaugeVec, label string) float64 {
+	ch := make(chan prometheus.Metric, 10)
+	g.Collect(ch)
+	close(ch)
+	for m := range ch {
+		var metric dto.Metric
+		_ = m.Write(&metric)
+		for _, lp := range metric.Label {
+			if lp.GetValue() == label && metric.Gauge != nil {
+				return *metric.Gauge.Value
+			}
+		}
+	}
+	return 0
+}
+
+func getCounterValue(c prometheus.Counter) float64 {
+	ch := make(chan prometheus.Metric, 1)
+	c.Collect(ch)
+	select {
+	case m := <-ch:
+		var metric dto.Metric
+		_ = m.Write(&metric)
+		if metric.Counter != nil {
+			return *metric.Counter.Value
+		}
+	default:
+	}
+	return 0
+}