Merge pull request #9819 from gyuho/mmm

*: promote db size metric to "'etcd" namespace

Author: gyuho

Documentation/op-guide/grafana.json

@@ -341,7 +341,7 @@
                         "steppedLine": false,
                         "targets": [
                             {
-                                "expr": "etcd_debugging_mvcc_db_total_size_in_bytes{job=\"$cluster\"}",
+                                "expr": "etcd_mvcc_db_total_size_in_bytes{job=\"$cluster\"}",
                                 "hide": false,
                                 "interval": "",
                                 "intervalFactor": 2,

Documentation/op-guide/maintenance.md

@@ -149,7 +149,9 @@ $ ETCDCTL_API=3 etcdctl put newkey 123
 OK
 ```
 
-The metric `etcd_debugging_mvcc_db_total_size_in_use_in_bytes` indicates the actual database usage after a history compaction, while `etcd_debugging_mvcc_db_total_size_in_bytes` shows the database size including free space waiting for defragmentation. The latter increases only when the former is close to it, meaning when both of these metrics are close to the quota, a history compaction is required to avoid triggering the space quota.
+The metric `etcd_mvcc_db_total_size_in_use_in_bytes` indicates the actual database usage after a history compaction, while `etcd_debugging_mvcc_db_total_size_in_bytes` shows the database size including free space waiting for defragmentation. The latter increases only when the former is close to it, meaning when both of these metrics are close to the quota, a history compaction is required to avoid triggering the space quota.
+
+`etcd_debugging_mvcc_db_total_size_in_bytes` is renamed to `etcd_mvcc_db_total_size_in_bytes` from v3.4.
 
 ## Snapshot backup
 

Documentation/upgrades/upgrade_3_4.md

@@ -47,6 +47,19 @@ OK
 +etcd --peer-trusted-ca-file ca-peer.crt
 ```
 
+#### Promote `etcd_debugging_mvcc_db_total_size_in_bytes` Prometheus metrics
+
+v3.4 promotes `etcd_debugging_mvcc_db_total_size_in_bytes` Prometheus metrics to `etcd_mvcc_db_total_size_in_bytes`, in order to encourage etcd storage monitoring.
+
+`etcd_debugging_mvcc_db_total_size_in_bytes` is still served in v3.4 for backward compatibilities. It will be completely deprecated in v3.5.
+
+```diff
+-etcd_debugging_mvcc_db_total_size_in_bytes
++etcd_mvcc_db_total_size_in_bytes
+```
+
+Note that `etcd_debugging_*` namespace metrics have been marked as experimental. As we improve monitoring guide, we will promote more metrics.
+
 #### Deprecating `etcd --log-output` flag (now `--log-outputs`)
 
 Rename [`etcd --log-output` to `--log-outputs`](https://github.com/coreos/etcd/pull/9624) to support multiple log outputs. **`etcd --logger=capnslog` does not support multiple log outputs.**

Documentation/upgrades/upgrade_3_5.md

@@ -14,6 +14,17 @@ Before [starting an upgrade](#upgrade-procedure), read through the rest of this
 
 Highlighted breaking changes in 3.5.
 
+#### Deprecate `etcd_debugging_mvcc_db_total_size_in_bytes` Prometheus metrics
+
+v3.4 promoted `etcd_debugging_mvcc_db_total_size_in_bytes` Prometheus metrics to `etcd_mvcc_db_total_size_in_bytes`, in order to encourage etcd storage monitoring. And v3.5 completely deprcates `etcd_debugging_mvcc_db_total_size_in_bytes`.
+
+```diff
+-etcd_debugging_mvcc_db_total_size_in_bytes
++etcd_mvcc_db_total_size_in_bytes
+```
+
+Note that `etcd_debugging_*` namespace metrics have been marked as experimental. As we improve monitoring guide, we will promote more metrics.
+
 #### Deprecated in `etcd --logger capnslog`
 
 v3.4 defaults to `--logger=zap` in order to support multiple log outputs and structured logging.

integration/metrics_test.go

@@ -40,8 +40,16 @@ func TestMetricDbSizeBoot(t *testing.T) {
 	}
 }
 
-// TestMetricDbSizeDefrag checks that the db size metric is set after defrag.
 func TestMetricDbSizeDefrag(t *testing.T) {
+	testMetricDbSizeDefrag(t, "etcd")
+}
+
+func TestMetricDbSizeDefragDebugging(t *testing.T) {
+	testMetricDbSizeDefrag(t, "etcd_debugging")
+}
+
+// testMetricDbSizeDefrag checks that the db size metric is set after defrag.
+func testMetricDbSizeDefrag(t *testing.T, name string) {
 	defer testutil.AfterTest(t)
 	clus := NewClusterV3(t, &ClusterConfig{Size: 1})
 	defer clus.Terminate(t)
@@ -63,7 +71,7 @@ func TestMetricDbSizeDefrag(t *testing.T) {
 	time.Sleep(500 * time.Millisecond)
 
 	expected := numPuts * len(putreq.Value)
-	beforeDefrag, err := clus.Members[0].Metric("etcd_debugging_mvcc_db_total_size_in_bytes")
+	beforeDefrag, err := clus.Members[0].Metric(name + "_mvcc_db_total_size_in_bytes")
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -74,7 +82,7 @@ func TestMetricDbSizeDefrag(t *testing.T) {
 	if bv < expected {
 		t.Fatalf("expected db size greater than %d, got %d", expected, bv)
 	}
-	beforeDefragInUse, err := clus.Members[0].Metric("etcd_debugging_mvcc_db_total_size_in_use_in_bytes")
+	beforeDefragInUse, err := clus.Members[0].Metric("etcd_mvcc_db_total_size_in_use_in_bytes")
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -98,7 +106,7 @@ func TestMetricDbSizeDefrag(t *testing.T) {
 	}
 	time.Sleep(500 * time.Millisecond)
 
-	afterCompactionInUse, err := clus.Members[0].Metric("etcd_debugging_mvcc_db_total_size_in_use_in_bytes")
+	afterCompactionInUse, err := clus.Members[0].Metric("etcd_mvcc_db_total_size_in_use_in_bytes")
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -113,7 +121,7 @@ func TestMetricDbSizeDefrag(t *testing.T) {
 	// defrag should give freed space back to fs
 	mc.Defragment(context.TODO(), &pb.DefragmentRequest{})
 
-	afterDefrag, err := clus.Members[0].Metric("etcd_debugging_mvcc_db_total_size_in_bytes")
+	afterDefrag, err := clus.Members[0].Metric(name + "_mvcc_db_total_size_in_bytes")
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -125,7 +133,7 @@ func TestMetricDbSizeDefrag(t *testing.T) {
 		t.Fatalf("expected less than %d, got %d after defrag", bv, av)
 	}
 
-	afterDefragInUse, err := clus.Members[0].Metric("etcd_debugging_mvcc_db_total_size_in_use_in_bytes")
+	afterDefragInUse, err := clus.Members[0].Metric("etcd_mvcc_db_total_size_in_use_in_bytes")
 	if err != nil {
 		t.Fatal(err)
 	}

mvcc/kvstore.go

@@ -323,6 +323,9 @@ func (s *store) restore() error {
 	reportDbTotalSizeInBytesMu.Lock()
 	reportDbTotalSizeInBytes = func() float64 { return float64(b.Size()) }
 	reportDbTotalSizeInBytesMu.Unlock()
+	reportDbTotalSizeInBytesDebuggingMu.Lock()
+	reportDbTotalSizeInBytesDebugging = func() float64 { return float64(b.Size()) }
+	reportDbTotalSizeInBytesDebuggingMu.Unlock()
 	reportDbTotalSizeInUseInBytesMu.Lock()
 	reportDbTotalSizeInUseInBytes = func() float64 { return float64(b.SizeInUse()) }
 	reportDbTotalSizeInUseInBytesMu.Unlock()

mvcc/metrics.go

@@ -146,7 +146,7 @@ var (
 		})
 
 	dbTotalSize = prometheus.NewGaugeFunc(prometheus.GaugeOpts{
-		Namespace: "etcd_debugging",
+		Namespace: "etcd",
 		Subsystem: "mvcc",
 		Name:      "db_total_size_in_bytes",
 		Help:      "Total size of the underlying database physically allocated in bytes.",
@@ -159,11 +159,28 @@ var (
 	)
 	// overridden by mvcc initialization
 	reportDbTotalSizeInBytesMu sync.RWMutex
-	reportDbTotalSizeInBytes   func() float64 = func() float64 { return 0 }
+	reportDbTotalSizeInBytes   = func() float64 { return 0 }
 
-	dbTotalSizeInUse = prometheus.NewGaugeFunc(prometheus.GaugeOpts{
+	// TODO: remove this in v3.5
+	dbTotalSizeDebugging = prometheus.NewGaugeFunc(prometheus.GaugeOpts{
 		Namespace: "etcd_debugging",
 		Subsystem: "mvcc",
+		Name:      "db_total_size_in_bytes",
+		Help:      "Total size of the underlying database physically allocated in bytes.",
+	},
+		func() float64 {
+			reportDbTotalSizeInBytesDebuggingMu.RLock()
+			defer reportDbTotalSizeInBytesDebuggingMu.RUnlock()
+			return reportDbTotalSizeInBytesDebugging()
+		},
+	)
+	// overridden by mvcc initialization
+	reportDbTotalSizeInBytesDebuggingMu sync.RWMutex
+	reportDbTotalSizeInBytesDebugging   = func() float64 { return 0 }
+
+	dbTotalSizeInUse = prometheus.NewGaugeFunc(prometheus.GaugeOpts{
+		Namespace: "etcd",
+		Subsystem: "mvcc",
 		Name:      "db_total_size_in_use_in_bytes",
 		Help:      "Total size of the underlying database logically in use in bytes.",
 	},
@@ -218,6 +235,7 @@ func init() {
 	prometheus.MustRegister(dbCompactionTotalMs)
 	prometheus.MustRegister(dbCompactionKeysCounter)
 	prometheus.MustRegister(dbTotalSize)
+	prometheus.MustRegister(dbTotalSizeDebugging)
 	prometheus.MustRegister(dbTotalSizeInUse)
 	prometheus.MustRegister(hashSec)
 	prometheus.MustRegister(hashRevSec)