builtins: Add ApproximateSpanBytes column to crdb_internal.sstable_metrics()

Added a column to indicate how many bytes the given keyspan makes up.

Informs: cockroachdb#94659
Release note: None

Author: Rahul Aggarwal

docs/generated/sql/functions.md

@@ -3308,7 +3308,7 @@ table. Returns an error if validation fails.</p>
 </span></td><td>Volatile</td></tr>
 <tr><td><a name="crdb_internal.set_vmodule"></a><code>crdb_internal.set_vmodule(vmodule_string: <a href="string.html">string</a>) &rarr; <a href="int.html">int</a></code></td><td><span class="funcdesc"><p>Set the equivalent of the <code>--vmodule</code> flag on the gateway node processing this request; it affords control over the logging verbosity of different files. Example syntax: <code>crdb_internal.set_vmodule('recordio=2,file=1,gfs*=3')</code>. Reset with: <code>crdb_internal.set_vmodule('')</code>. Raising the verbosity can severely affect performance.</p>
 </span></td><td>Volatile</td></tr>
-<tr><td><a name="crdb_internal.sstable_metrics"></a><code>crdb_internal.sstable_metrics(node_id: <a href="int.html">int</a>, store_id: <a href="int.html">int</a>, start_key: <a href="bytes.html">bytes</a>, end_key: <a href="bytes.html">bytes</a>) &rarr; tuple{int AS node_id,, int AS store_id, int AS level, int AS file_num, jsonb AS metrics}</code></td><td><span class="funcdesc"><p>Returns statistics for the sstables containing keys in the range start_key and end_key for the provided node id.</p>
+<tr><td><a name="crdb_internal.sstable_metrics"></a><code>crdb_internal.sstable_metrics(node_id: <a href="int.html">int</a>, store_id: <a href="int.html">int</a>, start_key: <a href="bytes.html">bytes</a>, end_key: <a href="bytes.html">bytes</a>) &rarr; tuple{int AS node_id,, int AS store_id, int AS level, int AS file_num, bytes AS approximate_span_<a href="bytes.html">bytes</a>, jsonb AS metrics}</code></td><td><span class="funcdesc"><p>Returns statistics for the sstables containing keys in the range start_key and end_key for the provided node id.</p>
 </span></td><td>Stable</td></tr>
 <tr><td><a name="crdb_internal.table_span"></a><code>crdb_internal.table_span(table_id: <a href="int.html">int</a>) &rarr; <a href="bytes.html">bytes</a>[]</code></td><td><span class="funcdesc"><p>This function returns the span that contains the keys for the given table.</p>
 </span></td><td>Leakproof</td></tr>

pkg/sql/sem/builtins/fixed_oids.go

@@ -2424,7 +2424,7 @@ var builtinOidsArray = []string{
 	2451: `st_asmvtgeom(geometry: geometry, bbox: box2d) -> geometry`,
 	2452: `crdb_internal.repaired_descriptor(descriptor: bytes, valid_descriptor_ids: int[], valid_job_ids: int[]) -> bytes`,
 	2453: `crdb_internal.reset_activity_tables() -> bool`,
-	2454: `crdb_internal.sstable_metrics(node_id: int, store_id: int, start_key: bytes, end_key: bytes) -> tuple{int AS node_id,, int AS store_id, int AS level, int AS file_num, jsonb AS metrics}`,
+	2454: `crdb_internal.sstable_metrics(node_id: int, store_id: int, start_key: bytes, end_key: bytes) -> tuple{int AS node_id,, int AS store_id, int AS level, int AS file_num, bytes AS approximate_span_bytes, jsonb AS metrics}`,
 	2455: `crdb_internal.repair_catalog_corruption(descriptor_id: int, corruption: string) -> bool`,
 	2456: `crdb_internal.merge_aggregated_stmt_metadata(input: jsonb[]) -> jsonb`,
 }

pkg/sql/sem/builtins/generator_builtins.go

@@ -3125,8 +3125,8 @@ func (tssi *tableSpanStatsIterator) ResolvedType() *types.T {
 }
 
 var tableMetricsGeneratorType = types.MakeLabeledTuple(
-	[]*types.T{types.Int, types.Int, types.Int, types.Int, types.Json},
-	[]string{"node_id,", "store_id", "level", "file_num", "metrics"},
+	[]*types.T{types.Int, types.Int, types.Int, types.Int, types.Bytes, types.Json},
+	[]string{"node_id,", "store_id", "level", "file_num", "approximate_span_bytes", "metrics"},
 )
 
 // tableMetricsIterator implements tree.ValueGenerator; it returns a set of
@@ -3186,6 +3186,7 @@ func (tmi *tableMetricsIterator) Values() (tree.Datums, error) {
 		tree.NewDInt(tree.DInt(tmi.storeID)),
 		tree.NewDInt(tree.DInt(metricsInfo.Level)),
 		tree.NewDInt(tree.DInt(metricsInfo.TableID)),
+		tree.NewDBytes(tree.DBytes(metricsInfo.ApproximateSpanBytes)),
 		metricsJson,
 	}, nil
 }

pkg/sql/sem/builtins/generator_builtins_test.go

@@ -90,6 +90,7 @@ func TestGetSSTableMetricsMultiNode(t *testing.T) {
 	var storeID int
 	var level int
 	var fileNum int
+	var approximateSpanBytes []byte
 	var metrics []byte
 
 	for idx, id := range tc.NodeIDs() {
@@ -121,7 +122,7 @@ func TestGetSSTableMetricsMultiNode(t *testing.T) {
 			nodeIDArg, storeIDArg))
 
 		for rows.Next() {
-			require.NoError(t, rows.Scan(&nodeID, &storeID, &level, &fileNum, &metrics))
+			require.NoError(t, rows.Scan(&nodeID, &storeID, &level, &fileNum, &approximateSpanBytes, &metrics))
 			require.NoError(t, json.Unmarshal(metrics, &enginepb.SSTableMetricsInfo{}))
 			require.Equal(t, nodeID, nodeIDArg)
 			require.Equal(t, storeID, storeIDArg)
@@ -137,7 +138,7 @@ func TestGetSSTableMetricsMultiNode(t *testing.T) {
 			nodeIDArg, storeIDArg))
 
 		for rows.Next() {
-			require.NoError(t, rows.Scan(&nodeID, &storeID, &level, &fileNum, &metrics))
+			require.NoError(t, rows.Scan(&nodeID, &storeID, &level, &fileNum, &approximateSpanBytes, &metrics))
 			require.NoError(t, json.Unmarshal(metrics, &enginepb.SSTableMetricsInfo{}))
 			require.Equal(t, nodeID, nodeIDArg)
 			require.Equal(t, storeID, storeIDArg)
@@ -183,10 +184,11 @@ func TestGetSSTableMetricsSingleNode(t *testing.T) {
 	var storeID int
 	var level int
 	var fileNum int
+	var approximateSpanBytes []byte
 	var metrics []byte
 
 	for rows.Next() {
-		require.NoError(t, rows.Scan(&nodeID, &storeID, &level, &fileNum, &metrics))
+		require.NoError(t, rows.Scan(&nodeID, &storeID, &level, &fileNum, &approximateSpanBytes, &metrics))
 		require.NoError(t, json.Unmarshal(metrics, &enginepb.SSTableMetricsInfo{}))
 		require.Equal(t, nodeID, nodeIDArg)
 		require.Equal(t, storeID, storeIDArg)

pkg/storage/enginepb/rocksdb.proto

@@ -22,6 +22,8 @@ message SSTableMetricsInfo {
   uint64 table_id = 2 [(gogoproto.customname) = "TableID"];
   // table_info_json contains sstable properties, encoded as JSON
   bytes table_info_json = 3 [(gogoproto.customname) = "TableInfoJSON"];
+  // approximate_span_bytes represents the total number of bytes that overlap the given keyspan
+  bytes approximate_span_bytes = 4 [(gogoproto.customname) = "ApproximateSpanBytes"];
 }
 
 // SSTUserProperties contains the user-added properties of a single sstable.

pkg/storage/pebble.go

@@ -2010,7 +2010,7 @@ func (p *Pebble) PreIngestDelay(ctx context.Context) {
 
 // GetTableMetrics implements the Engine interface.
 func (p *Pebble) GetTableMetrics(start, end roachpb.Key) ([]enginepb.SSTableMetricsInfo, error) {
-	tableInfo, err := p.db.SSTables(pebble.WithKeyRangeFilter(start, end))
+	tableInfo, err := p.db.SSTables(pebble.WithKeyRangeFilter(start, end), pebble.WithProperties(), pebble.WithApproximateSpanBytes())
 
 	if err != nil {
 		return []enginepb.SSTableMetricsInfo{}, err
@@ -2032,7 +2032,8 @@ func (p *Pebble) GetTableMetrics(start, end roachpb.Key) ([]enginepb.SSTableMetr
 			}
 
 			tableID := sstableInfo.TableInfo.FileNum
-			metricsInfo = append(metricsInfo, enginepb.SSTableMetricsInfo{TableID: uint64(tableID), Level: int32(level), TableInfoJSON: marshalTableInfo})
+			approximateSpanBytes := sstableInfo.Properties.UserProperties["approximate-span-bytes"]
+			metricsInfo = append(metricsInfo, enginepb.SSTableMetricsInfo{TableID: uint64(tableID), Level: int32(level), ApproximateSpanBytes: []byte(approximateSpanBytes), TableInfoJSON: marshalTableInfo})
 		}
 	}
 	return metricsInfo, nil