db: add KeyStatistics.LatestKindsCount

Also give one example of how to interpret these statistics. And clarify
that SnapshotPinned* are not counting what they are claiming to count.

db.go

@@ -2749,14 +2749,40 @@ func (d *DB) SetCreatorID(creatorID uint64) error {
 
 // KeyStatistics keeps track of the number of keys that have been pinned by a
 // snapshot as well as counts of the different key kinds in the lsm.
+//
+// One way of using the accumulated stats, when we only have sets and dels,
+// and say the counts are represented as del_count, set_count,
+// del_latest_count, set_latest_count, snapshot_pinned_count.
+//
+//   - del_latest_count + set_latest_count is the set of unique user keys
+//     (unique).
+//
+//   - set_latest_count is the set of live unique user keys (live_unique).
+//
+//   - Garbage is del_count + set_count - live_unique.
+//
+//   - If everything were in the LSM, del_count+set_count-snapshot_pinned_count
+//     would also be the set of unique user keys (note that
+//     snapshot_pinned_count is counting something different -- see comment below).
+//     But snapshot_pinned_count only counts keys in the LSM so the excess here
+//     must be keys in memtables.
 type KeyStatistics struct {
-	// when a compaction determines a key is obsolete, but cannot elide the key
-	// because it's required by an open snapshot.
+	// TODO(sumeer): the SnapshotPinned* are incorrect in that these older
+	// versions can be in a different level. Either fix the accounting or
+	// rename these fields.
+
+	// SnapshotPinnedKeys represents obsolete keys that cannot be elided during
+	// a compaction, because they are required by an open snapshot.
 	SnapshotPinnedKeys int
-	// the total number of bytes of all snapshot pinned keys.
+	// SnapshotPinnedKeysBytes is the total number of bytes of all snapshot
+	// pinned keys.
 	SnapshotPinnedKeysBytes uint64
-	// Note: these fields are currently only populated for point keys (including range deletes).
+	// KindsCount is the count for each kind of key. It includes point keys,
+	// range deletes and range keys.
 	KindsCount [InternalKeyKindMax + 1]int
+	// LatestKindsCount is the count for each kind of key when it is the latest
+	// kind for a user key. It is only populated for point keys.
+	LatestKindsCount [InternalKeyKindMax + 1]int
 }
 
 // LSMKeyStatistics is used by DB.ScanStatistics.
@@ -2801,7 +2827,8 @@ func (d *DB) ScanStatistics(
 			// pinned by a snapshot.
 			size := uint64(key.Size())
 			kind := key.Kind()
-			if iterInfo.Kind == IteratorLevelLSM && d.equal(prevKey.UserKey, key.UserKey) {
+			sameKey := d.equal(prevKey.UserKey, key.UserKey)
+			if iterInfo.Kind == IteratorLevelLSM && sameKey {
 				stats.Levels[iterInfo.Level].SnapshotPinnedKeys++
 				stats.Levels[iterInfo.Level].SnapshotPinnedKeysBytes += size
 				stats.Accumulated.SnapshotPinnedKeys++
@@ -2810,6 +2837,12 @@ func (d *DB) ScanStatistics(
 			if iterInfo.Kind == IteratorLevelLSM {
 				stats.Levels[iterInfo.Level].KindsCount[kind]++
 			}
+			if !sameKey {
+				if iterInfo.Kind == IteratorLevelLSM {
+					stats.Levels[iterInfo.Level].LatestKindsCount[kind]++
+				}
+				stats.Accumulated.LatestKindsCount[kind]++
+			}
 
 			stats.Accumulated.KindsCount[kind]++
 			prevKey.CopyFrom(*key)

scan_internal_test.go

@@ -182,6 +182,9 @@ func TestScanStatistics(t *testing.T) {
 				}
 				for _, kind := range keyKindsToDisplay {
 					fmt.Fprintf(&b, "  %s key count: %d\n", kind.String(), stats.Levels[lvl].KindsCount[kind])
+					if stats.Levels[lvl].LatestKindsCount[kind] > 0 {
+						fmt.Fprintf(&b, "  %s latest count: %d\n", kind.String(), stats.Levels[lvl].LatestKindsCount[kind])
+					}
 				}
 			}
 
@@ -191,6 +194,9 @@ func TestScanStatistics(t *testing.T) {
 			}
 			for _, kind := range keyKindsToDisplay {
 				fmt.Fprintf(&b, "  %s key count: %d\n", kind.String(), stats.Accumulated.KindsCount[kind])
+				if stats.Accumulated.LatestKindsCount[kind] > 0 {
+					fmt.Fprintf(&b, "  %s latest count: %d\n", kind.String(), stats.Accumulated.LatestKindsCount[kind])
+				}
 			}
 			return b.String()
 		default:

testdata/scan_statistics

@@ -12,6 +12,7 @@ scan-statistics lower=b upper=f keys=(SET)
 ----
 Aggregate:
   SET key count: 2
+  SET latest count: 2
 
 flush
 ----
@@ -20,15 +21,19 @@ scan-statistics lower=b upper=e keys=(SET) levels=(0)
 ----
 Level 0:
   SET key count: 1
+  SET latest count: 1
 Aggregate:
   SET key count: 1
+  SET latest count: 1
 
 scan-statistics lower=b upper=f keys=(SET) levels=(0)
 ----
 Level 0:
   SET key count: 2
+  SET latest count: 2
 Aggregate:
   SET key count: 2
+  SET latest count: 2
 
 scan-statistics lower=f upper=l keys=(SET)
 ----
@@ -49,9 +54,11 @@ scan-statistics lower=b upper=f keys=(SET, DEL) levels=(0)
 Level 0:
   SET key count: 2
   DEL key count: 2
+  DEL latest count: 2
 Aggregate:
   SET key count: 2
   DEL key count: 2
+  DEL latest count: 2
 
 reset
 ----
@@ -81,8 +88,10 @@ scan-statistics lower=b upper=f keys=(SET) levels=(6)
 ----
 Level 6:
   SET key count: 1
+  SET latest count: 1
 Aggregate:
   SET key count: 1
+  SET latest count: 1
 
 batch commit
 set c a
@@ -96,10 +105,13 @@ scan-statistics lower=b upper=f keys=(SET) levels=(0, 6)
 ----
 Level 0:
   SET key count: 1
+  SET latest count: 1
 Level 6:
   SET key count: 1
+  SET latest count: 1
 Aggregate:
   SET key count: 2
+  SET latest count: 2
 
 reset
 ----