db: fix RangeDeletionBytesEstimate

Fix RangeDeletionBytesEstimate to incorporate an estimate of the referenced
blob values.

Fix #5501.

Author: jbowens

table_stats.go

@@ -11,6 +11,7 @@ import (
 	"slices"
 	"time"
 
+	"github.com/cockroachdb/crlib/crmath"
 	"github.com/cockroachdb/crlib/crtime"
 	"github.com/cockroachdb/errors"
 	"github.com/cockroachdb/pebble/internal/base"
@@ -487,7 +488,7 @@ func (d *DB) loadTableRangeDelStats(
 		// the size of the range key block relative to the overall size of the
 		// table is expected to be small.
 		if level == numLevels-1 && meta.SmallestSeqNum < maxRangeDeleteSeqNum {
-			size, err := r.EstimateDiskUsage(start, end, env, meta.IterTransforms())
+			size, err := estimateDiskUsageInTableAndBlobReferences(r, start, end, env, meta)
 			if err != nil {
 				return nil, 0, err
 			}
@@ -689,7 +690,7 @@ func (d *DB) estimateReclaimedSizeBeneath(
 				// The range fully contains the file, so skip looking it up in table
 				// cache/looking at its indexes and add the full file size.
 				if updateEstimates {
-					estimate += file.Size
+					estimate += file.Size + file.EstimatedReferenceSize()
 				}
 				if updateHints && hintSeqNum > file.SmallestSeqNum {
 					hintSeqNum = file.SmallestSeqNum
@@ -705,7 +706,7 @@ func (d *DB) estimateReclaimedSizeBeneath(
 				var size uint64
 				err := d.fileCache.withReader(ctx, block.NoReadEnv, file,
 					func(r *sstable.Reader, env sstable.ReadEnv) (err error) {
-						size, err = r.EstimateDiskUsage(start, end, env, file.IterTransforms())
+						size, err = estimateDiskUsageInTableAndBlobReferences(r, start, end, env, file)
 						return err
 					})
 				if err != nil {
@@ -752,6 +753,24 @@ func sanityCheckStats(
 	}
 }
 
+// estimateDiskUsageInTableAndBlobReferences estimates the disk usage within a
+// sstable and its referenced values. The size of blob files is computed using
+// linear interpolation.
+func estimateDiskUsageInTableAndBlobReferences(
+	r *sstable.Reader, start, end []byte, env sstable.ReadEnv, meta *manifest.TableMetadata,
+) (uint64, error) {
+	size, err := r.EstimateDiskUsage(start, end, env, meta.IterTransforms())
+	if err != nil {
+		return 0, err
+	}
+
+	estimatedTableSize := max(size, 1)
+	originalTableSize := max(meta.Size, 1)
+	referenceSize := crmath.ScaleUint64(meta.EstimatedReferenceSize(),
+		estimatedTableSize, originalTableSize)
+	return size + referenceSize, nil
+}
+
 // maybeSetStatsFromProperties sets the table backing properties and attempts to
 // set the table stats from the properties, for a table that was created by an
 // ingestion or compaction.

testdata/compaction/value_separation

@@ -981,5 +981,5 @@ num-entries: 1
 num-deletions: 1
 num-range-key-sets: 0
 point-deletions-bytes-estimate: 0
-range-deletions-bytes-estimate: 797
+range-deletions-bytes-estimate: 4987
 compression: None:79