sstable/blob: track MVCCGarbageBytes in FileWriterStats

We now track the amount of uncompressed, likely MVCC garbage bytes in our
`FileWriterStats`.

Author: annrpom

blob_rewrite_test.go

@@ -270,7 +270,7 @@ func TestBlobRewriteRandomized(t *testing.T) {
 			if rng.IntN(20) == 0 {
 				blobWriter.FlushForTesting()
 			}
-			handles[i] = blobWriter.AddValue(values[i])
+			handles[i] = blobWriter.AddValue(values[i], false /* isLikelyMVCCGarbage */)
 		}
 		stats, err := blobWriter.Close()
 		require.NoError(t, err)

file_cache_test.go

@@ -239,7 +239,7 @@ func (t *fileCacheTest) newTestHandle() (*fileCacheHandle, *fileCacheTestFS) {
 			t.Fatal(err)
 		}
 		bw := blob.NewFileWriter(fn, w, blob.FileWriterOptions{})
-		_ = bw.AddValue(xxx[:fn])
+		_ = bw.AddValue(xxx[:fn], false /* isLikelyMVCCGarbage */)
 		if _, err := bw.Close(); err != nil {
 			t.Fatal(err)
 		}

internal/blobtest/handles.go

@@ -261,14 +261,14 @@ func (bv *Values) WriteFiles(
 					BlockID:    handle.BlockID,
 					ValueID:    blob.BlockValueID(prevID),
 					ValueLen:   12,
-				}))
+				}), false /* isLikelyMVCCGarbage */)
 				prevID++
 			}
 
 			if value, ok := bv.trackedHandles[handle]; ok {
-				writer.AddValue([]byte(value))
+				writer.AddValue([]byte(value), false /* isLikelyMVCCGarbage */)
 			} else {
-				writer.AddValue(deriveValueFromHandle(handle))
+				writer.AddValue(deriveValueFromHandle(handle), false /* isLikelyMVCCGarbage */)
 			}
 		}
 		fileStats, err := writer.Close()

sstable/blob/blob.go

@@ -11,6 +11,7 @@ import (
 	"fmt"
 	"sync"
 
+	"github.com/cockroachdb/crlib/crhumanize"
 	"github.com/cockroachdb/errors"
 	"github.com/cockroachdb/pebble/internal/base"
 	"github.com/cockroachdb/pebble/internal/crc"
@@ -125,13 +126,14 @@ type FileWriterStats struct {
 	UncompressedValueBytes uint64
 	FileLen                uint64
 	Properties             FileProperties
+	MVCCGarbageBytes       uint64
 }
 
 // String implements the fmt.Stringer interface.
 func (s FileWriterStats) String() string {
 	var buf bytes.Buffer
-	fmt.Fprintf(&buf, "{BlockCount: %d, ValueCount: %d, UncompressedValueBytes: %d, FileLen: %d}",
-		s.BlockCount, s.ValueCount, s.UncompressedValueBytes, s.FileLen)
+	fmt.Fprintf(&buf, "{BlockCount: %d, ValueCount: %d, UncompressedValueBytes: %d (%s MVCCGarbage), FileLen: %d}",
+		s.BlockCount, s.ValueCount, s.UncompressedValueBytes, crhumanize.Percent(s.MVCCGarbageBytes, s.UncompressedValueBytes), s.FileLen)
 	return buf.String()
 }
 
@@ -187,14 +189,17 @@ var writerPool = sync.Pool{
 
 // AddValue adds the provided value to the blob file, returning a Handle
 // identifying the location of the value.
-func (w *FileWriter) AddValue(v []byte) Handle {
+func (w *FileWriter) AddValue(v []byte, isLikelyMVCCGarbage bool) Handle {
 	// Determine if we should first flush the block.
 	if sz := w.valuesEncoder.size(); w.flushGov.ShouldFlush(sz, sz+len(v)) {
 		w.flush()
 	}
 	valuesInBlock := w.valuesEncoder.Count()
 	w.stats.ValueCount++
 	w.stats.UncompressedValueBytes += uint64(len(v))
+	if isLikelyMVCCGarbage {
+		w.stats.MVCCGarbageBytes += uint64(len(v))
+	}
 	w.valuesEncoder.AddValue(v)
 	return Handle{
 		BlobFileID: base.BlobFileID(w.fileNum),

sstable/blob/blob_test.go

@@ -32,7 +32,7 @@ func TestBlobWriter(t *testing.T) {
 			obj = &objstorage.MemObj{}
 			w := NewFileWriter(000001, obj, opts)
 			for _, l := range crstrings.Lines(td.Input) {
-				h := w.AddValue([]byte(l))
+				h := w.AddValue([]byte(l), false /* isLikelyMVCCGarbage */)
 				fmt.Fprintf(&buf, "%-25s: %q\n", h, l)
 			}
 			stats, err := w.Close()
@@ -55,7 +55,7 @@ func TestBlobWriter(t *testing.T) {
 					w.beginNewVirtualBlock(BlockID(vBlockID))
 					vBlockID++
 				default:
-					h := w.AddValue([]byte(l))
+					h := w.AddValue([]byte(l), false /* isLikelyMVCCGarbage */)
 					fmt.Fprintf(&buf, "%-25s: %q\n", h, l)
 				}
 			}

sstable/blob/fetcher_test.go

@@ -99,7 +99,7 @@ func TestValueFetcher(t *testing.T) {
 			obj := &objstorage.MemObj{}
 			w := NewFileWriter(base.DiskFileNum(fileNum), obj, opts)
 			for _, l := range crstrings.Lines(td.Input) {
-				h := w.AddValue([]byte(l))
+				h := w.AddValue([]byte(l), false /* isLikelyMVCCGarbage */)
 				fmt.Fprintf(&buf, "%-25s: %q\n", h, l)
 			}
 			stats, err := w.Close()
@@ -191,7 +191,7 @@ func TestValueFetcherRetrieveRandomized(t *testing.T) {
 	for v := 4 << 20; v > 0; {
 		n := testutils.RandIntInRange(rng, 1, 4096)
 		val := testutils.RandBytes(rng, n)
-		h := w.AddValue(val)
+		h := w.AddValue(val, false /* isLikelyMVCCGarbage */)
 		handles = append(handles, h)
 		values = append(values, val)
 		v -= n
@@ -269,7 +269,7 @@ func benchmarkValueFetcherRetrieve(b *testing.B, valueSize int, cacheSize int64)
 	var handles []Handle
 	for v := valueSize; v > 0; {
 		n := testutils.RandIntInRange(rng, 1, 4096)
-		h := w.AddValue(testutils.RandBytes(rng, n))
+		h := w.AddValue(testutils.RandBytes(rng, n), false /* isLikelyMVCCGarbage */)
 		handles = append(handles, h)
 		v -= n
 	}

testdata/blob_rewrite

@@ -24,7 +24,7 @@ close-output
 # sync-data: 000002.blob
 # close: 000002.blob
 Blob file created: 000002 size:[105 (105B)] vals:[15 (15B)]
-{BlockCount: 1, ValueCount: 2, UncompressedValueBytes: 15, FileLen: 105}
+{BlockCount: 1, ValueCount: 2, UncompressedValueBytes: 15 (0% MVCCGarbage), FileLen: 105}
 blobrefs:[
  0: B000002 15
 ]
@@ -76,4 +76,4 @@ rewrite-blob 000002 000001 000003
 Successfully rewrote blob file 000002 to 000004
 Input SSTables: [000001 000003]
 SSTables with blob references: 2
-{BlockCount: 1, ValueCount: 2, UncompressedValueBytes: 15, FileLen: 106}
+{BlockCount: 1, ValueCount: 2, UncompressedValueBytes: 15 (0% MVCCGarbage), FileLen: 106}

testdata/value_separation_policy

@@ -48,7 +48,7 @@ close-output
 # sync-data: 000002.blob
 # close: 000002.blob
 Blob file created: 000002 size:[120 (120B)] vals:[29 (29B)]
-{BlockCount: 1, ValueCount: 3, UncompressedValueBytes: 29, FileLen: 120}
+{BlockCount: 1, ValueCount: 3, UncompressedValueBytes: 29 (0% MVCCGarbage), FileLen: 120}
 blobrefs:[
  0: B000002 29
 ]

value_separation.go

@@ -296,7 +296,7 @@ func (vs *writeNewBlobFiles) Add(
 	}
 
 	// Append the value to the blob file.
-	handle := vs.writer.AddValue(v)
+	handle := vs.writer.AddValue(v, isLikelyMVCCGarbage)
 
 	// Write the key and the handle to the sstable. We need to map the
 	// blob.Handle into a blob.InlineHandle. Everything is copied verbatim,
@@ -341,7 +341,8 @@ func (vs *writeNewBlobFiles) FinishOutput() (compact.ValueSeparationMetadata, er
 
 	return compact.ValueSeparationMetadata{
 		BlobReferences: manifest.BlobReferences{
-			manifest.MakeBlobReference(base.BlobFileID(vs.objMeta.DiskFileNum), stats.UncompressedValueBytes, stats.UncompressedValueBytes, meta),
+			manifest.MakeBlobReference(base.BlobFileID(vs.objMeta.DiskFileNum),
+				stats.UncompressedValueBytes, stats.UncompressedValueBytes, meta),
 		},
 		BlobReferenceSize:  stats.UncompressedValueBytes,
 		BlobReferenceDepth: 1,