index: Limit index files to 3000 pack files

cmd/restic/cmd_rebuild_index.go

@@ -73,12 +73,12 @@ func rebuildIndex(ctx context.Context, repo restic.Repository, ignorePacks resti
 		return err
 	}
 
-	id, err := idx.Save(ctx, repo, supersedes)
+	ids, err := idx.Save(ctx, repo, supersedes)
 	if err != nil {
 		return errors.Fatalf("unable to save index, last error was: %v", err)
 	}
 
-	Verbosef("saved new index as %v\n", id.Str())
+	Verbosef("saved new indexes as %v\n", ids)
 
 	Verbosef("remove %d old index files\n", len(supersedes))
 

internal/index/index.go

@@ -85,8 +85,8 @@ type blobJSON struct {
 }
 
 type indexJSON struct {
-	Supersedes restic.IDs  `json:"supersedes,omitempty"`
-	Packs      []*packJSON `json:"packs"`
+	Supersedes restic.IDs `json:"supersedes,omitempty"`
+	Packs      []packJSON `json:"packs"`
 }
 
 func loadIndexJSON(ctx context.Context, repo restic.Repository, id restic.ID) (*indexJSON, error) {
@@ -257,26 +257,24 @@ func (idx *Index) FindBlob(h restic.BlobHandle) (result []Location, err error) {
 	return result, nil
 }
 
+const maxEntries = 3000
+
 // Save writes the complete index to the repo.
-func (idx *Index) Save(ctx context.Context, repo restic.Repository, supersedes restic.IDs) (restic.ID, error) {
-	packs := make(map[restic.ID][]restic.Blob, len(idx.Packs))
-	for id, p := range idx.Packs {
-		packs[id] = p.Entries
-	}
+func (idx *Index) Save(ctx context.Context, repo restic.Repository, supersedes restic.IDs) (restic.IDs, error) {
+	debug.Log("pack files: %d\n", len(idx.Packs))
 
-	return Save(ctx, repo, packs, supersedes)
-}
+	var indexIDs []restic.ID
 
-// Save writes a new index containing the given packs.
-func Save(ctx context.Context, repo restic.Repository, packs map[restic.ID][]restic.Blob, supersedes restic.IDs) (restic.ID, error) {
-	idx := &indexJSON{
+	packs := 0
+	jsonIDX := &indexJSON{
 		Supersedes: supersedes,
-		Packs:      make([]*packJSON, 0, len(packs)),
+		Packs:      make([]packJSON, 0, maxEntries),
 	}
 
-	for packID, blobs := range packs {
-		b := make([]blobJSON, 0, len(blobs))
-		for _, blob := range blobs {
+	for packID, pack := range idx.Packs {
+		debug.Log("%04d add pack %v with %d entries", packs, packID, len(pack.Entries))
+		b := make([]blobJSON, 0, len(pack.Entries))
+		for _, blob := range pack.Entries {
 			b = append(b, blobJSON{
 				ID:     blob.ID,
 				Type:   blob.Type,
@@ -285,13 +283,35 @@ func Save(ctx context.Context, repo restic.Repository, packs map[restic.ID][]res
 			})
 		}
 
-		p := &packJSON{
+		p := packJSON{
 			ID:    packID,
 			Blobs: b,
 		}
 
-		idx.Packs = append(idx.Packs, p)
+		jsonIDX.Packs = append(jsonIDX.Packs, p)
+
+		packs++
+		if packs == maxEntries {
+			id, err := repo.SaveJSONUnpacked(ctx, restic.IndexFile, jsonIDX)
+			if err != nil {
+				return nil, err
+			}
+			debug.Log("saved new index as %v", id)
+
+			indexIDs = append(indexIDs, id)
+			packs = 0
+			jsonIDX.Packs = jsonIDX.Packs[:0]
+		}
 	}
 
-	return repo.SaveJSONUnpacked(ctx, restic.IndexFile, idx)
+	if packs > 0 {
+		id, err := repo.SaveJSONUnpacked(ctx, restic.IndexFile, jsonIDX)
+		if err != nil {
+			return nil, err
+		}
+		debug.Log("saved new index as %v", id)
+		indexIDs = append(indexIDs, id)
+	}
+
+	return indexIDs, nil
 }

internal/index/index_test.go

@@ -2,7 +2,6 @@ package index
 
 import (
 	"context"
-	"math/rand"
 	"testing"
 	"time"
 
@@ -177,12 +176,12 @@ func BenchmarkIndexSave(b *testing.B) {
 	b.ResetTimer()
 
 	for i := 0; i < b.N; i++ {
-		id, err := idx.Save(context.TODO(), repo, nil)
+		ids, err := idx.Save(context.TODO(), repo, nil)
 		if err != nil {
 			b.Fatalf("New() returned error %v", err)
 		}
 
-		b.Logf("saved as %v", id.Str())
+		b.Logf("saved as %v", ids)
 	}
 }
 
@@ -217,69 +216,18 @@ func loadIndex(t testing.TB, repo restic.Repository) *Index {
 	return idx
 }
 
-func TestSave(t *testing.T) {
-	repo, cleanup := createFilledRepo(t, 3, 0)
-	defer cleanup()
-
-	idx := loadIndex(t, repo)
-
-	packs := make(map[restic.ID][]restic.Blob)
-	for id := range idx.Packs {
-		if rand.Float32() < 0.5 {
-			packs[id] = idx.Packs[id].Entries
-		}
-	}
-
-	t.Logf("save %d/%d packs in a new index\n", len(packs), len(idx.Packs))
-
-	id, err := Save(context.TODO(), repo, packs, idx.IndexIDs.List())
-	if err != nil {
-		t.Fatalf("unable to save new index: %v", err)
-	}
-
-	t.Logf("new index saved as %v", id.Str())
-
-	for id := range idx.IndexIDs {
-		t.Logf("remove index %v", id.Str())
-		h := restic.Handle{Type: restic.IndexFile, Name: id.String()}
-		err = repo.Backend().Remove(context.TODO(), h)
-		if err != nil {
-			t.Errorf("error removing index %v: %v", id, err)
-		}
-	}
-
-	idx2 := loadIndex(t, repo)
-	t.Logf("load new index with %d packs", len(idx2.Packs))
-
-	if len(idx2.Packs) != len(packs) {
-		t.Errorf("wrong number of packs in new index, want %d, got %d", len(packs), len(idx2.Packs))
-	}
-
-	for id := range packs {
-		if _, ok := idx2.Packs[id]; !ok {
-			t.Errorf("pack %v is not contained in new index", id.Str())
-		}
-	}
-
-	for id := range idx2.Packs {
-		if _, ok := packs[id]; !ok {
-			t.Errorf("pack %v is not contained in new index", id.Str())
-		}
-	}
-}
-
 func TestIndexSave(t *testing.T) {
 	repo, cleanup := createFilledRepo(t, 3, 0)
 	defer cleanup()
 
 	idx := loadIndex(t, repo)
 
-	id, err := idx.Save(context.TODO(), repo, idx.IndexIDs.List())
+	ids, err := idx.Save(context.TODO(), repo, idx.IndexIDs.List())
 	if err != nil {
 		t.Fatalf("unable to save new index: %v", err)
 	}
 
-	t.Logf("new index saved as %v", id.Str())
+	t.Logf("new index saved as %v", ids)
 
 	for id := range idx.IndexIDs {
 		t.Logf("remove index %v", id.Str())