2
2
mirror of https://github.com/octoleo/restic.git synced 2024-11-25 06:07:44 +00:00

rebuild-index: correctly rebuild index for mixed packs

For mixed packs, data and tree blobs were stored in separate index
entries. This results in warning from the check command and maybe other
problems.
This commit is contained in:
Michael Eischer 2022-05-24 22:30:42 +02:00
parent bb5f196b09
commit 04c23fa95d
4 changed files with 72 additions and 14 deletions

View File

@ -0,0 +1,13 @@
Bugfix: Correctly rebuild index for legacy repositories
After running `rebuild-index` on a legacy repository containing mixed pack
files, that is pack files with store both metadata and file data, `check`
printed warnings like `pack 12345678 contained in several indexes: ...`. The
warning is not critical. It has been fixed by properly handling mixed pack
files while rebuilding the index.
Running `prune` for such legacy repositories will also fix the warning by
reorganizing the pack files which cause the warning.
https://github.com/restic/restic/pull/3772
https://forum.restic.net/t/help-in-recovery-a-corrupted-repository/5044/13

View File

@ -247,8 +247,8 @@ func (idx *Index) Each(ctx context.Context) <-chan restic.PackedBlob {
} }
type EachByPackResult struct { type EachByPackResult struct {
packID restic.ID PackID restic.ID
blobs []restic.Blob Blobs []restic.Blob
} }
// EachByPack returns a channel that yields all blobs known to the index // EachByPack returns a channel that yields all blobs known to the index
@ -269,22 +269,29 @@ func (idx *Index) EachByPack(ctx context.Context, packBlacklist restic.IDSet) <-
close(ch) close(ch)
}() }()
byPack := make(map[restic.ID][][]*indexEntry)
for typ := range idx.byType { for typ := range idx.byType {
byPack := make(map[restic.ID][]*indexEntry)
m := &idx.byType[typ] m := &idx.byType[typ]
m.foreach(func(e *indexEntry) bool { m.foreach(func(e *indexEntry) bool {
packID := idx.packs[e.packIndex] packID := idx.packs[e.packIndex]
if _, ok := byPack[packID]; !ok {
byPack[packID] = make([][]*indexEntry, restic.NumBlobTypes)
}
if !idx.final || !packBlacklist.Has(packID) { if !idx.final || !packBlacklist.Has(packID) {
byPack[packID] = append(byPack[packID], e) byPack[packID][typ] = append(byPack[packID][typ], e)
} }
return true return true
}) })
}
for packID, pack := range byPack { for packID, packByType := range byPack {
var result EachByPackResult var result EachByPackResult
result.packID = packID result.PackID = packID
for typ, pack := range packByType {
for _, e := range pack { for _, e := range pack {
result.blobs = append(result.blobs, idx.toPackedBlob(e, restic.BlobType(typ)).Blob) result.Blobs = append(result.Blobs, idx.toPackedBlob(e, restic.BlobType(typ)).Blob)
}
} }
select { select {
case <-ctx.Done(): case <-ctx.Done():
@ -292,7 +299,6 @@ func (idx *Index) EachByPack(ctx context.Context, packBlacklist restic.IDSet) <-
case ch <- result: case ch <- result:
} }
} }
}
}() }()
return ch return ch

View File

@ -574,3 +574,42 @@ func TestIndexHas(t *testing.T) {
rtest.Assert(t, !idx.Has(restic.NewRandomBlobHandle()), "Index reports having a data blob not added to it") rtest.Assert(t, !idx.Has(restic.NewRandomBlobHandle()), "Index reports having a data blob not added to it")
rtest.Assert(t, !idx.Has(restic.BlobHandle{ID: tests[0].ID, Type: restic.TreeBlob}), "Index reports having a tree blob added to it with the same id as a data blob") rtest.Assert(t, !idx.Has(restic.BlobHandle{ID: tests[0].ID, Type: restic.TreeBlob}), "Index reports having a tree blob added to it with the same id as a data blob")
} }
func TestMixedEachByPack(t *testing.T) {
idx := repository.NewIndex()
expected := make(map[restic.ID]int)
// create 50 packs with 2 blobs each
for i := 0; i < 50; i++ {
packID := restic.NewRandomID()
expected[packID] = 1
blobs := []restic.Blob{
{
BlobHandle: restic.BlobHandle{Type: restic.DataBlob, ID: restic.NewRandomID()},
Offset: 0,
Length: 42,
},
{
BlobHandle: restic.BlobHandle{Type: restic.TreeBlob, ID: restic.NewRandomID()},
Offset: 42,
Length: 43,
},
}
idx.StorePack(packID, blobs)
}
reported := make(map[restic.ID]int)
for bp := range idx.EachByPack(context.TODO(), restic.NewIDSet()) {
reported[bp.PackID]++
rtest.Equals(t, 2, len(bp.Blobs)) // correct blob count
if bp.Blobs[0].Offset > bp.Blobs[1].Offset {
bp.Blobs[1], bp.Blobs[0] = bp.Blobs[0], bp.Blobs[1]
}
b0 := bp.Blobs[0]
rtest.Assert(t, b0.Type == restic.DataBlob && b0.Offset == 0 && b0.Length == 42, "wrong blob", b0)
b1 := bp.Blobs[1]
rtest.Assert(t, b1.Type == restic.TreeBlob && b1.Offset == 42 && b1.Length == 43, "wrong blob", b1)
}
rtest.Equals(t, expected, reported)
}

View File

@ -339,7 +339,7 @@ func (mi *MasterIndex) Save(ctx context.Context, repo restic.SaverUnpacked, pack
debug.Log("adding index %d", i) debug.Log("adding index %d", i)
for pbs := range idx.EachByPack(ctx, packBlacklist) { for pbs := range idx.EachByPack(ctx, packBlacklist) {
newIndex.StorePack(pbs.packID, pbs.blobs) newIndex.StorePack(pbs.PackID, pbs.Blobs)
p.Add(1) p.Add(1)
if IndexFull(newIndex, mi.compress) { if IndexFull(newIndex, mi.compress) {
select { select {