2
2
mirror of https://github.com/octoleo/restic.git synced 2024-12-22 10:58:55 +00:00

stats: hardlinks only reduce restore within a snapshot

The `stats` command checks inodes to not count hardlinked files multiple
times into the restore size. This check applies across all snapshots and
not only within snapshots. As a result the result size was far too low
when calculating it for multiple snapshots and it would vary depending
on the order in which snapshots were listed.
This commit is contained in:
Michael Eischer 2022-05-07 14:33:53 +02:00
parent cc8a03b1d0
commit 2c07f7fff3
2 changed files with 13 additions and 9 deletions

View File

@ -0,0 +1,8 @@
Bugfix: `stats` fix restore size calculation for multiple snapshots
Since restic 0.10.0 the restore size calculated by the `stats` command for
multiple snapshots was too low. The hardlink detection was accidentally applied
across multiple snapshots and thus ignored many files. This has been fixed.
https://github.com/restic/restic/issues/3736
https://github.com/restic/restic/pull/3740

View File

@ -111,7 +111,6 @@ func runStats(gopts GlobalOptions, args []string) error {
// create a container for the stats (and other needed state) // create a container for the stats (and other needed state)
stats := &statsContainer{ stats := &statsContainer{
uniqueFiles: make(map[fileID]struct{}), uniqueFiles: make(map[fileID]struct{}),
uniqueInodes: make(map[uint64]struct{}),
fileBlobs: make(map[string]restic.IDSet), fileBlobs: make(map[string]restic.IDSet),
blobs: restic.NewBlobSet(), blobs: restic.NewBlobSet(),
snapshotsCount: 0, snapshotsCount: 0,
@ -175,7 +174,8 @@ func statsWalkSnapshot(ctx context.Context, snapshot *restic.Snapshot, repo rest
return restic.FindUsedBlobs(ctx, repo, restic.IDs{*snapshot.Tree}, stats.blobs, nil) return restic.FindUsedBlobs(ctx, repo, restic.IDs{*snapshot.Tree}, stats.blobs, nil)
} }
err := walker.Walk(ctx, repo, *snapshot.Tree, restic.NewIDSet(), statsWalkTree(repo, stats)) uniqueInodes := make(map[uint64]struct{})
err := walker.Walk(ctx, repo, *snapshot.Tree, restic.NewIDSet(), statsWalkTree(repo, stats, uniqueInodes))
if err != nil { if err != nil {
return fmt.Errorf("walking tree %s: %v", *snapshot.Tree, err) return fmt.Errorf("walking tree %s: %v", *snapshot.Tree, err)
} }
@ -183,7 +183,7 @@ func statsWalkSnapshot(ctx context.Context, snapshot *restic.Snapshot, repo rest
return nil return nil
} }
func statsWalkTree(repo restic.Repository, stats *statsContainer) walker.WalkFunc { func statsWalkTree(repo restic.Repository, stats *statsContainer, uniqueInodes map[uint64]struct{}) walker.WalkFunc {
return func(parentTreeID restic.ID, npath string, node *restic.Node, nodeErr error) (bool, error) { return func(parentTreeID restic.ID, npath string, node *restic.Node, nodeErr error) (bool, error) {
if nodeErr != nil { if nodeErr != nil {
return true, nodeErr return true, nodeErr
@ -242,8 +242,8 @@ func statsWalkTree(repo restic.Repository, stats *statsContainer) walker.WalkFun
// if inodes are present, only count each inode once // if inodes are present, only count each inode once
// (hard links do not increase restore size) // (hard links do not increase restore size)
if _, ok := stats.uniqueInodes[node.Inode]; !ok || node.Inode == 0 { if _, ok := uniqueInodes[node.Inode]; !ok || node.Inode == 0 {
stats.uniqueInodes[node.Inode] = struct{}{} uniqueInodes[node.Inode] = struct{}{}
stats.TotalSize += node.Size stats.TotalSize += node.Size
} }
@ -290,10 +290,6 @@ type statsContainer struct {
// contents (hashed sequence of content blob IDs) // contents (hashed sequence of content blob IDs)
uniqueFiles map[fileID]struct{} uniqueFiles map[fileID]struct{}
// uniqueInodes marks visited files according to their
// inode # (hashed sequence of inode numbers)
uniqueInodes map[uint64]struct{}
// fileBlobs maps a file name (path) to the set of // fileBlobs maps a file name (path) to the set of
// blobs that have been seen as a part of the file // blobs that have been seen as a part of the file
fileBlobs map[string]restic.IDSet fileBlobs map[string]restic.IDSet