fix(dedupe): run dedupe only for repositories found at startup

no need to run dedupe/restore blobs for images being pushed or synced while running dedupe task, they are already deduped/restored inline. Signed-off-by: Petu Eusebiu <[email protected]>
project-zot · Sep 22, 2023 · 4366b6d · 4366b6d
1 parent 8c55944
commit 4366b6d
Show file tree

Hide file tree

Showing 5 changed files with 51 additions and 7 deletions.
diff --git a/pkg/storage/common/common.go b/pkg/storage/common/common.go
@@ -971,14 +971,36 @@ type DedupeTaskGenerator struct {
 	and generating a task for each unprocessed one*/
 	lastDigests []godigest.Digest
 	done        bool
+	repos       []string // list of repos on which we run dedupe
 	Log         zlog.Logger
 }
 
 func (gen *DedupeTaskGenerator) Next() (scheduler.Task, error) {
 	var err error
 
+	/* at first run get from storage currently found repositories so that we skip the ones that gets synced/uploaded
+	while this generator runs, there are deduped/restored inline, no need to run dedupe/restore again */
+	if len(gen.repos) == 0 {
+		gen.repos, err = gen.ImgStore.GetRepositories()
+		if err != nil {
+			//nolint: dupword
+			gen.Log.Error().Err(err).Msg("dedupe rebuild: unable to to get list of repositories")
+
+			return nil, err
+		}
+	}
+
+	if len(gen.repos) == 0 {
+		gen.Log.Info().Msg("dedupe rebuild: no repositories found in storage, finished.")
+
+		// no repositories in storage, no need to continue
+		gen.done = true
+
+		return nil, nil
+	}
+
 	// get all blobs from storage.imageStore and group them by digest
-	gen.digest, gen.duplicateBlobs, err = gen.ImgStore.GetNextDigestWithBlobPaths(gen.lastDigests)
+	gen.digest, gen.duplicateBlobs, err = gen.ImgStore.GetNextDigestWithBlobPaths(gen.repos, gen.lastDigests)
 	if err != nil {
 		gen.Log.Error().Err(err).Msg("dedupe rebuild: failed to get next digest")
 

diff --git a/pkg/storage/imagestore/imagestore.go b/pkg/storage/imagestore/imagestore.go
@@ -1965,7 +1965,8 @@ func (is *ImageStore) RunGCPeriodically(interval time.Duration, sch *scheduler.S
 	sch.SubmitGenerator(generator, interval, scheduler.MediumPriority)
 }
 
-func (is *ImageStore) GetNextDigestWithBlobPaths(lastDigests []godigest.Digest) (godigest.Digest, []string, error) {
+func (is *ImageStore) GetNextDigestWithBlobPaths(repos []string, lastDigests []godigest.Digest,
+) (godigest.Digest, []string, error) {
 	var lockLatency time.Time
 
 	dir := is.rootDir
@@ -1983,8 +1984,18 @@ func (is *ImageStore) GetNextDigestWithBlobPaths(lastDigests []godigest.Digest)
 			return driver.ErrSkipDir
 		}
 
+		// skip blobs under .uploads
+		if strings.HasSuffix(fileInfo.Path(), storageConstants.BlobUploadDir) {
+			return driver.ErrSkipDir
+		}
+
 		if fileInfo.IsDir() {
-			return nil
+			// skip repositories not found in repos
+			repo := path.Base(fileInfo.Path())
+
+			if !zcommon.Contains(repos, repo) && repo != "blobs" && repo != "sha256" {
+				return driver.ErrSkipDir
+			}
 		}
 
 		blobDigest := godigest.NewDigestFromEncoded("sha256", path.Base(fileInfo.Path()))

diff --git a/pkg/storage/local/local_test.go b/pkg/storage/local/local_test.go
@@ -1200,6 +1200,17 @@ func TestDedupeLinks(t *testing.T) {
 					storageConstants.DefaultUntaggedImgeRetentionDelay, testCase.dedupe, true, log, metrics, nil, nil)
 			}
 
+			// run on empty image store
+			// switch dedupe to true from false
+			taskScheduler, cancel := runAndGetScheduler()
+
+			// rebuild with dedupe true
+			imgStore.RunDedupeBlobs(time.Duration(0), taskScheduler)
+			// wait until rebuild finishes
+			time.Sleep(1 * time.Second)
+
+			cancel()
+
 			// manifest1
 			upload, err := imgStore.NewBlobUpload("dedupe1")
 			So(err, ShouldBeNil)

diff --git a/pkg/storage/types/types.go b/pkg/storage/types/types.go
@@ -52,7 +52,7 @@ type ImageStore interface { //nolint:interfacebloat
 	RunGCPeriodically(interval time.Duration, sch *scheduler.Scheduler)
 	RunDedupeBlobs(interval time.Duration, sch *scheduler.Scheduler)
 	RunDedupeForDigest(digest godigest.Digest, dedupe bool, duplicateBlobs []string) error
-	GetNextDigestWithBlobPaths(lastDigests []godigest.Digest) (godigest.Digest, []string, error)
+	GetNextDigestWithBlobPaths(repos []string, lastDigests []godigest.Digest) (godigest.Digest, []string, error)
 	GetAllBlobs(repo string) ([]string, error)
 }
 

diff --git a/pkg/test/mocks/image_store_mock.go b/pkg/test/mocks/image_store_mock.go
@@ -50,7 +50,7 @@ type MockedImageStore struct {
 	RunGCPeriodicallyFn          func(interval time.Duration, sch *scheduler.Scheduler)
 	RunDedupeBlobsFn             func(interval time.Duration, sch *scheduler.Scheduler)
 	RunDedupeForDigestFn         func(digest godigest.Digest, dedupe bool, duplicateBlobs []string) error
-	GetNextDigestWithBlobPathsFn func(lastDigests []godigest.Digest) (godigest.Digest, []string, error)
+	GetNextDigestWithBlobPathsFn func(repos []string, lastDigests []godigest.Digest) (godigest.Digest, []string, error)
 	GetAllBlobsFn                func(repo string) ([]string, error)
 }
 
@@ -370,10 +370,10 @@ func (is MockedImageStore) RunDedupeForDigest(digest godigest.Digest, dedupe boo
 	return nil
 }
 
-func (is MockedImageStore) GetNextDigestWithBlobPaths(lastDigests []godigest.Digest,
+func (is MockedImageStore) GetNextDigestWithBlobPaths(repos []string, lastDigests []godigest.Digest,
 ) (godigest.Digest, []string, error) {
 	if is.GetNextDigestWithBlobPathsFn != nil {
-		return is.GetNextDigestWithBlobPathsFn(lastDigests)
+		return is.GetNextDigestWithBlobPathsFn(repos, lastDigests)
 	}
 
 	return "", []string{}, nil