Skip failed LFS objects and continue when migrating LFS

modules/lfs/errors.go

@@ -0,0 +1,60 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package lfs
+
+import (
+	"fmt"
+)
+
+type ErrLFS struct {
+	OID string
+	Err error
+}
+
+type ErrLFSDownload struct {
+	ErrLFS
+}
+
+type ErrLFSUpload struct {
+	ErrLFS
+}
+
+type ErrLFSVerify struct {
+	ErrLFS
+}
+
+func (e *ErrLFS) Error() string {
+	return fmt.Sprintf("LFS error for object %s: %v", e.OID, e.Err)
+}
+
+func (e *ErrLFS) Unwrap() error {
+	return e.Err
+}
+
+func (e *ErrLFSDownload) Error() string {
+	return fmt.Sprintf("LFS error while downloading [%s]: %s", e.OID, e.Err)
+}
+
+func (e *ErrLFSUpload) Error() string {
+	return fmt.Sprintf("LFS error while uploading [%s]: %s", e.OID, e.Err)
+}
+
+func (e *ErrLFSVerify) Error() string {
+	return fmt.Sprintf("LFS error while verifying [%s]: %s", e.OID, e.Err)
+}
+
+func (e *ErrLFSDownload) Is(target error) bool {
+	_, ok := target.(*ErrLFSDownload)
+	return ok
+}
+
+func (e *ErrLFSUpload) Is(target error) bool {
+	_, ok := target.(*ErrLFSUpload)
+	return ok
+}
+
+func (e *ErrLFSVerify) Is(target error) bool {
+	_, ok := target.(*ErrLFSVerify)
+	return ok
+}

modules/lfs/http_client.go

@@ -71,9 +71,12 @@
 
 	url := fmt.Sprintf("%s/objects/batch", c.endpoint)
 
-	request := &BatchRequest{operation, c.transferNames(), nil, objects}
 	payload := new(bytes.Buffer)
-	err := json.NewEncoder(payload).Encode(request)
+	err := json.NewEncoder(payload).Encode(&BatchRequest{
+		Operation: operation,
+		Transfers: c.transferNames(),
+		Objects:   objects,
+	})
 	if err != nil {
 		log.Error("Error encoding json: %v", err)
 		return nil, err
@@ -134,15 +137,19 @@
 		return fmt.Errorf("TransferAdapter not found: %s", result.Transfer)
 	}
 
+	var resultErrors []error
 	for _, object := range result.Objects {
 		if object.Error != nil {
-			objectError := errors.New(object.Error.Message)
-			log.Trace("Error on object %v: %v", object.Pointer, objectError)
+			log.Trace("Error on object %v: %v", object.Pointer, object.Error.Message)
 			if uc != nil {
+				objectError := &ErrLFSUpload{ErrLFS{OID: object.Oid, Err: object.Error}}
+				resultErrors = append(resultErrors, objectError)
 				if _, err := uc(object.Pointer, objectError); err != nil {
 					return err
 				}
 			} else {
+				objectError := &ErrLFSDownload{ErrLFS{OID: object.Oid, Err: object.Error}}
+				resultErrors = append(resultErrors, objectError)
 				if err := dc(object.Pointer, nil, objectError); err != nil {
 					return err
 				}
@@ -164,18 +171,33 @@
 
 			content, err := uc(object.Pointer, nil)
 			if err != nil {
-				return err
+				if errors.Is(err, &ErrLFSUpload{}) {
+					resultErrors = append(resultErrors, err)
+					continue
+				} else {
+					return err
+				}
 			}
 
 			err = transferAdapter.Upload(ctx, link, object.Pointer, content)
 			if err != nil {
-				return err
+				if errors.Is(err, &ErrLFSUpload{}) {
+					resultErrors = append(resultErrors, err)
+					continue
+				} else {
+					return err
+				}
 			}
 
 			link, ok = object.Actions["verify"]
 			if ok {
 				if err := transferAdapter.Verify(ctx, link, object.Pointer); err != nil {
-					return err
+					if errors.Is(err, &ErrLFSVerify{}) {
+						resultErrors = append(resultErrors, err)
+						continue
+					} else {
+						return err
+					}
 				}
 			}
 		} else {
@@ -191,16 +213,26 @@
 
 			content, err := transferAdapter.Download(ctx, link)
 			if err != nil {
-				return err
+				if errors.Is(err, &ErrLFSDownload{}) {
+					resultErrors = append(resultErrors, err)
+					continue
+				} else {
+					return err
+				}
 			}
 
 			if err := dc(object.Pointer, content, nil); err != nil {
-				return err
+				if errors.Is(err, &ErrLFSDownload{}) {
+					resultErrors = append(resultErrors, err)
+					continue
+				} else {
+					return err
+				}
 			}
 		}
 	}
 
-	return nil
+	return errors.Join(resultErrors...)
 }
 
 // createRequest creates a new request, and sets the headers.

modules/lfs/shared.go

@@ -4,6 +4,7 @@
 package lfs
 
 import (
+	"fmt"
 	"time"
 )
 
@@ -64,6 +65,10 @@ type ObjectError struct {
 	Message string `json:"message"`
 }
 
+func (oe *ObjectError) Error() string {
+	return fmt.Sprintf("code: %d, message: %s", oe.Code, oe.Message)
+}
+
 // PointerBlob associates a Git blob with a Pointer.
 type PointerBlob struct {
 	Hash string

modules/repository/repo.go

@@ -5,6 +5,7 @@
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"io"
 	"strings"
@@ -178,9 +179,16 @@
 	errChan := make(chan error, 1)
 	go lfs.SearchPointerBlobs(ctx, gitRepo, pointerChan, errChan)
 
+	var lfsDownloadErrors []error
+
 	downloadObjects := func(pointers []lfs.Pointer) error {
 		err := lfsClient.Download(ctx, pointers, func(p lfs.Pointer, content io.ReadCloser, objectError error) error {
 			if objectError != nil {
+				if errors.Is(objectError, &lfs.ErrLFSDownload{}) {
+					log.Trace("Repo[%-v]: Ignoring LFS error: %v", repo, objectError)
+					lfsDownloadErrors = append(lfsDownloadErrors, objectError)
+					return nil
+				}
 				return objectError
 			}
 
@@ -208,10 +216,13 @@
 			default:
 			}
 		}
+		if errors.Is(err, &lfs.ErrLFSDownload{}) {
+			return nil
+		}
 		return err
 	}
 
-	var batch []lfs.Pointer
+	var downloadBatch []lfs.Pointer
 	for pointerBlob := range pointerChan {
 		meta, err := git_model.GetLFSMetaObjectByOid(ctx, repo.ID, pointerBlob.Oid)
 		if err != nil && err != git_model.ErrLFSObjectNotExist {
@@ -233,28 +244,32 @@
 
 		if exist {
 			log.Trace("Repo[%-v]: LFS object %-v already present; creating meta object", repo, pointerBlob.Pointer)
-			_, err := git_model.NewLFSMetaObject(ctx, repo.ID, pointerBlob.Pointer)
+			m, err := git_model.NewLFSMetaObject(ctx, repo.ID, pointerBlob.Pointer)
 			if err != nil {
 				log.Error("Repo[%-v]: Error creating LFS meta object %-v: %v", repo, pointerBlob.Pointer, err)
 				return err
 			}
+			if m.Existing {
+				log.Trace("Repo[%-v]: LFS meta object %-v was already present", repo, m)
+			}
+			continue
 		} else {
 			if setting.LFS.MaxFileSize > 0 && pointerBlob.Size > setting.LFS.MaxFileSize {
 				log.Info("Repo[%-v]: LFS object %-v download denied because of LFS_MAX_FILE_SIZE=%d < size %d", repo, pointerBlob.Pointer, setting.LFS.MaxFileSize, pointerBlob.Size)
 				continue
 			}
 
-			batch = append(batch, pointerBlob.Pointer)
-			if len(batch) >= lfsClient.BatchSize() {
-				if err := downloadObjects(batch); err != nil {
+			downloadBatch = append(downloadBatch, pointerBlob.Pointer)
+			if len(downloadBatch) >= lfsClient.BatchSize() {
+				if err := downloadObjects(downloadBatch); err != nil {
 					return err
 				}
-				batch = nil
+				downloadBatch = nil
 			}
 		}
 	}
-	if len(batch) > 0 {
-		if err := downloadObjects(batch); err != nil {
+	if len(downloadBatch) > 0 {
+		if err := downloadObjects(downloadBatch); err != nil {
 			return err
 		}
 	}
@@ -265,7 +280,7 @@
 		return err
 	}
 
-	return nil
+	return errors.Join(lfsDownloadErrors...)
 }
 
 // shortRelease to reduce load memory, this struct can replace repo_model.Release

services/repository/migrate.go

@@ -169,6 +169,7 @@ func MigrateRepositoryGitData(ctx context.Context, u *user_model.User,
 			lfsClient := lfs.NewClient(endpoint, httpTransport)
 			if err = repo_module.StoreMissingLfsObjectsInRepository(ctx, repo, gitRepo, lfsClient); err != nil {
 				log.Error("Failed to store missing LFS objects for repository: %v", err)
+				// TODO: check for lfs.ErrLFSDownload errors and display them
 			}
 		}
 	}