Rework repository archive (#14723)

* Use storage to store archive files

* Fix backend lint

* Add archiver table on database

* Finish archive download

* Fix test

* Add database migrations

* Add status for archiver

* Fix lint

* Add queue

* Add doctor to check and delete old archives

* Improve archive queue

* Fix tests

* improve archive storage

* Delete repo archives

* Add missing fixture

* fix fixture

* Fix fixture

* Fix test

* Fix archiver cleaning

* Fix bug

* Add docs for repository archive storage

* remove repo-archive configuration

* Fix test

* Fix test

* Fix lint

Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>

1 files changed, 47 insertions, 50 deletions

diff --git a/models/repo.go b/models/repo.go
index dc4e03a28a..2baf6e9bdd 100644
--- a/models/repo.go
+++ b/models/repo.go
@@ -1587,6 +1587,22 @@ func DeleteRepository(doer *User, uid, repoID int64) error {
 		return err
 	}
 
+	// Remove archives
+	var archives []*RepoArchiver
+	if err = sess.Where("repo_id=?", repoID).Find(&archives); err != nil {
+		return err
+	}
+
+	for _, v := range archives {
+		v.Repo = repo
+		p, _ := v.RelativePath()
+		removeStorageWithNotice(sess, storage.RepoArchives, "Delete repo archive file", p)
+	}
+
+	if _, err := sess.Delete(&RepoArchiver{RepoID: repoID}); err != nil {
+		return err
+	}
+
 	if repo.NumForks > 0 {
 		if _, err = sess.Exec("UPDATE `repository` SET fork_id=0,is_fork=? WHERE fork_id=?", false, repo.ID); err != nil {
 			log.Error("reset 'fork_id' and 'is_fork': %v", err)
@@ -1768,64 +1784,45 @@ func DeleteRepositoryArchives(ctx context.Context) error {
 func DeleteOldRepositoryArchives(ctx context.Context, olderThan time.Duration) error {
 	log.Trace("Doing: ArchiveCleanup")
 
-	if err := x.Where("id > 0").Iterate(new(Repository), func(idx int, bean interface{}) error {
-		return deleteOldRepositoryArchives(ctx, olderThan, idx, bean)
-	}); err != nil {
-		log.Trace("Error: ArchiveClean: %v", err)
-		return err
-	}
-
-	log.Trace("Finished: ArchiveCleanup")
-	return nil
-}
-
-func deleteOldRepositoryArchives(ctx context.Context, olderThan time.Duration, idx int, bean interface{}) error {
-	repo := bean.(*Repository)
-	basePath := filepath.Join(repo.RepoPath(), "archives")
-
-	for _, ty := range []string{"zip", "targz"} {
-		select {
-		case <-ctx.Done():
-			return ErrCancelledf("before deleting old repository archives with filetype %s for %s", ty, repo.FullName())
-		default:
-		}
-
-		path := filepath.Join(basePath, ty)
-		file, err := os.Open(path)
-		if err != nil {
-			if !os.IsNotExist(err) {
-				log.Warn("Unable to open directory %s: %v", path, err)
-				return err
-			}
-
-			// If the directory doesn't exist, that's okay.
-			continue
-		}
-
-		files, err := file.Readdir(0)
-		file.Close()
+	for {
+		var archivers []RepoArchiver
+		err := x.Where("created_unix < ?", time.Now().Add(-olderThan).Unix()).
+			Asc("created_unix").
+			Limit(100).
+			Find(&archivers)
 		if err != nil {
-			log.Warn("Unable to read directory %s: %v", path, err)
+			log.Trace("Error: ArchiveClean: %v", err)
 			return err
 		}
 
-		minimumOldestTime := time.Now().Add(-olderThan)
-		for _, info := range files {
-			if info.ModTime().Before(minimumOldestTime) && !info.IsDir() {
-				select {
-				case <-ctx.Done():
-					return ErrCancelledf("before deleting old repository archive file %s with filetype %s for %s", info.Name(), ty, repo.FullName())
-				default:
-				}
-				toDelete := filepath.Join(path, info.Name())
-				// This is a best-effort purge, so we do not check error codes to confirm removal.
-				if err = util.Remove(toDelete); err != nil {
-					log.Trace("Unable to delete %s, but proceeding: %v", toDelete, err)
-				}
+		for _, archiver := range archivers {
+			if err := deleteOldRepoArchiver(ctx, &archiver); err != nil {
+				return err
 			}
 		}
+		if len(archivers) < 100 {
+			break
+		}
 	}
 
+	log.Trace("Finished: ArchiveCleanup")
+	return nil
+}
+
+var delRepoArchiver = new(RepoArchiver)
+
+func deleteOldRepoArchiver(ctx context.Context, archiver *RepoArchiver) error {
+	p, err := archiver.RelativePath()
+	if err != nil {
+		return err
+	}
+	_, err = x.ID(archiver.ID).Delete(delRepoArchiver)
+	if err != nil {
+		return err
+	}
+	if err := storage.RepoArchives.Delete(p); err != nil {
+		log.Error("delete repo archive file failed: %v", err)
+	}
 	return nil
 }