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>

custom/conf/app.example.ini

@@ -2048,6 +2048,16 @@ PATH =
 ;; storage type
 ;STORAGE_TYPE = local
 
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;; settings for repository archives, will override storage setting
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;[storage.repo-archive]
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;; storage type
+;STORAGE_TYPE = local
+
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 ;; lfs storage will override storage

docs/content/doc/advanced/config-cheat-sheet.en-us.md

@@ -995,6 +995,23 @@ MINIO_USE_SSL = false
 
 And used by `[attachment]`, `[lfs]` and etc. as `STORAGE_TYPE`.
 
+## Repository Archive Storage (`storage.repo-archive`)
+
+Configuration for repository archive storage. It will inherit from default `[storage]` or
+`[storage.xxx]` when set `STORAGE_TYPE` to `xxx`. The default of `PATH`
+is `data/repo-archive` and the default of `MINIO_BASE_PATH` is `repo-archive/`.
+
+- `STORAGE_TYPE`: **local**: Storage type for repo archive, `local` for local disk or `minio` for s3 compatible object storage service or other name defined with `[storage.xxx]`
+- `SERVE_DIRECT`: **false**: Allows the storage driver to redirect to authenticated URLs to serve files directly. Currently, only Minio/S3 is supported via signed URLs, local does nothing.
+- `PATH`: **./data/repo-archive**: Where to store archive files, only available when `STORAGE_TYPE` is `local`.
+- `MINIO_ENDPOINT`: **localhost:9000**: Minio endpoint to connect only available when `STORAGE_TYPE` is `minio`
+- `MINIO_ACCESS_KEY_ID`: Minio accessKeyID to connect only available when `STORAGE_TYPE` is `minio`
+- `MINIO_SECRET_ACCESS_KEY`: Minio secretAccessKey to connect only available when `STORAGE_TYPE is` `minio`
+- `MINIO_BUCKET`: **gitea**: Minio bucket to store the lfs only available when `STORAGE_TYPE` is `minio`
+- `MINIO_LOCATION`: **us-east-1**: Minio location to create bucket only available when `STORAGE_TYPE` is `minio`
+- `MINIO_BASE_PATH`: **repo-archive/**: Minio base path on the bucket only available when `STORAGE_TYPE` is `minio`
+- `MINIO_USE_SSL`: **false**: Minio enabled ssl only available when `STORAGE_TYPE` is `minio`
+
 ## Other (`other`)
 
 - `SHOW_FOOTER_BRANDING`: **false**: Show Gitea branding in the footer.

docs/content/doc/advanced/config-cheat-sheet.zh-cn.md

@@ -382,6 +382,21 @@ MINIO_USE_SSL = false
 
 然后你在 `[attachment]`, `[lfs]` 等中可以把这个名字用作 `STORAGE_TYPE` 的值。
 
+## Repository Archive Storage (`storage.repo-archive`)
+
+Repository archive 的存储配置。 如果 `STORAGE_TYPE` 为空，则此配置将从 `[storage]` 继承。如果不为 `local` 或者 `minio` 而为 `xxx`， 则从 `[storage.xxx]` 继承。当继承时， `PATH` 默认为 `data/repo-archive`，`MINIO_BASE_PATH` 默认为 `repo-archive/`。
+
+- `STORAGE_TYPE`: **local**: Repository archive 的存储类型，`local` 将存储到磁盘，`minio` 将存储到 s3 兼容的对象服务。
+- `SERVE_DIRECT`: **false**: 允许直接重定向到存储系统。当前，仅 Minio/S3 是支持的。
+- `PATH`: 存放 Repository archive 上传的文件的地方，默认是 `data/repo-archive`。
+- `MINIO_ENDPOINT`: **localhost:9000**: Minio 地址，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+- `MINIO_ACCESS_KEY_ID`: Minio accessKeyID，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+- `MINIO_SECRET_ACCESS_KEY`: Minio secretAccessKey，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+- `MINIO_BUCKET`: **gitea**: Minio bucket，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+- `MINIO_LOCATION`: **us-east-1**: Minio location ，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+- `MINIO_BASE_PATH`: **repo-archive/**: Minio base path ，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+- `MINIO_USE_SSL`: **false**: Minio 是否启用 ssl ，仅当 `STORAGE_TYPE` 为 `minio` 时有效。
+
 ## Other (`other`)
 
 - `SHOW_FOOTER_BRANDING`: 为真则在页面底部显示Gitea的字样。

integrations/gitea-repositories-meta/user27/repo49.git/refs/heads/test/archive

@@ -0,0 +1 @@
+aacbdfe9e1c4b47f60abe81849045fa4e96f1d75

models/fixtures/repo_archiver.yml

@@ -0,0 +1 @@
+[] # empty

models/migrations/migrations.go

@@ -319,6 +319,8 @@ var migrations = []Migration{
 	NewMigration("Create PushMirror table", createPushMirrorTable),
 	// v184 -> v185
 	NewMigration("Rename Task errors to message", renameTaskErrorsToMessage),
+	// v185 -> v186
+	NewMigration("Add new table repo_archiver", addRepoArchiver),
 }
 
 // GetCurrentDBVersion returns the current db version

models/migrations/v181.go

@@ -1,3 +1,4 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
 // Use of this source code is governed by a MIT-style
 // license that can be found in the LICENSE file.
 

models/migrations/v185.go

@@ -0,0 +1,22 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package migrations
+
+import (
+	"xorm.io/xorm"
+)
+
+func addRepoArchiver(x *xorm.Engine) error {
+	// RepoArchiver represents all archivers
+	type RepoArchiver struct {
+		ID          int64 `xorm:"pk autoincr"`
+		RepoID      int64 `xorm:"index unique(s)"`
+		Type        int   `xorm:"unique(s)"`
+		Status      int
+		CommitID    string `xorm:"VARCHAR(40) unique(s)"`
+		CreatedUnix int64  `xorm:"INDEX NOT NULL created"`
+	}
+	return x.Sync2(new(RepoArchiver))
+}

models/models.go

@@ -136,6 +136,7 @@ func init() {
 		new(RepoTransfer),
 		new(IssueIndex),
 		new(PushMirror),
+		new(RepoArchiver),
 	)
 
 	gonicNames := []string{"SSL", "UID"}

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
 }
 

models/repo_archiver.go

@@ -0,0 +1,86 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package models
+
+import (
+	"fmt"
+
+	"code.gitea.io/gitea/modules/git"
+	"code.gitea.io/gitea/modules/timeutil"
+)
+
+// RepoArchiverStatus represents repo archive status
+type RepoArchiverStatus int
+
+// enumerate all repo archive statuses
+const (
+	RepoArchiverGenerating = iota // the archiver is generating
+	RepoArchiverReady             // it's ready
+)
+
+// RepoArchiver represents all archivers
+type RepoArchiver struct {
+	ID          int64           `xorm:"pk autoincr"`
+	RepoID      int64           `xorm:"index unique(s)"`
+	Repo        *Repository     `xorm:"-"`
+	Type        git.ArchiveType `xorm:"unique(s)"`
+	Status      RepoArchiverStatus
+	CommitID    string             `xorm:"VARCHAR(40) unique(s)"`
+	CreatedUnix timeutil.TimeStamp `xorm:"INDEX NOT NULL created"`
+}
+
+// LoadRepo loads repository
+func (archiver *RepoArchiver) LoadRepo() (*Repository, error) {
+	if archiver.Repo != nil {
+		return archiver.Repo, nil
+	}
+
+	var repo Repository
+	has, err := x.ID(archiver.RepoID).Get(&repo)
+	if err != nil {
+		return nil, err
+	}
+	if !has {
+		return nil, ErrRepoNotExist{
+			ID: archiver.RepoID,
+		}
+	}
+	return &repo, nil
+}
+
+// RelativePath returns relative path
+func (archiver *RepoArchiver) RelativePath() (string, error) {
+	repo, err := archiver.LoadRepo()
+	if err != nil {
+		return "", err
+	}
+
+	return fmt.Sprintf("%s/%s/%s.%s", repo.FullName(), archiver.CommitID[:2], archiver.CommitID, archiver.Type.String()), nil
+}
+
+// GetRepoArchiver get an archiver
+func GetRepoArchiver(ctx DBContext, repoID int64, tp git.ArchiveType, commitID string) (*RepoArchiver, error) {
+	var archiver RepoArchiver
+	has, err := ctx.e.Where("repo_id=?", repoID).And("`type`=?", tp).And("commit_id=?", commitID).Get(&archiver)
+	if err != nil {
+		return nil, err
+	}
+	if has {
+		return &archiver, nil
+	}
+	return nil, nil
+}
+
+// AddRepoArchiver adds an archiver
+func AddRepoArchiver(ctx DBContext, archiver *RepoArchiver) error {
+	_, err := ctx.e.Insert(archiver)
+	return err
+}
+
+// UpdateRepoArchiverStatus updates archiver's status
+func UpdateRepoArchiverStatus(ctx DBContext, archiver *RepoArchiver) error {
+	_, err := ctx.e.ID(archiver.ID).Cols("status").Update(archiver)
+	return err
+}

models/unit_tests.go

@@ -74,6 +74,8 @@ func MainTest(m *testing.M, pathToGiteaRoot string) {
 
 	setting.RepoAvatar.Storage.Path = filepath.Join(setting.AppDataPath, "repo-avatars")
 
+	setting.RepoArchive.Storage.Path = filepath.Join(setting.AppDataPath, "repo-archive")
+
 	if err = storage.Init(); err != nil {
 		fatalTestError("storage.Init: %v\n", err)
 	}

modules/context/context.go

@@ -380,6 +380,21 @@ func (ctx *Context) ServeFile(file string, names ...string) {
 	http.ServeFile(ctx.Resp, ctx.Req, file)
 }
 
+// ServeStream serves file via io stream
+func (ctx *Context) ServeStream(rd io.Reader, name string) {
+	ctx.Resp.Header().Set("Content-Description", "File Transfer")
+	ctx.Resp.Header().Set("Content-Type", "application/octet-stream")
+	ctx.Resp.Header().Set("Content-Disposition", "attachment; filename="+name)
+	ctx.Resp.Header().Set("Content-Transfer-Encoding", "binary")
+	ctx.Resp.Header().Set("Expires", "0")
+	ctx.Resp.Header().Set("Cache-Control", "must-revalidate")
+	ctx.Resp.Header().Set("Pragma", "public")
+	_, err := io.Copy(ctx.Resp, rd)
+	if err != nil {
+		ctx.ServerError("Download file failed", err)
+	}
+}
+
 // Error returned an error to web browser
 func (ctx *Context) Error(status int, contents ...string) {
 	var v = http.StatusText(status)