images.go
| 1 | package web |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "encoding/json" |
| 6 | "errors" |
| 7 | "fmt" |
| 8 | "io/fs" |
| 9 | "log" |
| 10 | "net/http" |
| 11 | "os" |
| 12 | "path/filepath" |
| 13 | |
| 14 | "github.com/go-chi/chi/v5" |
| 15 | |
| 16 | "hearthforge/internal/web/views" |
| 17 | ) |
| 18 | |
| 19 | // imageRoutes registers the Images tab. Reads follow REGISTRY_PULL like the |
| 20 | // registry itself; deletes are admin-only. |
| 21 | func (s *Server) imageRoutes(r chi.Router) { |
| 22 | r.Get("/{repo}/images", s.imageList) |
| 23 | r.Group(func(r chi.Router) { |
| 24 | r.Use(s.requireAdmin) |
| 25 | r.Post("/{repo}/images/delete", s.deleteImage) |
| 26 | r.Post("/{repo}/images/delete-all", s.deleteAllImages) |
| 27 | }) |
| 28 | } |
| 29 | |
| 30 | func (s *Server) imageList(w http.ResponseWriter, r *http.Request) { |
| 31 | repo, ok := s.visibleRepo(w, r) |
| 32 | if !ok { |
| 33 | return |
| 34 | } |
| 35 | u := User(r) |
| 36 | if !s.Cfg.CanPullImages(repo.IsPrivate, u != nil, u != nil && u.IsAdmin) { |
| 37 | http.Error(w, "Images are not visible to you", http.StatusForbidden) |
| 38 | return |
| 39 | } |
| 40 | tags, err := s.DB.ListImageTags(r.Context(), repo.ID) |
| 41 | if err != nil { |
| 42 | http.Error(w, "Database error", http.StatusInternalServerError) |
| 43 | return |
| 44 | } |
| 45 | sizes, err := s.DB.ImageSizes(r.Context(), repo.ID) |
| 46 | if err != nil { |
| 47 | http.Error(w, "Database error", http.StatusInternalServerError) |
| 48 | return |
| 49 | } |
| 50 | views.Render(w, http.StatusOK, views.ImageList(s.Cfg, u, repo, tags, sizes)) |
| 51 | } |
| 52 | |
| 53 | // deleteImage removes one tag, or the whole image path when no tag is given. |
| 54 | // A manifest left without tags goes too, with the layers only it used. |
| 55 | func (s *Server) deleteImage(w http.ResponseWriter, r *http.Request) { |
| 56 | repo, ok := s.adminRepo(w, r) |
| 57 | if !ok { |
| 58 | return |
| 59 | } |
| 60 | image := r.FormValue("image") |
| 61 | tag := r.FormValue("tag") |
| 62 | if tag == "" { |
| 63 | if err := s.purgeImages(r, repo.ID, &image); err != nil { |
| 64 | http.Error(w, "Database error", http.StatusInternalServerError) |
| 65 | return |
| 66 | } |
| 67 | redirectTo(w, r, "/"+repo.Name+"/images") |
| 68 | return |
| 69 | } |
| 70 | s.registryMu.Lock() |
| 71 | digest, err := s.DB.DeleteTag(r.Context(), repo.ID, image, tag) |
| 72 | if err == nil && digest != "" { |
| 73 | s.dropUnused(r.Context(), repo.ID, image, []string{digest}, nil) |
| 74 | } |
| 75 | s.registryMu.Unlock() |
| 76 | if err != nil { |
| 77 | http.Error(w, "Database error", http.StatusInternalServerError) |
| 78 | return |
| 79 | } |
| 80 | redirectTo(w, r, "/"+repo.Name+"/images") |
| 81 | } |
| 82 | |
| 83 | func (s *Server) deleteAllImages(w http.ResponseWriter, r *http.Request) { |
| 84 | repo, ok := s.adminRepo(w, r) |
| 85 | if !ok { |
| 86 | return |
| 87 | } |
| 88 | if err := s.purgeImages(r, repo.ID, nil); err != nil { |
| 89 | http.Error(w, "Database error", http.StatusInternalServerError) |
| 90 | return |
| 91 | } |
| 92 | redirectTo(w, r, "/"+repo.Name+"/images") |
| 93 | } |
| 94 | |
| 95 | // purgeImages drops one image path, or every image when image is nil, and |
| 96 | // removes the files nothing else uses. |
| 97 | func (s *Server) purgeImages(r *http.Request, repoID int64, image *string) error { |
| 98 | s.registryMu.Lock() |
| 99 | defer s.registryMu.Unlock() |
| 100 | digests, err := s.DB.RepoDigests(r.Context(), repoID, image) |
| 101 | if err != nil { |
| 102 | return err |
| 103 | } |
| 104 | if err := s.DB.DeleteImages(r.Context(), repoID, image); err != nil { |
| 105 | return err |
| 106 | } |
| 107 | for _, d := range digests { |
| 108 | s.removeUnreferenced(r.Context(), d) |
| 109 | } |
| 110 | return nil |
| 111 | } |
| 112 | |
| 113 | // deleteRepoRow removes a repository row and its image files. The cascade |
| 114 | // drops the index rows; the files nothing else uses go afterwards. Every |
| 115 | // place that deletes a repo goes through here. |
| 116 | func (s *Server) deleteRepoRow(r *http.Request, repoID int64) error { |
| 117 | s.registryMu.Lock() |
| 118 | defer s.registryMu.Unlock() |
| 119 | digests, err := s.DB.RepoDigests(r.Context(), repoID, nil) |
| 120 | if err != nil { |
| 121 | return err |
| 122 | } |
| 123 | if err := s.DB.DeleteRepo(r.Context(), repoID); err != nil { |
| 124 | return err |
| 125 | } |
| 126 | for _, d := range digests { |
| 127 | s.removeUnreferenced(r.Context(), d) |
| 128 | } |
| 129 | return nil |
| 130 | } |
| 131 | |
| 132 | // dropUnused deletes the given manifests that no tag and no other manifest |
| 133 | // of the image use. Then it unlinks the given blobs and the blobs of the |
| 134 | // deleted manifests that no remaining manifest uses. Children of a deleted |
| 135 | // index go the same way. The caller holds registryMu. Errors are only |
| 136 | // logged: the change that triggered the cleanup is already stored, and the |
| 137 | // startup sweep removes what is left. |
| 138 | // |
| 139 | // ponytail: a push in flight that shares a layer only with a dropped |
| 140 | // manifest fails its manifest PUT with MANIFEST_BLOB_UNKNOWN. Two CI |
| 141 | // pipelines that push the same image path can hit this. A retry works. |
| 142 | func (s *Server) dropUnused(ctx context.Context, repoID int64, image string, manifests, blobs []string) { |
| 143 | if err := s.tryDropUnused(context.WithoutCancel(ctx), repoID, image, manifests, blobs); err != nil { |
| 144 | log.Printf("registry cleanup of repo %d image %q: %v", repoID, image, err) |
| 145 | } |
| 146 | } |
| 147 | |
| 148 | func (s *Server) tryDropUnused(ctx context.Context, repoID int64, image string, manifests, blobs []string) error { |
| 149 | for len(manifests) > 0 { |
| 150 | used, err := s.imageRefs(ctx, repoID, image) |
| 151 | if err != nil { |
| 152 | return err |
| 153 | } |
| 154 | var children []string |
| 155 | for _, d := range manifests { |
| 156 | if used[d] { |
| 157 | continue |
| 158 | } |
| 159 | // A broken file only leaves its blobs linked until the sweep. |
| 160 | refs, _ := s.readManifestRefs(d) |
| 161 | gone, err := s.DB.DeleteUntaggedManifest(ctx, repoID, image, d) |
| 162 | if err != nil { |
| 163 | return err |
| 164 | } |
| 165 | if !gone { |
| 166 | continue |
| 167 | } |
| 168 | s.removeUnreferenced(ctx, d) |
| 169 | blobs = append(blobs, refs.blobs()...) |
| 170 | children = append(children, refs.children()...) |
| 171 | } |
| 172 | manifests = children |
| 173 | } |
| 174 | if len(blobs) == 0 { |
| 175 | return nil |
| 176 | } |
| 177 | used, err := s.imageRefs(ctx, repoID, image) |
| 178 | if err != nil { |
| 179 | return err |
| 180 | } |
| 181 | for _, b := range blobs { |
| 182 | if used[b] { |
| 183 | continue |
| 184 | } |
| 185 | if err := s.DB.UnlinkBlob(ctx, repoID, image, b); err != nil { |
| 186 | return err |
| 187 | } |
| 188 | s.removeUnreferenced(ctx, b) |
| 189 | } |
| 190 | return nil |
| 191 | } |
| 192 | |
| 193 | // imageRefs collects every blob and child manifest digest that the image's |
| 194 | // manifests reference. It fails on a manifest it cannot read, because a |
| 195 | // missing entry would make a used blob look unused. |
| 196 | func (s *Server) imageRefs(ctx context.Context, repoID int64, image string) (map[string]bool, error) { |
| 197 | digests, err := s.DB.ManifestDigests(ctx, repoID, image) |
| 198 | if err != nil { |
| 199 | return nil, err |
| 200 | } |
| 201 | used := map[string]bool{} |
| 202 | for _, d := range digests { |
| 203 | refs, err := s.readManifestRefs(d) |
| 204 | if err != nil { |
| 205 | return nil, fmt.Errorf("manifest %s: %w", d, err) |
| 206 | } |
| 207 | for _, r := range append(refs.blobs(), refs.children()...) { |
| 208 | used[r] = true |
| 209 | } |
| 210 | } |
| 211 | return used, nil |
| 212 | } |
| 213 | |
| 214 | func (s *Server) readManifestRefs(digest string) (manifestRefs, error) { |
| 215 | var refs manifestRefs |
| 216 | body, err := os.ReadFile(s.blobPath(digest)) |
| 217 | if err != nil { |
| 218 | return refs, err |
| 219 | } |
| 220 | return refs, json.Unmarshal(body, &refs) |
| 221 | } |
| 222 | |
| 223 | // SweepRegistry drops untagged manifests and unused blob links of every |
| 224 | // image, then deletes blob files that no index row references. It must run |
| 225 | // when no push is in flight, because a push links its layers before its |
| 226 | // manifest. |
| 227 | func (s *Server) SweepRegistry(ctx context.Context) error { |
| 228 | s.registryMu.Lock() |
| 229 | defer s.registryMu.Unlock() |
| 230 | images, err := s.DB.RegistryImages(ctx) |
| 231 | if err != nil { |
| 232 | return err |
| 233 | } |
| 234 | for _, img := range images { |
| 235 | manifests, err := s.DB.ManifestDigests(ctx, img.RepoID, img.Image) |
| 236 | if err != nil { |
| 237 | return err |
| 238 | } |
| 239 | blobs, err := s.DB.BlobDigests(ctx, img.RepoID, img.Image) |
| 240 | if err != nil { |
| 241 | return err |
| 242 | } |
| 243 | s.dropUnused(ctx, img.RepoID, img.Image, manifests, blobs) |
| 244 | } |
| 245 | root, err := os.OpenRoot(filepath.Join(s.Cfg.RegistryDir(), "blobs")) |
| 246 | if errors.Is(err, fs.ErrNotExist) { |
| 247 | return nil |
| 248 | } |
| 249 | if err != nil { |
| 250 | return err |
| 251 | } |
| 252 | defer func() { _ = root.Close() }() |
| 253 | files := 0 |
| 254 | err = fs.WalkDir(root.FS(), ".", func(path string, e fs.DirEntry, err error) error { |
| 255 | if err != nil || e.IsDir() { |
| 256 | return err |
| 257 | } |
| 258 | // Keep in sync with blobPath: blobs/<algorithm>/<hex>. |
| 259 | digest := filepath.Base(filepath.Dir(path)) + ":" + e.Name() |
| 260 | if used, err := s.DB.DigestReferenced(ctx, digest); err != nil || used { |
| 261 | return nil |
| 262 | } |
| 263 | if root.Remove(path) == nil { |
| 264 | files++ |
| 265 | } |
| 266 | return nil |
| 267 | }) |
| 268 | if files > 0 { |
| 269 | log.Printf("registry sweep: removed %d unused files", files) |
| 270 | } |
| 271 | return err |
| 272 | } |
| 273 |