|
| 1 | +package garbagecollector |
| 2 | + |
| 3 | +import ( |
| 4 | + "context" |
| 5 | + "errors" |
| 6 | + "fmt" |
| 7 | + "github.com/octohelm/x/ptr" |
| 8 | + "log/slog" |
| 9 | + "time" |
| 10 | + |
| 11 | + "github.com/distribution/reference" |
| 12 | + "github.com/go-courier/logr" |
| 13 | + manifestv1 "github.com/octohelm/crkit/pkg/apis/manifest/v1" |
| 14 | + "github.com/octohelm/crkit/pkg/content" |
| 15 | + "github.com/octohelm/crkit/pkg/content/fs/driver" |
| 16 | + "github.com/opencontainers/go-digest" |
| 17 | +) |
| 18 | + |
| 19 | +func MarkAndSweepExcludeModifiedIn( |
| 20 | + ctx context.Context, |
| 21 | + namespace content.Namespace, d driver.Driver, |
| 22 | + excludeModifiedIn time.Duration, |
| 23 | + dryRun bool, |
| 24 | +) error { |
| 25 | + repositoryNameIterable, ok := namespace.(content.RepositoryNameIterable) |
| 26 | + if !ok { |
| 27 | + return &content.ErrNotImplemented{Reason: errors.New("RepositoryNameIterable of Namespace")} |
| 28 | + } |
| 29 | + |
| 30 | + blobDigestIterable, ok := namespace.(content.DigestIterable) |
| 31 | + if !ok { |
| 32 | + return &content.ErrNotImplemented{Reason: errors.New("DigestIterable of Namespace")} |
| 33 | + } |
| 34 | + |
| 35 | + stabled := time.Now().Add(-excludeModifiedIn) |
| 36 | + |
| 37 | + c := &collector{ |
| 38 | + Vacuum: NewVacuum(d, dryRun), |
| 39 | + namespace: namespace, |
| 40 | + used: map[digest.Digest]struct{}{}, |
| 41 | + recentlyActivated: func(modTime time.Time) bool { |
| 42 | + return modTime.After(stabled) |
| 43 | + }, |
| 44 | + } |
| 45 | + |
| 46 | + return c.MarkAndSweep(ctx, repositoryNameIterable, blobDigestIterable) |
| 47 | +} |
| 48 | + |
| 49 | +type collector struct { |
| 50 | + Vacuum |
| 51 | + namespace content.Namespace |
| 52 | + used map[digest.Digest]struct{} |
| 53 | + recentlyActivated func(modTime time.Time) bool |
| 54 | +} |
| 55 | + |
| 56 | +func (c *collector) count(dgst digest.Digest) { |
| 57 | + c.used[dgst] = struct{}{} |
| 58 | +} |
| 59 | + |
| 60 | +func (c *collector) referenced(dgst digest.Digest) bool { |
| 61 | + _, ok := c.used[dgst] |
| 62 | + return ok |
| 63 | +} |
| 64 | + |
| 65 | +func (c *collector) referencedOrRecentlyActivated(ld content.LinkedDigest) bool { |
| 66 | + if c.recentlyActivated(ld.ModTime) { |
| 67 | + c.count(ld.Digest) |
| 68 | + |
| 69 | + return true |
| 70 | + } |
| 71 | + |
| 72 | + _, ok := c.used[ld.Digest] |
| 73 | + return ok |
| 74 | +} |
| 75 | + |
| 76 | +func (c *collector) MarkAndSweep(pctx context.Context, repositoryNameIterable content.RepositoryNameIterable, blobDigestIterable content.DigestIterable) error { |
| 77 | + ctx, l := logr.FromContext(pctx).Start(pctx, "MarkAndSweep") |
| 78 | + defer l.End() |
| 79 | + |
| 80 | + for named, err := range repositoryNameIterable.RepositoryNames(ctx) { |
| 81 | + if err != nil { |
| 82 | + return err |
| 83 | + } |
| 84 | + |
| 85 | + if err := c.markAndSweepRepository(ctx, named); err != nil { |
| 86 | + return fmt.Errorf("failed to mark and sweep repository %s: %w", named, err) |
| 87 | + } |
| 88 | + } |
| 89 | + |
| 90 | + for d, err := range blobDigestIterable.Digests(ctx) { |
| 91 | + if err != nil { |
| 92 | + return err |
| 93 | + } |
| 94 | + |
| 95 | + l.WithValues(slog.String("blob", string(d))).Debug("checking") |
| 96 | + |
| 97 | + if c.referenced(d) { |
| 98 | + continue |
| 99 | + } |
| 100 | + |
| 101 | + if err := c.RemoveBlob(ctx, d); err != nil { |
| 102 | + return fmt.Errorf("failed to remove blob %s: %w", d, err) |
| 103 | + } |
| 104 | + } |
| 105 | + |
| 106 | + return nil |
| 107 | +} |
| 108 | + |
| 109 | +func (c *collector) markAndSweepRepository(ctx context.Context, named reference.Named) error { |
| 110 | + l := logr.FromContext(ctx) |
| 111 | + |
| 112 | + repository, err := c.namespace.Repository(ctx, named) |
| 113 | + if err != nil { |
| 114 | + return fmt.Errorf("failed to construct repository: %w", err) |
| 115 | + } |
| 116 | + |
| 117 | + tagService, err := repository.Tags(ctx) |
| 118 | + if err != nil { |
| 119 | + return fmt.Errorf("failed to tag service: %w", err) |
| 120 | + } |
| 121 | + |
| 122 | + manifestService, err := repository.Manifests(ctx) |
| 123 | + if err != nil { |
| 124 | + return fmt.Errorf("failed to manifest service: %w", err) |
| 125 | + } |
| 126 | + |
| 127 | + blobStore, err := repository.Blobs(ctx) |
| 128 | + if err != nil { |
| 129 | + return fmt.Errorf("failed to blob store: %w", err) |
| 130 | + } |
| 131 | + |
| 132 | + manifestDigestIterable, ok := manifestService.(content.LinkedDigestIterable) |
| 133 | + if !ok { |
| 134 | + return &content.ErrNotImplemented{Reason: errors.New("LinkedDigestIterable of ManifestService")} |
| 135 | + } |
| 136 | + |
| 137 | + layerDigestIterable, ok := blobStore.(content.LinkedDigestIterable) |
| 138 | + if !ok { |
| 139 | + return &content.ErrNotImplemented{Reason: errors.New("LinkedDigestIterable of BlobStore")} |
| 140 | + } |
| 141 | + |
| 142 | + allTags, err := tagService.All(ctx) |
| 143 | + if err != nil { |
| 144 | + return fmt.Errorf("failed to get all tags: %w", err) |
| 145 | + } |
| 146 | + |
| 147 | + for _, tag := range allTags { |
| 148 | + d, err := tagService.Get(ctx, tag) |
| 149 | + if err != nil { |
| 150 | + return err |
| 151 | + } |
| 152 | + |
| 153 | + if err := c.markManifest(ctx, manifestService, d.Digest); err != nil { |
| 154 | + return err |
| 155 | + } |
| 156 | + } |
| 157 | + |
| 158 | + l.WithValues(slog.String("name", named.String())).Info("marking") |
| 159 | + |
| 160 | + for ld, err := range manifestDigestIterable.LinkedDigests(ctx) { |
| 161 | + if err != nil { |
| 162 | + return err |
| 163 | + } |
| 164 | + |
| 165 | + l.WithValues( |
| 166 | + slog.String("name", named.String()), |
| 167 | + slog.String("manifest", string(ld.Digest)), |
| 168 | + ).Info("checking") |
| 169 | + |
| 170 | + if c.referencedOrRecentlyActivated(ld) { |
| 171 | + continue |
| 172 | + } |
| 173 | + |
| 174 | + if err := c.RemoveManifest(ctx, named, ld.Digest, allTags); err != nil { |
| 175 | + return fmt.Errorf("failed to remove manifest %s@%s: %w", named, ld.Digest, err) |
| 176 | + } |
| 177 | + } |
| 178 | + |
| 179 | + for ld, err := range layerDigestIterable.LinkedDigests(ctx) { |
| 180 | + if err != nil { |
| 181 | + return err |
| 182 | + } |
| 183 | + |
| 184 | + l.WithValues( |
| 185 | + slog.String("name", named.String()), |
| 186 | + slog.String("layer", string(ld.Digest)), |
| 187 | + ).Debug("checking") |
| 188 | + |
| 189 | + if c.referencedOrRecentlyActivated(ld) { |
| 190 | + continue |
| 191 | + } |
| 192 | + |
| 193 | + if err := c.RemoveLayer(ctx, named, ld.Digest); err != nil { |
| 194 | + return fmt.Errorf("failed to remove layer %s@%s: %w", named, ld.Digest, err) |
| 195 | + } |
| 196 | + } |
| 197 | + |
| 198 | + return nil |
| 199 | +} |
| 200 | + |
| 201 | +func (c *collector) markManifest(ctx context.Context, manifestService content.ManifestService, manifestDigest digest.Digest) error { |
| 202 | + m, err := manifestService.Get(ctx, manifestDigest) |
| 203 | + if err != nil { |
| 204 | + return err |
| 205 | + } |
| 206 | + |
| 207 | + c.count(manifestDigest) |
| 208 | + |
| 209 | + switch m.Type() { |
| 210 | + case manifestv1.DockerMediaTypeManifestList, manifestv1.MediaTypeImageIndex: |
| 211 | + for d := range m.References() { |
| 212 | + if err := c.markManifest(ctx, manifestService, d.Digest); err != nil { |
| 213 | + // skip for partial cached |
| 214 | + if errors.As(err, ptr.Ptr(&content.ErrManifestUnknownRevision{})) { |
| 215 | + continue |
| 216 | + } |
| 217 | + return err |
| 218 | + } |
| 219 | + } |
| 220 | + default: |
| 221 | + for d := range m.References() { |
| 222 | + c.count(d.Digest) |
| 223 | + } |
| 224 | + } |
| 225 | + |
| 226 | + return nil |
| 227 | +} |
0 commit comments