| 1 | // Package gc provides garbage collection for go-ipfs. |
| 2 | package gc |
| 3 | |
| 4 | import ( |
| 5 | "context" |
| 6 | "errors" |
| 7 | "fmt" |
| 8 | "strings" |
| 9 | |
| 10 | bserv "github.com/ipfs/boxo/blockservice" |
| 11 | bstore "github.com/ipfs/boxo/blockstore" |
| 12 | offline "github.com/ipfs/boxo/exchange/offline" |
| 13 | dag "github.com/ipfs/boxo/ipld/merkledag" |
| 14 | pin "github.com/ipfs/boxo/pinning/pinner" |
| 15 | "github.com/ipfs/boxo/verifcid" |
| 16 | cid "github.com/ipfs/go-cid" |
| 17 | dstore "github.com/ipfs/go-datastore" |
| 18 | ipld "github.com/ipfs/go-ipld-format" |
| 19 | logging "github.com/ipfs/go-log/v2" |
| 20 | ) |
| 21 | |
| 22 | var log = logging.Logger("gc") |
| 23 | |
| 24 | // Result represents an incremental output from a garbage collection |
| 25 | // run. It contains either an error, or the cid of a removed object. |
| 26 | type Result struct { |
| 27 | KeyRemoved cid.Cid |
| 28 | Error error |
| 29 | } |
| 30 | |
| 31 | // converts a set of CIDs with different codecs to a set of CIDs with the raw codec. |
| 32 | func toRawCids(set *cid.Set) (*cid.Set, error) { |
| 33 | newSet := cid.NewSet() |
| 34 | err := set.ForEach(func(c cid.Cid) error { |
| 35 | newSet.Add(cid.NewCidV1(cid.Raw, c.Hash())) |
| 36 | return nil |
| 37 | }) |
| 38 | return newSet, err |
| 39 | } |
| 40 | |
| 41 | // GC performs a mark and sweep garbage collection of the blocks in the blockstore |
| 42 | // first, it creates a 'marked' set and adds to it the following: |
| 43 | // - all recursively pinned blocks, plus all of their descendants (recursively) |
| 44 | // - bestEffortRoots, plus all of its descendants (recursively) |
| 45 | // - all directly pinned blocks |
| 46 | // - all blocks utilized internally by the pinner |
| 47 | // |
| 48 | // The routine then iterates over every block in the blockstore and |
| 49 | // deletes any block that is not found in the marked set. |
| 50 | func GC(ctx context.Context, bs bstore.GCBlockstore, dstor dstore.Datastore, pn pin.Pinner, bestEffortRoots []cid.Cid) <-chan Result { |
| 51 | ctx, cancel := context.WithCancel(ctx) |
| 52 | |
| 53 | unlocker := bs.GCLock(ctx) |
| 54 | |
| 55 | bsrv := bserv.New(bs, offline.Exchange(bs)) |
| 56 | ds := dag.NewDAGService(bsrv) |
| 57 | |
| 58 | output := make(chan Result, 128) |
| 59 | |
| 60 | go func() { |
| 61 | defer cancel() |
| 62 | defer close(output) |
| 63 | defer unlocker.Unlock(ctx) |
| 64 | |
| 65 | gcs, err := ColoredSet(ctx, pn, ds, bestEffortRoots, output) |
| 66 | if err != nil { |
| 67 | select { |
| 68 | case output <- Result{Error: err}: |
| 69 | case <-ctx.Done(): |
| 70 | } |
| 71 | return |
| 72 | } |
| 73 | |
| 74 | // The blockstore reports raw blocks. We need to remove the codecs from the CIDs. |
| 75 | gcs, err = toRawCids(gcs) |
| 76 | if err != nil { |
| 77 | select { |
| 78 | case output <- Result{Error: err}: |
| 79 | case <-ctx.Done(): |
| 80 | } |
| 81 | return |
| 82 | } |
| 83 | |
| 84 | keychain, err := bs.AllKeysChan(ctx) |
| 85 | if err != nil { |
| 86 | select { |
| 87 | case output <- Result{Error: err}: |
| 88 | case <-ctx.Done(): |
| 89 | } |
| 90 | return |
| 91 | } |
| 92 | |
| 93 | errors := false |
| 94 | var removed uint64 |
| 95 | |
| 96 | loop: |
| 97 | for ctx.Err() == nil { // select may not notice that we're "done". |
| 98 | select { |
| 99 | case k, ok := <-keychain: |
| 100 | if !ok { |
| 101 | break loop |
| 102 | } |
| 103 | // NOTE: assumes that all CIDs returned by the keychain are _raw_ CIDv1 CIDs. |
| 104 | // This means we keep the block as long as we want it somewhere (CIDv1, CIDv0, Raw, other...). |
| 105 | if !gcs.Has(k) { |
| 106 | err := bs.DeleteBlock(ctx, k) |
| 107 | removed++ |
| 108 | if err != nil { |
| 109 | errors = true |
| 110 | select { |
| 111 | case output <- Result{Error: &CannotDeleteBlockError{k, err}}: |
| 112 | case <-ctx.Done(): |
| 113 | break loop |
| 114 | } |
| 115 | // continue as error is non-fatal |
| 116 | continue loop |
| 117 | } |
| 118 | select { |
| 119 | case output <- Result{KeyRemoved: k}: |
| 120 | case <-ctx.Done(): |
| 121 | break loop |
| 122 | } |
| 123 | } |
| 124 | case <-ctx.Done(): |
| 125 | break loop |
| 126 | } |
| 127 | } |
| 128 | if errors { |
| 129 | select { |
| 130 | case output <- Result{Error: ErrCannotDeleteSomeBlocks}: |
| 131 | case <-ctx.Done(): |
| 132 | return |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | gds, ok := dstor.(dstore.GCDatastore) |
| 137 | if !ok { |
| 138 | return |
| 139 | } |
| 140 | |
| 141 | err = gds.CollectGarbage(ctx) |
| 142 | if err != nil { |
| 143 | select { |
| 144 | case output <- Result{Error: err}: |
| 145 | case <-ctx.Done(): |
| 146 | } |
| 147 | return |
| 148 | } |
| 149 | }() |
| 150 | |
| 151 | return output |
| 152 | } |
| 153 | |
| 154 | // Descendants recursively finds all the descendants of the given roots and |
| 155 | // adds them to the given cid.Set, using the provided dag.GetLinks function |
| 156 | // to walk the tree. |
| 157 | func Descendants(ctx context.Context, getLinks dag.GetLinks, set *cid.Set, roots <-chan pin.StreamedPin) error { |
| 158 | verifyGetLinks := func(ctx context.Context, c cid.Cid) ([]*ipld.Link, error) { |
| 159 | err := verifcid.ValidateCid(verifcid.DefaultAllowlist, c) |
| 160 | if err != nil { |
| 161 | return nil, err |
| 162 | } |
| 163 | |
| 164 | return getLinks(ctx, c) |
| 165 | } |
| 166 | |
| 167 | verboseCidError := func(err error) error { |
| 168 | if strings.Contains(err.Error(), verifcid.ErrDigestTooSmall.Error()) || |
| 169 | strings.Contains(err.Error(), verifcid.ErrPossiblyInsecureHashFunction.Error()) { |
| 170 | err = fmt.Errorf("\"%s\"\nPlease run 'ipfs pin verify'"+ // nolint |
| 171 | " to list insecure hashes. If you want to read them,"+ |
| 172 | " please downgrade your go-ipfs to 0.4.13\n", err) |
| 173 | log.Error(err) |
| 174 | } |
| 175 | return err |
| 176 | } |
| 177 | |
| 178 | for { |
| 179 | select { |
| 180 | case <-ctx.Done(): |
| 181 | return ctx.Err() |
| 182 | case wrapper, ok := <-roots: |
| 183 | if !ok { |
| 184 | return nil |
| 185 | } |
| 186 | if wrapper.Err != nil { |
| 187 | return wrapper.Err |
| 188 | } |
| 189 | |
| 190 | // Walk recursively walks the dag and adds the keys to the given set |
| 191 | err := dag.Walk(ctx, verifyGetLinks, wrapper.Pin.Key, func(k cid.Cid) bool { |
| 192 | return set.Visit(toCidV1(k)) |
| 193 | }, dag.Concurrent()) |
| 194 | if err != nil { |
| 195 | err = verboseCidError(err) |
| 196 | return err |
| 197 | } |
| 198 | } |
| 199 | } |
| 200 | } |
| 201 | |
| 202 | // toCidV1 converts any CIDv0s to CIDv1s. |
| 203 | func toCidV1(c cid.Cid) cid.Cid { |
| 204 | if c.Version() == 0 { |
| 205 | return cid.NewCidV1(c.Type(), c.Hash()) |
| 206 | } |
| 207 | return c |
| 208 | } |
| 209 | |
| 210 | // ColoredSet computes the set of nodes in the graph that are pinned by the |
| 211 | // pins in the given pinner. |
| 212 | func ColoredSet(ctx context.Context, pn pin.Pinner, ng ipld.NodeGetter, bestEffortRoots []cid.Cid, output chan<- Result) (*cid.Set, error) { |
| 213 | // KeySet currently implemented in memory, in the future, may be bloom filter or |
| 214 | // disk backed to conserve memory. |
| 215 | errors := false |
| 216 | gcs := cid.NewSet() |
| 217 | getLinks := func(ctx context.Context, cid cid.Cid) ([]*ipld.Link, error) { |
| 218 | links, err := ipld.GetLinks(ctx, ng, cid) |
| 219 | if err != nil { |
| 220 | errors = true |
| 221 | select { |
| 222 | case output <- Result{Error: &CannotFetchLinksError{cid, err}}: |
| 223 | case <-ctx.Done(): |
| 224 | return nil, ctx.Err() |
| 225 | } |
| 226 | } |
| 227 | return links, nil |
| 228 | } |
| 229 | rkeys := pn.RecursiveKeys(ctx, false) |
| 230 | err := Descendants(ctx, getLinks, gcs, rkeys) |
| 231 | if err != nil { |
| 232 | errors = true |
| 233 | select { |
| 234 | case output <- Result{Error: err}: |
| 235 | case <-ctx.Done(): |
| 236 | return nil, ctx.Err() |
| 237 | } |
| 238 | } |
| 239 | |
| 240 | bestEffortGetLinks := func(ctx context.Context, cid cid.Cid) ([]*ipld.Link, error) { |
| 241 | links, err := ipld.GetLinks(ctx, ng, cid) |
| 242 | if err != nil && !ipld.IsNotFound(err) { |
| 243 | errors = true |
| 244 | select { |
| 245 | case output <- Result{Error: &CannotFetchLinksError{cid, err}}: |
| 246 | case <-ctx.Done(): |
| 247 | return nil, ctx.Err() |
| 248 | } |
| 249 | } |
| 250 | return links, nil |
| 251 | } |
| 252 | bestEffortRootsChan := make(chan pin.StreamedPin) |
| 253 | go func() { |
| 254 | defer close(bestEffortRootsChan) |
| 255 | for _, root := range bestEffortRoots { |
| 256 | select { |
| 257 | case <-ctx.Done(): |
| 258 | return |
| 259 | case bestEffortRootsChan <- pin.StreamedPin{Pin: pin.Pinned{Key: root}}: |
| 260 | } |
| 261 | } |
| 262 | }() |
| 263 | err = Descendants(ctx, bestEffortGetLinks, gcs, bestEffortRootsChan) |
| 264 | if err != nil { |
| 265 | errors = true |
| 266 | select { |
| 267 | case output <- Result{Error: err}: |
| 268 | case <-ctx.Done(): |
| 269 | return nil, ctx.Err() |
| 270 | } |
| 271 | } |
| 272 | |
| 273 | dkeys := pn.DirectKeys(ctx, false) |
| 274 | for k := range dkeys { |
| 275 | if k.Err != nil { |
| 276 | return nil, k.Err |
| 277 | } |
| 278 | gcs.Add(toCidV1(k.Pin.Key)) |
| 279 | } |
| 280 | |
| 281 | ikeys := pn.InternalPins(ctx, false) |
| 282 | err = Descendants(ctx, getLinks, gcs, ikeys) |
| 283 | if err != nil { |
| 284 | errors = true |
| 285 | select { |
| 286 | case output <- Result{Error: err}: |
| 287 | case <-ctx.Done(): |
| 288 | return nil, ctx.Err() |
| 289 | } |
| 290 | } |
| 291 | |
| 292 | if errors { |
| 293 | return nil, ErrCannotFetchAllLinks |
| 294 | } |
| 295 | |
| 296 | return gcs, nil |
| 297 | } |
| 298 | |
| 299 | // ErrCannotFetchAllLinks is returned as the last Result in the GC output |
| 300 | // channel when there was an error creating the marked set because of a |
| 301 | // problem when finding descendants. |
| 302 | var ErrCannotFetchAllLinks = errors.New("garbage collection aborted: could not retrieve some links") |
| 303 | |
| 304 | // ErrCannotDeleteSomeBlocks is returned when removing blocks marked for |
| 305 | // deletion fails as the last Result in GC output channel. |
| 306 | var ErrCannotDeleteSomeBlocks = errors.New("garbage collection incomplete: could not delete some blocks") |
| 307 | |
| 308 | // CannotFetchLinksError provides detailed information about which links |
| 309 | // could not be fetched and can appear as a Result in the GC output channel. |
| 310 | type CannotFetchLinksError struct { |
| 311 | Key cid.Cid |
| 312 | Err error |
| 313 | } |
| 314 | |
| 315 | // Error implements the error interface for this type with a useful |
| 316 | // message. |
| 317 | func (e *CannotFetchLinksError) Error() string { |
| 318 | return fmt.Sprintf("could not retrieve links for %s: %s", e.Key, e.Err) |
| 319 | } |
| 320 | |
| 321 | // CannotDeleteBlockError provides detailed information about which |
| 322 | // blocks could not be deleted and can appear as a Result in the GC output |
| 323 | // channel. |
| 324 | type CannotDeleteBlockError struct { |
| 325 | Key cid.Cid |
| 326 | Err error |
| 327 | } |
| 328 | |
| 329 | // Error implements the error interface for this type with a |
| 330 | // useful message. |
| 331 | func (e *CannotDeleteBlockError) Error() string { |
| 332 | return fmt.Sprintf("could not remove %s: %s", e.Key, e.Err) |
| 333 | } |