You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

storage_load_package.go 11 kB

1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
1 year ago
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366
  1. package task
  2. import (
  3. "fmt"
  4. "io"
  5. "math"
  6. "os"
  7. "path/filepath"
  8. "time"
  9. "github.com/samber/lo"
  10. "gitlink.org.cn/cloudream/common/pkgs/bitmap"
  11. "gitlink.org.cn/cloudream/common/pkgs/logger"
  12. "gitlink.org.cn/cloudream/common/pkgs/task"
  13. cdssdk "gitlink.org.cn/cloudream/common/sdks/storage"
  14. "gitlink.org.cn/cloudream/common/utils/io2"
  15. "gitlink.org.cn/cloudream/common/utils/reflect2"
  16. "gitlink.org.cn/cloudream/common/utils/sort2"
  17. "gitlink.org.cn/cloudream/storage/common/consts"
  18. stgglb "gitlink.org.cn/cloudream/storage/common/globals"
  19. stgmod "gitlink.org.cn/cloudream/storage/common/models"
  20. "gitlink.org.cn/cloudream/storage/common/pkgs/distlock/reqbuilder"
  21. "gitlink.org.cn/cloudream/storage/common/pkgs/ec"
  22. coormq "gitlink.org.cn/cloudream/storage/common/pkgs/mq/coordinator"
  23. "gitlink.org.cn/cloudream/storage/common/pkgs/storage/types"
  24. "gitlink.org.cn/cloudream/storage/common/utils"
  25. )
  26. type StorageLoadPackage struct {
  27. PackagePath string
  28. LocalBase string
  29. RemoteBase string
  30. userID cdssdk.UserID
  31. packageID cdssdk.PackageID
  32. storageID cdssdk.StorageID
  33. pinnedBlocks []stgmod.ObjectBlock
  34. }
  35. func NewStorageLoadPackage(userID cdssdk.UserID, packageID cdssdk.PackageID, storageID cdssdk.StorageID) *StorageLoadPackage {
  36. return &StorageLoadPackage{
  37. userID: userID,
  38. packageID: packageID,
  39. storageID: storageID,
  40. }
  41. }
  42. func (t *StorageLoadPackage) Execute(task *task.Task[TaskContext], ctx TaskContext, complete CompleteFn) {
  43. startTime := time.Now()
  44. log := logger.WithType[StorageLoadPackage]("Task")
  45. log.WithField("TaskID", task.ID()).
  46. Infof("begin to load package %v to %v", t.packageID, t.storageID)
  47. err := t.do(task, ctx)
  48. if err == nil {
  49. log.WithField("TaskID", task.ID()).
  50. Infof("loading success, cost: %v", time.Since(startTime))
  51. } else {
  52. log.WithField("TaskID", task.ID()).
  53. Warnf("loading package: %v, cost: %v", err, time.Since(startTime))
  54. }
  55. complete(err, CompleteOption{
  56. RemovingDelay: time.Minute,
  57. })
  58. }
  59. func (t *StorageLoadPackage) do(task *task.Task[TaskContext], ctx TaskContext) error {
  60. coorCli, err := stgglb.CoordinatorMQPool.Acquire()
  61. if err != nil {
  62. return fmt.Errorf("new coordinator client: %w", err)
  63. }
  64. defer stgglb.CoordinatorMQPool.Release(coorCli)
  65. getStgResp, err := coorCli.GetStorageDetails(coormq.ReqGetStorageDetails([]cdssdk.StorageID{t.storageID}))
  66. if err != nil {
  67. return fmt.Errorf("request to coordinator: %w", err)
  68. }
  69. if getStgResp.Storages[0] == nil {
  70. return fmt.Errorf("storage not found")
  71. }
  72. if getStgResp.Storages[0].Shared == nil {
  73. return fmt.Errorf("storage has shared storage")
  74. }
  75. t.PackagePath = utils.MakeLoadedPackagePath(t.userID, t.packageID)
  76. fullLocalPath := filepath.Join(getStgResp.Storages[0].Shared.LoadBase, t.PackagePath)
  77. if err = os.MkdirAll(fullLocalPath, 0755); err != nil {
  78. return fmt.Errorf("creating output directory: %w", err)
  79. }
  80. getObjectDetails, err := coorCli.GetPackageObjectDetails(coormq.ReqGetPackageObjectDetails(t.packageID))
  81. if err != nil {
  82. return fmt.Errorf("getting package object details: %w", err)
  83. }
  84. shardstore, err := ctx.stgMgr.GetShardStore(t.storageID)
  85. if err != nil {
  86. return fmt.Errorf("get shard store of storage %v: %w", t.storageID, err)
  87. }
  88. mutex, err := reqbuilder.NewBuilder().
  89. // 提前占位
  90. Metadata().StoragePackage().CreateOne(t.userID, t.storageID, t.packageID).
  91. // 保护在storage目录中下载的文件
  92. Storage().Buzy(t.storageID).
  93. // 保护下载文件时同时保存到IPFS的文件
  94. Shard().Buzy(t.storageID).
  95. MutexLock(ctx.distlock)
  96. if err != nil {
  97. return fmt.Errorf("acquire locks failed, err: %w", err)
  98. }
  99. defer mutex.Unlock()
  100. for _, obj := range getObjectDetails.Objects {
  101. err := t.downloadOne(coorCli, shardstore, fullLocalPath, obj)
  102. if err != nil {
  103. return err
  104. }
  105. ctx.accessStat.AddAccessCounter(obj.Object.ObjectID, t.packageID, t.storageID, 1)
  106. }
  107. _, err = coorCli.StoragePackageLoaded(coormq.NewStoragePackageLoaded(t.userID, t.storageID, t.packageID, t.pinnedBlocks))
  108. if err != nil {
  109. return fmt.Errorf("loading package to storage: %w", err)
  110. }
  111. // TODO 要防止下载的临时文件被删除
  112. return err
  113. }
  114. func (t *StorageLoadPackage) downloadOne(coorCli *coormq.Client, shardStore types.ShardStore, dir string, obj stgmod.ObjectDetail) error {
  115. var file io.ReadCloser
  116. switch red := obj.Object.Redundancy.(type) {
  117. case *cdssdk.NoneRedundancy:
  118. reader, err := t.downloadNoneOrRepObject(shardStore, obj)
  119. if err != nil {
  120. return fmt.Errorf("downloading object: %w", err)
  121. }
  122. file = reader
  123. case *cdssdk.RepRedundancy:
  124. reader, err := t.downloadNoneOrRepObject(shardStore, obj)
  125. if err != nil {
  126. return fmt.Errorf("downloading rep object: %w", err)
  127. }
  128. file = reader
  129. case *cdssdk.ECRedundancy:
  130. reader, pinnedBlocks, err := t.downloadECObject(coorCli, shardStore, obj, red)
  131. if err != nil {
  132. return fmt.Errorf("downloading ec object: %w", err)
  133. }
  134. file = reader
  135. t.pinnedBlocks = append(t.pinnedBlocks, pinnedBlocks...)
  136. default:
  137. return fmt.Errorf("unknow redundancy type: %v", reflect2.TypeOfValue(obj.Object.Redundancy))
  138. }
  139. defer file.Close()
  140. fullPath := filepath.Join(dir, obj.Object.Path)
  141. lastDirPath := filepath.Dir(fullPath)
  142. if err := os.MkdirAll(lastDirPath, 0755); err != nil {
  143. return fmt.Errorf("creating object last dir: %w", err)
  144. }
  145. outputFile, err := os.Create(fullPath)
  146. if err != nil {
  147. return fmt.Errorf("creating object file: %w", err)
  148. }
  149. defer outputFile.Close()
  150. if _, err := io.Copy(outputFile, file); err != nil {
  151. return fmt.Errorf("writting object to file: %w", err)
  152. }
  153. return nil
  154. }
  155. func (t *StorageLoadPackage) downloadNoneOrRepObject(shardStore types.ShardStore, obj stgmod.ObjectDetail) (io.ReadCloser, error) {
  156. if len(obj.Blocks) == 0 && len(obj.PinnedAt) == 0 {
  157. return nil, fmt.Errorf("no storage has this object")
  158. }
  159. file, err := shardStore.Open(types.NewOpen(obj.Object.FileHash))
  160. if err != nil {
  161. return nil, err
  162. }
  163. return file, nil
  164. }
  165. func (t *StorageLoadPackage) downloadECObject(coorCli *coormq.Client, shardStore types.ShardStore, obj stgmod.ObjectDetail, ecRed *cdssdk.ECRedundancy) (io.ReadCloser, []stgmod.ObjectBlock, error) {
  166. allStorages, err := t.sortDownloadStorages(coorCli, obj)
  167. if err != nil {
  168. return nil, nil, err
  169. }
  170. bsc, blocks := t.getMinReadingBlockSolution(allStorages, ecRed.K)
  171. osc, _ := t.getMinReadingObjectSolution(allStorages, ecRed.K)
  172. if bsc < osc {
  173. var fileStrs []io.ReadCloser
  174. rs, err := ec.NewStreamRs(ecRed.K, ecRed.N, ecRed.ChunkSize)
  175. if err != nil {
  176. return nil, nil, fmt.Errorf("new rs: %w", err)
  177. }
  178. for i := range blocks {
  179. str, err := shardStore.Open(types.NewOpen(blocks[i].Block.FileHash))
  180. if err != nil {
  181. for i -= 1; i >= 0; i-- {
  182. fileStrs[i].Close()
  183. }
  184. return nil, nil, fmt.Errorf("donwloading file: %w", err)
  185. }
  186. fileStrs = append(fileStrs, str)
  187. }
  188. fileReaders, filesCloser := io2.ToReaders(fileStrs)
  189. var indexes []int
  190. for _, b := range blocks {
  191. indexes = append(indexes, b.Block.Index)
  192. }
  193. outputs, outputsCloser := io2.ToReaders(rs.ReconstructData(fileReaders, indexes))
  194. return io2.AfterReadClosed(io2.Length(io2.ChunkedJoin(outputs, int(ecRed.ChunkSize)), obj.Object.Size), func(c io.ReadCloser) {
  195. filesCloser()
  196. outputsCloser()
  197. }), nil, nil
  198. }
  199. // bsc >= osc,如果osc是MaxFloat64,那么bsc也一定是,也就意味着没有足够块来恢复文件
  200. if osc == math.MaxFloat64 {
  201. return nil, nil, fmt.Errorf("no enough blocks to reconstruct the file, want %d, get only %d", ecRed.K, len(blocks))
  202. }
  203. // 如果是直接读取的文件,那么就不需要Pin文件块
  204. str, err := shardStore.Open(types.NewOpen(obj.Object.FileHash))
  205. return str, nil, err
  206. }
  207. type downloadStorageInfo struct {
  208. Storage stgmod.StorageDetail
  209. ObjectPinned bool
  210. Blocks []stgmod.ObjectBlock
  211. Distance float64
  212. }
  213. func (t *StorageLoadPackage) sortDownloadStorages(coorCli *coormq.Client, obj stgmod.ObjectDetail) ([]*downloadStorageInfo, error) {
  214. var stgIDs []cdssdk.StorageID
  215. for _, id := range obj.PinnedAt {
  216. if !lo.Contains(stgIDs, id) {
  217. stgIDs = append(stgIDs, id)
  218. }
  219. }
  220. for _, b := range obj.Blocks {
  221. if !lo.Contains(stgIDs, b.StorageID) {
  222. stgIDs = append(stgIDs, b.StorageID)
  223. }
  224. }
  225. getStgs, err := coorCli.GetStorageDetails(coormq.ReqGetStorageDetails(stgIDs))
  226. if err != nil {
  227. return nil, fmt.Errorf("getting storage details: %w", err)
  228. }
  229. allStgs := make(map[cdssdk.StorageID]stgmod.StorageDetail)
  230. for _, stg := range getStgs.Storages {
  231. allStgs[stg.Storage.StorageID] = *stg
  232. }
  233. downloadStorageMap := make(map[cdssdk.StorageID]*downloadStorageInfo)
  234. for _, id := range obj.PinnedAt {
  235. storage, ok := downloadStorageMap[id]
  236. if !ok {
  237. mod := allStgs[id]
  238. storage = &downloadStorageInfo{
  239. Storage: mod,
  240. ObjectPinned: true,
  241. Distance: t.getStorageDistance(mod),
  242. }
  243. downloadStorageMap[id] = storage
  244. }
  245. storage.ObjectPinned = true
  246. }
  247. for _, b := range obj.Blocks {
  248. storage, ok := downloadStorageMap[b.StorageID]
  249. if !ok {
  250. mod := allStgs[b.StorageID]
  251. storage = &downloadStorageInfo{
  252. Storage: mod,
  253. Distance: t.getStorageDistance(mod),
  254. }
  255. downloadStorageMap[b.StorageID] = storage
  256. }
  257. storage.Blocks = append(storage.Blocks, b)
  258. }
  259. return sort2.Sort(lo.Values(downloadStorageMap), func(left, right *downloadStorageInfo) int {
  260. return sort2.Cmp(left.Distance, right.Distance)
  261. }), nil
  262. }
  263. type downloadBlock struct {
  264. Storage stgmod.StorageDetail
  265. Block stgmod.ObjectBlock
  266. }
  267. func (t *StorageLoadPackage) getMinReadingBlockSolution(sortedStorages []*downloadStorageInfo, k int) (float64, []downloadBlock) {
  268. gotBlocksMap := bitmap.Bitmap64(0)
  269. var gotBlocks []downloadBlock
  270. dist := float64(0.0)
  271. for _, n := range sortedStorages {
  272. for _, b := range n.Blocks {
  273. if !gotBlocksMap.Get(b.Index) {
  274. gotBlocks = append(gotBlocks, downloadBlock{
  275. Storage: n.Storage,
  276. Block: b,
  277. })
  278. gotBlocksMap.Set(b.Index, true)
  279. dist += n.Distance
  280. }
  281. if len(gotBlocks) >= k {
  282. return dist, gotBlocks
  283. }
  284. }
  285. }
  286. return math.MaxFloat64, gotBlocks
  287. }
  288. func (t *StorageLoadPackage) getMinReadingObjectSolution(sortedStorages []*downloadStorageInfo, k int) (float64, *stgmod.StorageDetail) {
  289. dist := math.MaxFloat64
  290. var downloadStg *stgmod.StorageDetail
  291. for _, n := range sortedStorages {
  292. if n.ObjectPinned && float64(k)*n.Distance < dist {
  293. dist = float64(k) * n.Distance
  294. stg := n.Storage
  295. downloadStg = &stg
  296. }
  297. }
  298. return dist, downloadStg
  299. }
  300. func (t *StorageLoadPackage) getStorageDistance(stg stgmod.StorageDetail) float64 {
  301. if stgglb.Local.HubID != nil {
  302. if stg.MasterHub.HubID == *stgglb.Local.HubID {
  303. return consts.StorageDistanceSameStorage
  304. }
  305. }
  306. if stg.MasterHub.LocationID == stgglb.Local.LocationID {
  307. return consts.StorageDistanceSameLocation
  308. }
  309. return consts.StorageDistanceOther
  310. }

本项目旨在将云际存储公共基础设施化,使个人及企业可低门槛使用高效的云际存储服务(安装开箱即用云际存储客户端即可,无需关注其他组件的部署),同时支持用户灵活便捷定制云际存储的功能细节。