worker.go 9.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398
  1. package worker
  2. import (
  3. "bytes"
  4. "errors"
  5. "fmt"
  6. "html/template"
  7. "net/http"
  8. "path/filepath"
  9. "time"
  10. "github.com/gin-gonic/gin"
  11. . "github.com/tuna/tunasync/internal"
  12. )
  13. var tunasyncWorker *Worker
  14. // A Worker is a instance of tunasync worker
  15. type Worker struct {
  16. cfg *Config
  17. providers map[string]mirrorProvider
  18. jobs map[string]*mirrorJob
  19. managerChan chan jobMessage
  20. semaphore chan empty
  21. schedule *scheduleQueue
  22. httpEngine *gin.Engine
  23. httpClient *http.Client
  24. }
  25. // GetTUNASyncWorker returns a singalton worker
  26. func GetTUNASyncWorker(cfg *Config) *Worker {
  27. if tunasyncWorker != nil {
  28. return tunasyncWorker
  29. }
  30. w := &Worker{
  31. cfg: cfg,
  32. providers: make(map[string]mirrorProvider),
  33. jobs: make(map[string]*mirrorJob),
  34. managerChan: make(chan jobMessage, 32),
  35. semaphore: make(chan empty, cfg.Global.Concurrent),
  36. schedule: newScheduleQueue(),
  37. }
  38. if cfg.Manager.CACert != "" {
  39. httpClient, err := CreateHTTPClient(cfg.Manager.CACert)
  40. if err != nil {
  41. logger.Error("Error initializing HTTP client: %s", err.Error())
  42. return nil
  43. }
  44. w.httpClient = httpClient
  45. }
  46. w.initJobs()
  47. w.makeHTTPServer()
  48. tunasyncWorker = w
  49. return w
  50. }
  51. func (w *Worker) initProviders() {
  52. c := w.cfg
  53. formatLogDir := func(logDir string, m mirrorConfig) string {
  54. tmpl, err := template.New("logDirTmpl-" + m.Name).Parse(logDir)
  55. if err != nil {
  56. panic(err)
  57. }
  58. var formatedLogDir bytes.Buffer
  59. tmpl.Execute(&formatedLogDir, m)
  60. return formatedLogDir.String()
  61. }
  62. for _, mirror := range c.Mirrors {
  63. logDir := mirror.LogDir
  64. mirrorDir := mirror.MirrorDir
  65. if logDir == "" {
  66. logDir = c.Global.LogDir
  67. }
  68. if mirrorDir == "" {
  69. mirrorDir = filepath.Join(
  70. c.Global.MirrorDir, mirror.Name,
  71. )
  72. }
  73. if mirror.Interval == 0 {
  74. mirror.Interval = c.Global.Interval
  75. }
  76. logDir = formatLogDir(logDir, mirror)
  77. var provider mirrorProvider
  78. switch mirror.Provider {
  79. case ProvCommand:
  80. pc := cmdConfig{
  81. name: mirror.Name,
  82. upstreamURL: mirror.Upstream,
  83. command: mirror.Command,
  84. workingDir: mirrorDir,
  85. logDir: logDir,
  86. logFile: filepath.Join(logDir, "latest.log"),
  87. interval: time.Duration(mirror.Interval) * time.Minute,
  88. env: mirror.Env,
  89. }
  90. p, err := newCmdProvider(pc)
  91. if err != nil {
  92. panic(err)
  93. }
  94. provider = p
  95. case ProvRsync:
  96. rc := rsyncConfig{
  97. name: mirror.Name,
  98. upstreamURL: mirror.Upstream,
  99. rsyncCmd: mirror.Command,
  100. password: mirror.Password,
  101. excludeFile: mirror.ExcludeFile,
  102. workingDir: mirrorDir,
  103. logDir: logDir,
  104. logFile: filepath.Join(logDir, "latest.log"),
  105. useIPv6: mirror.UseIPv6,
  106. interval: time.Duration(mirror.Interval) * time.Minute,
  107. }
  108. p, err := newRsyncProvider(rc)
  109. if err != nil {
  110. panic(err)
  111. }
  112. provider = p
  113. case ProvTwoStageRsync:
  114. rc := twoStageRsyncConfig{
  115. name: mirror.Name,
  116. stage1Profile: mirror.Stage1Profile,
  117. upstreamURL: mirror.Upstream,
  118. rsyncCmd: mirror.Command,
  119. password: mirror.Password,
  120. excludeFile: mirror.ExcludeFile,
  121. workingDir: mirrorDir,
  122. logDir: logDir,
  123. logFile: filepath.Join(logDir, "latest.log"),
  124. useIPv6: mirror.UseIPv6,
  125. interval: time.Duration(mirror.Interval) * time.Minute,
  126. }
  127. p, err := newTwoStageRsyncProvider(rc)
  128. if err != nil {
  129. panic(err)
  130. }
  131. provider = p
  132. default:
  133. panic(errors.New("Invalid mirror provider"))
  134. }
  135. provider.AddHook(newLogLimiter(provider))
  136. w.providers[provider.Name()] = provider
  137. }
  138. }
  139. func (w *Worker) initJobs() {
  140. w.initProviders()
  141. for name, provider := range w.providers {
  142. w.jobs[name] = newMirrorJob(provider)
  143. }
  144. }
  145. // Ctrl server receives commands from the manager
  146. func (w *Worker) makeHTTPServer() {
  147. s := gin.New()
  148. s.Use(gin.Recovery())
  149. s.POST("/", func(c *gin.Context) {
  150. var cmd WorkerCmd
  151. if err := c.BindJSON(&cmd); err != nil {
  152. c.JSON(http.StatusBadRequest, gin.H{"msg": "Invalid request"})
  153. return
  154. }
  155. job, ok := w.jobs[cmd.MirrorID]
  156. if !ok {
  157. c.JSON(http.StatusNotFound, gin.H{"msg": fmt.Sprintf("Mirror ``%s'' not found", cmd.MirrorID)})
  158. return
  159. }
  160. logger.Info("Received command: %v", cmd)
  161. // if job disabled, start them first
  162. switch cmd.Cmd {
  163. case CmdStart, CmdRestart:
  164. if job.State() == stateDisabled {
  165. go job.Run(w.managerChan, w.semaphore)
  166. }
  167. }
  168. switch cmd.Cmd {
  169. case CmdStart:
  170. job.ctrlChan <- jobStart
  171. case CmdRestart:
  172. job.ctrlChan <- jobRestart
  173. case CmdStop:
  174. // if job is disabled, no goroutine would be there
  175. // receiving this signal
  176. if job.State() != stateDisabled {
  177. w.schedule.Remove(job.Name())
  178. job.ctrlChan <- jobStop
  179. }
  180. case CmdDisable:
  181. if job.State() != stateDisabled {
  182. w.schedule.Remove(job.Name())
  183. job.ctrlChan <- jobDisable
  184. <-job.disabled
  185. }
  186. case CmdPing:
  187. job.ctrlChan <- jobStart
  188. default:
  189. c.JSON(http.StatusNotAcceptable, gin.H{"msg": "Invalid Command"})
  190. return
  191. }
  192. c.JSON(http.StatusOK, gin.H{"msg": "OK"})
  193. })
  194. w.httpEngine = s
  195. }
  196. func (w *Worker) runHTTPServer() {
  197. addr := fmt.Sprintf("%s:%d", w.cfg.Server.Addr, w.cfg.Server.Port)
  198. httpServer := &http.Server{
  199. Addr: addr,
  200. Handler: w.httpEngine,
  201. ReadTimeout: 10 * time.Second,
  202. WriteTimeout: 10 * time.Second,
  203. }
  204. if w.cfg.Server.SSLCert == "" && w.cfg.Server.SSLKey == "" {
  205. if err := httpServer.ListenAndServe(); err != nil {
  206. panic(err)
  207. }
  208. } else {
  209. if err := httpServer.ListenAndServeTLS(w.cfg.Server.SSLCert, w.cfg.Server.SSLKey); err != nil {
  210. panic(err)
  211. }
  212. }
  213. }
  214. // Run runs worker forever
  215. func (w *Worker) Run() {
  216. w.registorWorker()
  217. go w.runHTTPServer()
  218. w.runSchedule()
  219. }
  220. func (w *Worker) runSchedule() {
  221. mirrorList := w.fetchJobStatus()
  222. unset := make(map[string]bool)
  223. for name := range w.jobs {
  224. unset[name] = true
  225. }
  226. // Fetch mirror list stored in the manager
  227. // put it on the scheduled time
  228. // if it's disabled, ignore it
  229. for _, m := range mirrorList {
  230. if job, ok := w.jobs[m.Name]; ok {
  231. delete(unset, m.Name)
  232. switch m.Status {
  233. case Disabled:
  234. job.SetState(stateDisabled)
  235. continue
  236. case Paused:
  237. job.SetState(statePaused)
  238. go job.Run(w.managerChan, w.semaphore)
  239. continue
  240. default:
  241. job.SetState(stateReady)
  242. go job.Run(w.managerChan, w.semaphore)
  243. stime := m.LastUpdate.Add(job.provider.Interval())
  244. logger.Debug("Scheduling job %s @%s", job.Name(), stime.Format("2006-01-02 15:04:05"))
  245. w.schedule.AddJob(stime, job)
  246. }
  247. }
  248. }
  249. // some new jobs may be added
  250. // which does not exist in the
  251. // manager's mirror list
  252. for name := range unset {
  253. job := w.jobs[name]
  254. job.SetState(stateReady)
  255. go job.Run(w.managerChan, w.semaphore)
  256. w.schedule.AddJob(time.Now(), job)
  257. }
  258. for {
  259. select {
  260. case jobMsg := <-w.managerChan:
  261. // got status update from job
  262. job := w.jobs[jobMsg.name]
  263. if job.State() != stateReady {
  264. logger.Info("Job %s state is not ready, skip adding new schedule", jobMsg.name)
  265. continue
  266. }
  267. // syncing status is only meaningful when job
  268. // is running. If it's paused or disabled
  269. // a sync failure signal would be emitted
  270. // which needs to be ignored
  271. w.updateStatus(jobMsg)
  272. // only successful or the final failure msg
  273. // can trigger scheduling
  274. if jobMsg.schedule {
  275. schedTime := time.Now().Add(job.provider.Interval())
  276. logger.Info(
  277. "Next scheduled time for %s: %s",
  278. job.Name(),
  279. schedTime.Format("2006-01-02 15:04:05"),
  280. )
  281. w.schedule.AddJob(schedTime, job)
  282. }
  283. case <-time.Tick(5 * time.Second):
  284. // check schedule every 5 seconds
  285. if job := w.schedule.Pop(); job != nil {
  286. job.ctrlChan <- jobStart
  287. }
  288. }
  289. }
  290. }
  291. // Name returns worker name
  292. func (w *Worker) Name() string {
  293. return w.cfg.Global.Name
  294. }
  295. // URL returns the url to http server of the worker
  296. func (w *Worker) URL() string {
  297. proto := "https"
  298. if w.cfg.Server.SSLCert == "" && w.cfg.Server.SSLKey == "" {
  299. proto = "http"
  300. }
  301. return fmt.Sprintf("%s://%s:%d/", proto, w.cfg.Server.Hostname, w.cfg.Server.Port)
  302. }
  303. func (w *Worker) registorWorker() {
  304. url := fmt.Sprintf(
  305. "%s/workers",
  306. w.cfg.Manager.APIBase,
  307. )
  308. msg := WorkerStatus{
  309. ID: w.Name(),
  310. URL: w.URL(),
  311. }
  312. if _, err := PostJSON(url, msg, w.httpClient); err != nil {
  313. logger.Error("Failed to register worker")
  314. }
  315. }
  316. func (w *Worker) updateStatus(jobMsg jobMessage) {
  317. url := fmt.Sprintf(
  318. "%s/workers/%s/jobs/%s",
  319. w.cfg.Manager.APIBase,
  320. w.Name(),
  321. jobMsg.name,
  322. )
  323. p := w.providers[jobMsg.name]
  324. smsg := MirrorStatus{
  325. Name: jobMsg.name,
  326. Worker: w.cfg.Global.Name,
  327. IsMaster: true,
  328. Status: jobMsg.status,
  329. Upstream: p.Upstream(),
  330. Size: "unknown",
  331. ErrorMsg: jobMsg.msg,
  332. }
  333. if _, err := PostJSON(url, smsg, w.httpClient); err != nil {
  334. logger.Error("Failed to update mirror(%s) status: %s", jobMsg.name, err.Error())
  335. }
  336. }
  337. func (w *Worker) fetchJobStatus() []MirrorStatus {
  338. var mirrorList []MirrorStatus
  339. url := fmt.Sprintf(
  340. "%s/workers/%s/jobs",
  341. w.cfg.Manager.APIBase,
  342. w.Name(),
  343. )
  344. if _, err := GetJSON(url, &mirrorList, w.httpClient); err != nil {
  345. logger.Error("Failed to fetch job status: %s", err.Error())
  346. }
  347. return mirrorList
  348. }