sink_node.go 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462
  1. package nodes
  2. import (
  3. "bytes"
  4. "encoding/json"
  5. "fmt"
  6. "github.com/emqx/kuiper/common"
  7. "github.com/emqx/kuiper/common/templates"
  8. "github.com/emqx/kuiper/plugins"
  9. "github.com/emqx/kuiper/xstream/api"
  10. "github.com/emqx/kuiper/xstream/sinks"
  11. "sync"
  12. "text/template"
  13. "time"
  14. )
  15. type SinkNode struct {
  16. *defaultSinkNode
  17. //static
  18. sinkType string
  19. mutex sync.RWMutex
  20. //configs (also static for sinks)
  21. options map[string]interface{}
  22. isMock bool
  23. //states varies after restart
  24. sinks []api.Sink
  25. tch chan struct{} //channel to trigger cache saved, will be trigger by checkpoint only
  26. }
  27. func NewSinkNode(name string, sinkType string, props map[string]interface{}) *SinkNode {
  28. bufferLength := 1024
  29. if c, ok := props["bufferLength"]; ok {
  30. if t, err := common.ToInt(c); err != nil || t <= 0 {
  31. //invalid property bufferLength
  32. } else {
  33. bufferLength = t
  34. }
  35. }
  36. return &SinkNode{
  37. defaultSinkNode: &defaultSinkNode{
  38. input: make(chan interface{}, bufferLength),
  39. defaultNode: &defaultNode{
  40. name: name,
  41. concurrency: 1,
  42. ctx: nil,
  43. },
  44. },
  45. sinkType: sinkType,
  46. options: props,
  47. }
  48. }
  49. //Only for mock source, do not use it in production
  50. func NewSinkNodeWithSink(name string, sink api.Sink, props map[string]interface{}) *SinkNode {
  51. return &SinkNode{
  52. defaultSinkNode: &defaultSinkNode{
  53. input: make(chan interface{}, 1024),
  54. defaultNode: &defaultNode{
  55. name: name,
  56. concurrency: 1,
  57. ctx: nil,
  58. },
  59. },
  60. sinks: []api.Sink{sink},
  61. options: props,
  62. isMock: true,
  63. }
  64. }
  65. func (m *SinkNode) Open(ctx api.StreamContext, result chan<- error) {
  66. m.ctx = ctx
  67. logger := ctx.GetLogger()
  68. logger.Debugf("open sink node %s", m.name)
  69. if m.qos >= api.AtLeastOnce {
  70. m.tch = make(chan struct{})
  71. }
  72. go func() {
  73. if c, ok := m.options["concurrency"]; ok {
  74. if t, err := common.ToInt(c); err != nil || t <= 0 {
  75. logger.Warnf("invalid type for concurrency property, should be positive integer but found %t", c)
  76. } else {
  77. m.concurrency = t
  78. }
  79. }
  80. runAsync := false
  81. if c, ok := m.options["runAsync"]; ok {
  82. if t, ok := c.(bool); !ok {
  83. logger.Warnf("invalid type for runAsync property, should be bool but found %t", c)
  84. } else {
  85. runAsync = t
  86. }
  87. }
  88. retryInterval := 1000
  89. if c, ok := m.options["retryInterval"]; ok {
  90. if t, err := common.ToInt(c); err != nil || t < 0 {
  91. logger.Warnf("invalid type for retryInterval property, should be positive integer but found %t", c)
  92. } else {
  93. retryInterval = t
  94. }
  95. }
  96. retryCount := 3
  97. if c, ok := m.options["retryCount"]; ok {
  98. if t, err := common.ToInt(c); err != nil || t < 0 {
  99. logger.Warnf("invalid type for retryCount property, should be positive integer but found %t", c)
  100. } else {
  101. retryCount = t
  102. }
  103. }
  104. cacheLength := 1024
  105. if c, ok := m.options["cacheLength"]; ok {
  106. if t, err := common.ToInt(c); err != nil || t < 0 {
  107. logger.Warnf("invalid type for cacheLength property, should be positive integer but found %t", c)
  108. } else {
  109. cacheLength = t
  110. }
  111. }
  112. cacheSaveInterval := 1000
  113. if c, ok := m.options["cacheSaveInterval"]; ok {
  114. if t, err := common.ToInt(c); err != nil || t < 0 {
  115. logger.Warnf("invalid type for cacheSaveInterval property, should be positive integer but found %t", c)
  116. } else {
  117. cacheSaveInterval = t
  118. }
  119. }
  120. omitIfEmpty := false
  121. if c, ok := m.options["omitIfEmpty"]; ok {
  122. if t, ok := c.(bool); !ok {
  123. logger.Warnf("invalid type for omitIfEmpty property, should be a bool value 'true/false'.", c)
  124. } else {
  125. omitIfEmpty = t
  126. }
  127. }
  128. sendSingle := false
  129. if c, ok := m.options["sendSingle"]; ok {
  130. if t, ok := c.(bool); !ok {
  131. logger.Warnf("invalid type for sendSingle property, should be a bool value 'true/false'.", c)
  132. } else {
  133. sendSingle = t
  134. }
  135. }
  136. var tp *template.Template = nil
  137. if c, ok := m.options["dataTemplate"]; ok {
  138. if t, ok := c.(string); !ok {
  139. logger.Warnf("invalid type for dateTemplate property, should be a string value.", c)
  140. } else {
  141. funcMap := template.FuncMap{
  142. "json": templates.JsonMarshal,
  143. "base64": templates.Base64Encode,
  144. "add": templates.Add,
  145. }
  146. temp, err := template.New("sink").Funcs(funcMap).Parse(t)
  147. if err != nil {
  148. msg := fmt.Sprintf("property dataTemplate %v is invalid: %v", t, err)
  149. result <- fmt.Errorf(msg)
  150. logger.Warnf(msg)
  151. return
  152. } else {
  153. tp = temp
  154. }
  155. }
  156. }
  157. m.reset()
  158. logger.Infof("open sink node %d instances", m.concurrency)
  159. for i := 0; i < m.concurrency; i++ { // workers
  160. go func(instance int) {
  161. var sink api.Sink
  162. var err error
  163. if !m.isMock {
  164. logger.Debugf("Trying to get sink for rule %s with options %v\n", ctx.GetRuleId(), m.options)
  165. sink, err = getSink(m.sinkType, m.options)
  166. if err != nil {
  167. m.drainError(result, err, ctx, logger)
  168. return
  169. }
  170. logger.Debugf("Successfully get the sink %s", m.sinkType)
  171. m.mutex.Lock()
  172. m.sinks = append(m.sinks, sink)
  173. m.mutex.Unlock()
  174. logger.Debugf("Now is to open sink for rule %s.\n", ctx.GetRuleId())
  175. if err := sink.Open(ctx); err != nil {
  176. m.drainError(result, err, ctx, logger)
  177. return
  178. }
  179. logger.Debugf("Successfully open sink for rule %s.\n", ctx.GetRuleId())
  180. } else {
  181. sink = m.sinks[instance]
  182. }
  183. stats, err := NewStatManager("sink", ctx)
  184. if err != nil {
  185. m.drainError(result, err, ctx, logger)
  186. return
  187. }
  188. m.mutex.Lock()
  189. m.statManagers = append(m.statManagers, stats)
  190. m.mutex.Unlock()
  191. if common.Config.Sink.DisableCache {
  192. for {
  193. select {
  194. case data := <-m.input:
  195. if newdata, processed := m.preprocess(data); processed {
  196. break
  197. } else {
  198. data = newdata
  199. }
  200. stats.SetBufferLength(int64(len(m.input)))
  201. if runAsync {
  202. go doCollect(sink, data, stats, omitIfEmpty, sendSingle, tp, ctx)
  203. } else {
  204. doCollect(sink, data, stats, omitIfEmpty, sendSingle, tp, ctx)
  205. }
  206. case <-ctx.Done():
  207. logger.Infof("sink node %s instance %d done", m.name, instance)
  208. if err := sink.Close(ctx); err != nil {
  209. logger.Warnf("close sink node %s instance %d fails: %v", m.name, instance, err)
  210. }
  211. return
  212. case <-m.tch:
  213. logger.Debugf("rule %s sink receive checkpoint, do nothing", ctx.GetRuleId())
  214. }
  215. }
  216. } else {
  217. logger.Infof("Creating sink cache")
  218. var cache *Cache
  219. if m.qos >= api.AtLeastOnce {
  220. cache = NewCheckpointbasedCache(m.input, cacheLength, m.tch, result, ctx)
  221. } else {
  222. cache = NewTimebasedCache(m.input, cacheLength, cacheSaveInterval, result, ctx)
  223. }
  224. for {
  225. select {
  226. case data := <-cache.Out:
  227. if newdata, processed := m.preprocess(data.data); processed {
  228. break
  229. } else {
  230. data.data = newdata
  231. }
  232. stats.SetBufferLength(int64(len(m.input)))
  233. if runAsync {
  234. go doCollectCacheTuple(sink, data, stats, retryInterval, retryCount, omitIfEmpty, sendSingle, tp, cache.Complete, ctx)
  235. } else {
  236. doCollectCacheTuple(sink, data, stats, retryInterval, retryCount, omitIfEmpty, sendSingle, tp, cache.Complete, ctx)
  237. }
  238. case <-ctx.Done():
  239. logger.Infof("sink node %s instance %d done", m.name, instance)
  240. if err := sink.Close(ctx); err != nil {
  241. logger.Warnf("close sink node %s instance %d fails: %v", m.name, instance, err)
  242. }
  243. return
  244. }
  245. }
  246. }
  247. }(i)
  248. }
  249. }()
  250. }
  251. func (m *SinkNode) reset() {
  252. if !m.isMock {
  253. m.sinks = nil
  254. }
  255. m.statManagers = nil
  256. }
  257. func extractInput(v []byte) ([]map[string]interface{}, error) {
  258. var j []map[string]interface{}
  259. if err := json.Unmarshal(v, &j); err != nil {
  260. return nil, fmt.Errorf("fail to decode the input %s as json: %v", v, err)
  261. }
  262. return j, nil
  263. }
  264. func doCollect(sink api.Sink, item interface{}, stats StatManager, omitIfEmpty bool, sendSingle bool, tp *template.Template, ctx api.StreamContext) {
  265. stats.IncTotalRecordsIn()
  266. stats.ProcessTimeStart()
  267. defer stats.ProcessTimeEnd()
  268. logger := ctx.GetLogger()
  269. outdatas := getOutData(stats, ctx, item, omitIfEmpty, sendSingle, tp)
  270. for _, outdata := range outdatas {
  271. if err := sink.Collect(ctx, outdata); err != nil {
  272. stats.IncTotalExceptions()
  273. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), outdata, err)
  274. } else {
  275. stats.IncTotalRecordsOut()
  276. }
  277. }
  278. }
  279. func getOutData(stats StatManager, ctx api.StreamContext, item interface{}, omitIfEmpty bool, sendSingle bool, tp *template.Template) [][]byte {
  280. logger := ctx.GetLogger()
  281. var outdatas [][]byte
  282. switch val := item.(type) {
  283. case []byte:
  284. if omitIfEmpty && string(val) == "[{}]" {
  285. return nil
  286. }
  287. var (
  288. err error
  289. j []map[string]interface{}
  290. )
  291. if sendSingle || tp != nil {
  292. j, err = extractInput(val)
  293. if err != nil {
  294. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  295. stats.IncTotalExceptions()
  296. return nil
  297. }
  298. logger.Debugf("receive %d records", len(j))
  299. }
  300. if !sendSingle {
  301. if tp != nil {
  302. var output bytes.Buffer
  303. err := tp.Execute(&output, j)
  304. if err != nil {
  305. logger.Warnf("sink node %s instance %d publish %s decode template error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  306. stats.IncTotalExceptions()
  307. return nil
  308. }
  309. outdatas = append(outdatas, output.Bytes())
  310. } else {
  311. outdatas = [][]byte{val}
  312. }
  313. } else {
  314. for _, r := range j {
  315. if tp != nil {
  316. var output bytes.Buffer
  317. err := tp.Execute(&output, r)
  318. if err != nil {
  319. logger.Warnf("sink node %s instance %d publish %s decode template error: %v", ctx.GetOpId(), ctx.GetInstanceId(), val, err)
  320. stats.IncTotalExceptions()
  321. return nil
  322. }
  323. outdatas = append(outdatas, output.Bytes())
  324. } else {
  325. if ot, e := json.Marshal(r); e != nil {
  326. logger.Warnf("sink node %s instance %d publish %s marshal error: %v", ctx.GetOpId(), ctx.GetInstanceId(), r, e)
  327. stats.IncTotalExceptions()
  328. return nil
  329. } else {
  330. outdatas = append(outdatas, ot)
  331. }
  332. }
  333. }
  334. }
  335. case error:
  336. outdatas = [][]byte{[]byte(fmt.Sprintf(`[{"error":"%s"}]`, val.Error()))}
  337. default:
  338. outdatas = [][]byte{[]byte(fmt.Sprintf(`[{"error":"result is not a string but found %#v"}]`, val))}
  339. }
  340. return outdatas
  341. }
  342. func doCollectCacheTuple(sink api.Sink, item *CacheTuple, stats StatManager, retryInterval, retryCount int, omitIfEmpty bool, sendSingle bool, tp *template.Template, signalCh chan<- int, ctx api.StreamContext) {
  343. stats.IncTotalRecordsIn()
  344. stats.ProcessTimeStart()
  345. defer stats.ProcessTimeEnd()
  346. logger := ctx.GetLogger()
  347. outdatas := getOutData(stats, ctx, item.data, omitIfEmpty, sendSingle, tp)
  348. for _, outdata := range outdatas {
  349. outerloop:
  350. for {
  351. select {
  352. case <-ctx.Done():
  353. logger.Infof("sink node %s instance %d stops data resending", ctx.GetOpId(), ctx.GetInstanceId())
  354. return
  355. default:
  356. if err := sink.Collect(ctx, outdata); err != nil {
  357. stats.IncTotalExceptions()
  358. logger.Warnf("sink node %s instance %d publish %s error: %v", ctx.GetOpId(), ctx.GetInstanceId(), outdata, err)
  359. if retryInterval > 0 && retryCount > 0 {
  360. retryCount--
  361. time.Sleep(time.Duration(retryInterval) * time.Millisecond)
  362. logger.Debugf("try again")
  363. } else {
  364. break outerloop
  365. }
  366. } else {
  367. logger.Debugf("success")
  368. stats.IncTotalRecordsOut()
  369. select {
  370. case signalCh <- item.index:
  371. default:
  372. logger.Warnf("sink cache missing response for %d", item.index)
  373. }
  374. break outerloop
  375. }
  376. }
  377. }
  378. }
  379. }
  380. func doGetSink(name string, action map[string]interface{}) (api.Sink, error) {
  381. var (
  382. s api.Sink
  383. err error
  384. )
  385. switch name {
  386. case "log":
  387. s = sinks.NewLogSink()
  388. case "logToMemory":
  389. s = sinks.NewLogSinkToMemory()
  390. case "mqtt":
  391. s = &sinks.MQTTSink{}
  392. case "rest":
  393. s = &sinks.RestSink{}
  394. case "nop":
  395. s = &sinks.NopSink{}
  396. default:
  397. s, err = plugins.GetSink(name)
  398. if err != nil {
  399. return nil, err
  400. }
  401. }
  402. err = s.Configure(action)
  403. if err != nil {
  404. return nil, err
  405. }
  406. return s, nil
  407. }
  408. //Override defaultNode
  409. func (m *SinkNode) AddOutput(_ chan<- interface{}, name string) error {
  410. return fmt.Errorf("fail to add output %s, sink %s cannot add output", name, m.name)
  411. }
  412. //Override defaultNode
  413. func (m *SinkNode) Broadcast(_ interface{}) error {
  414. return fmt.Errorf("sink %s cannot add broadcast", m.name)
  415. }
  416. func (m *SinkNode) drainError(errCh chan<- error, err error, ctx api.StreamContext, logger api.Logger) {
  417. go func() {
  418. select {
  419. case errCh <- err:
  420. ctx.GetLogger().Errorf("error in sink %s", err)
  421. case <-ctx.Done():
  422. m.close(ctx, logger)
  423. }
  424. }()
  425. }
  426. func (m *SinkNode) close(ctx api.StreamContext, logger api.Logger) {
  427. for _, s := range m.sinks {
  428. if err := s.Close(ctx); err != nil {
  429. logger.Warnf("close sink fails: %v", err)
  430. }
  431. }
  432. if m.tch != nil {
  433. close(m.tch)
  434. m.tch = nil
  435. }
  436. }
  437. // Only called when checkpoint enabled
  438. func (m *SinkNode) SaveCache() {
  439. m.tch <- struct{}{}
  440. }