src/pools/utils.ts

   1 import cluster, { Worker as ClusterWorker } from 'node:cluster'
   2 import { existsSync } from 'node:fs'
   3 import { env } from 'node:process'
   4 import {
   5   SHARE_ENV,
   6   Worker as ThreadWorker,
   7   type WorkerOptions
   8 } from 'node:worker_threads'
   9
  10 import type { MessageValue, Task } from '../utility-types.js'
  11 import { average, isPlainObject, max, median, min } from '../utils.js'
  12 import type { TasksQueueOptions } from './pool.js'
  13 import {
  14   type MeasurementStatisticsRequirements,
  15   WorkerChoiceStrategies,
  16   type WorkerChoiceStrategy
  17 } from './selection-strategies/selection-strategies-types.js'
  18 import type { WorkerChoiceStrategiesContext } from './selection-strategies/worker-choice-strategies-context.js'
  19 import {
  20   type IWorker,
  21   type IWorkerNode,
  22   type MeasurementStatistics,
  23   type WorkerNodeOptions,
  24   type WorkerType,
  25   WorkerTypes,
  26   type WorkerUsage
  27 } from './worker.js'
  28
  29 /**
  30  * Default measurement statistics requirements.
  31  */
  32 export const DEFAULT_MEASUREMENT_STATISTICS_REQUIREMENTS: MeasurementStatisticsRequirements =
  33   {
  34     aggregate: false,
  35     average: false,
  36     median: false
  37   }
  38
  39 export const getDefaultTasksQueueOptions = (
  40   poolMaxSize: number
  41 ): Required<TasksQueueOptions> => {
  42   return {
  43     size: Math.pow(poolMaxSize, 2),
  44     concurrency: 1,
  45     taskStealing: true,
  46     tasksStealingOnBackPressure: false,
  47     tasksFinishedTimeout: 2000
  48   }
  49 }
  50
  51 export const checkFilePath = (filePath: string | undefined): void => {
  52   if (filePath == null) {
  53     throw new TypeError('The worker file path must be specified')
  54   }
  55   if (typeof filePath !== 'string') {
  56     throw new TypeError('The worker file path must be a string')
  57   }
  58   if (!existsSync(filePath)) {
  59     throw new Error(`Cannot find the worker file '${filePath}'`)
  60   }
  61 }
  62
  63 export const checkDynamicPoolSize = (
  64   min: number,
  65   max: number | undefined
  66 ): void => {
  67   if (max == null) {
  68     throw new TypeError(
  69       'Cannot instantiate a dynamic pool without specifying the maximum pool size'
  70     )
  71   } else if (!Number.isSafeInteger(max)) {
  72     throw new TypeError(
  73       'Cannot instantiate a dynamic pool with a non safe integer maximum pool size'
  74     )
  75   } else if (min > max) {
  76     throw new RangeError(
  77       'Cannot instantiate a dynamic pool with a maximum pool size inferior to the minimum pool size'
  78     )
  79   } else if (max === 0) {
  80     throw new RangeError(
  81       'Cannot instantiate a dynamic pool with a maximum pool size equal to zero'
  82     )
  83   } else if (min === max) {
  84     throw new RangeError(
  85       'Cannot instantiate a dynamic pool with a minimum pool size equal to the maximum pool size. Use a fixed pool instead'
  86     )
  87   }
  88 }
  89
  90 export const checkValidPriority = (priority: number | undefined): void => {
  91   if (priority != null && !Number.isSafeInteger(priority)) {
  92     throw new TypeError(`Invalid property 'priority': '${priority}'`)
  93   }
  94   if (
  95     priority != null &&
  96     Number.isSafeInteger(priority) &&
  97     (priority < -20 || priority > 19)
  98   ) {
  99     throw new RangeError("Property 'priority' must be between -20 and 19")
 100   }
 101 }
 102
 103 export const checkValidWorkerChoiceStrategy = (
 104   workerChoiceStrategy: WorkerChoiceStrategy | undefined
 105 ): void => {
 106   if (
 107     workerChoiceStrategy != null &&
 108     !Object.values(WorkerChoiceStrategies).includes(workerChoiceStrategy)
 109   ) {
 110     throw new Error(`Invalid worker choice strategy '${workerChoiceStrategy}'`)
 111   }
 112 }
 113
 114 export const checkValidTasksQueueOptions = (
 115   tasksQueueOptions: TasksQueueOptions | undefined
 116 ): void => {
 117   if (tasksQueueOptions != null && !isPlainObject(tasksQueueOptions)) {
 118     throw new TypeError('Invalid tasks queue options: must be a plain object')
 119   }
 120   if (
 121     tasksQueueOptions?.concurrency != null &&
 122     !Number.isSafeInteger(tasksQueueOptions.concurrency)
 123   ) {
 124     throw new TypeError(
 125       'Invalid worker node tasks concurrency: must be an integer'
 126     )
 127   }
 128   if (
 129     tasksQueueOptions?.concurrency != null &&
 130     tasksQueueOptions.concurrency <= 0
 131   ) {
 132     throw new RangeError(
 133       `Invalid worker node tasks concurrency: ${tasksQueueOptions.concurrency} is a negative integer or zero`
 134     )
 135   }
 136   if (
 137     tasksQueueOptions?.size != null &&
 138     !Number.isSafeInteger(tasksQueueOptions.size)
 139   ) {
 140     throw new TypeError(
 141       'Invalid worker node tasks queue size: must be an integer'
 142     )
 143   }
 144   if (tasksQueueOptions?.size != null && tasksQueueOptions.size <= 0) {
 145     throw new RangeError(
 146       `Invalid worker node tasks queue size: ${tasksQueueOptions.size} is a negative integer or zero`
 147     )
 148   }
 149 }
 150
 151 export const checkWorkerNodeArguments = (
 152   type: WorkerType | undefined,
 153   filePath: string | undefined,
 154   opts: WorkerNodeOptions | undefined
 155 ): void => {
 156   if (type == null) {
 157     throw new TypeError('Cannot construct a worker node without a worker type')
 158   }
 159   if (!Object.values(WorkerTypes).includes(type)) {
 160     throw new TypeError(
 161       `Cannot construct a worker node with an invalid worker type '${type}'`
 162     )
 163   }
 164   checkFilePath(filePath)
 165   if (opts == null) {
 166     throw new TypeError(
 167       'Cannot construct a worker node without worker node options'
 168     )
 169   }
 170   if (!isPlainObject(opts)) {
 171     throw new TypeError(
 172       'Cannot construct a worker node with invalid worker node options: must be a plain object'
 173     )
 174   }
 175   if (opts.tasksQueueBackPressureSize == null) {
 176     throw new TypeError(
 177       'Cannot construct a worker node without a tasks queue back pressure size option'
 178     )
 179   }
 180   if (!Number.isSafeInteger(opts.tasksQueueBackPressureSize)) {
 181     throw new TypeError(
 182       'Cannot construct a worker node with a tasks queue back pressure size option that is not an integer'
 183     )
 184   }
 185   if (opts.tasksQueueBackPressureSize <= 0) {
 186     throw new RangeError(
 187       'Cannot construct a worker node with a tasks queue back pressure size option that is not a positive integer'
 188     )
 189   }
 190   if (opts.tasksQueueBucketSize == null) {
 191     throw new TypeError(
 192       'Cannot construct a worker node without a tasks queue bucket size option'
 193     )
 194   }
 195   if (!Number.isSafeInteger(opts.tasksQueueBucketSize)) {
 196     throw new TypeError(
 197       'Cannot construct a worker node with a tasks queue bucket size option that is not an integer'
 198     )
 199   }
 200   if (opts.tasksQueueBucketSize <= 0) {
 201     throw new RangeError(
 202       'Cannot construct a worker node with a tasks queue bucket size option that is not a positive integer'
 203     )
 204   }
 205   if (opts.tasksQueuePriority == null) {
 206     throw new TypeError(
 207       'Cannot construct a worker node without a tasks queue priority option'
 208     )
 209   }
 210   if (typeof opts.tasksQueuePriority !== 'boolean') {
 211     throw new TypeError(
 212       'Cannot construct a worker node with a tasks queue priority option that is not a boolean'
 213     )
 214   }
 215 }
 216
 217 /**
 218  * Updates the given measurement statistics.
 219  *
 220  * @param measurementStatistics - The measurement statistics to update.
 221  * @param measurementRequirements - The measurement statistics requirements.
 222  * @param measurementValue - The measurement value.
 223  * @internal
 224  */
 225 const updateMeasurementStatistics = (
 226   measurementStatistics: MeasurementStatistics,
 227   measurementRequirements: MeasurementStatisticsRequirements | undefined,
 228   measurementValue: number | undefined
 229 ): void => {
 230   if (
 231     measurementRequirements != null &&
 232     measurementValue != null &&
 233     measurementRequirements.aggregate
 234   ) {
 235     measurementStatistics.aggregate =
 236       (measurementStatistics.aggregate ?? 0) + measurementValue
 237     measurementStatistics.minimum = min(
 238       measurementValue,
 239       measurementStatistics.minimum ?? Number.POSITIVE_INFINITY
 240     )
 241     measurementStatistics.maximum = max(
 242       measurementValue,
 243       measurementStatistics.maximum ?? Number.NEGATIVE_INFINITY
 244     )
 245     if (measurementRequirements.average || measurementRequirements.median) {
 246       measurementStatistics.history.put(measurementValue)
 247       if (measurementRequirements.average) {
 248         measurementStatistics.average = average(
 249           measurementStatistics.history.toArray()
 250         )
 251       } else if (measurementStatistics.average != null) {
 252         delete measurementStatistics.average
 253       }
 254       if (measurementRequirements.median) {
 255         measurementStatistics.median = median(
 256           measurementStatistics.history.toArray()
 257         )
 258       } else if (measurementStatistics.median != null) {
 259         delete measurementStatistics.median
 260       }
 261     }
 262   }
 263 }
 264 if (env.NODE_ENV === 'test') {
 265   // eslint-disable-next-line @typescript-eslint/no-unsafe-member-access
 266   exports.updateMeasurementStatistics = updateMeasurementStatistics
 267 }
 268
 269 export const updateWaitTimeWorkerUsage = <
 270   Worker extends IWorker,
 271   Data = unknown,
 272   Response = unknown
 273 >(
 274     workerChoiceStrategiesContext:
 275     | WorkerChoiceStrategiesContext<Worker, Data, Response>
 276     | undefined,
 277     workerUsage: WorkerUsage,
 278     task: Task<Data>
 279   ): void => {
 280   const timestamp = performance.now()
 281   const taskWaitTime = timestamp - (task.timestamp ?? timestamp)
 282   updateMeasurementStatistics(
 283     workerUsage.waitTime,
 284     workerChoiceStrategiesContext?.getTaskStatisticsRequirements().waitTime,
 285     taskWaitTime
 286   )
 287 }
 288
 289 export const updateTaskStatisticsWorkerUsage = <Response = unknown>(
 290   workerUsage: WorkerUsage,
 291   message: MessageValue<Response>
 292 ): void => {
 293   const workerTaskStatistics = workerUsage.tasks
 294   if (
 295     // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
 296     workerTaskStatistics.executing != null &&
 297     workerTaskStatistics.executing > 0
 298   ) {
 299     --workerTaskStatistics.executing
 300   }
 301   if (message.workerError == null) {
 302     ++workerTaskStatistics.executed
 303   } else {
 304     ++workerTaskStatistics.failed
 305   }
 306 }
 307
 308 export const updateRunTimeWorkerUsage = <
 309   Worker extends IWorker,
 310   Data = unknown,
 311   Response = unknown
 312 >(
 313     workerChoiceStrategiesContext:
 314     | WorkerChoiceStrategiesContext<Worker, Data, Response>
 315     | undefined,
 316     workerUsage: WorkerUsage,
 317     message: MessageValue<Response>
 318   ): void => {
 319   if (message.workerError != null) {
 320     return
 321   }
 322   updateMeasurementStatistics(
 323     workerUsage.runTime,
 324     workerChoiceStrategiesContext?.getTaskStatisticsRequirements().runTime,
 325     message.taskPerformance?.runTime ?? 0
 326   )
 327 }
 328
 329 export const updateEluWorkerUsage = <
 330   Worker extends IWorker,
 331   Data = unknown,
 332   Response = unknown
 333 >(
 334     workerChoiceStrategiesContext:
 335     | WorkerChoiceStrategiesContext<Worker, Data, Response>
 336     | undefined,
 337     workerUsage: WorkerUsage,
 338     message: MessageValue<Response>
 339   ): void => {
 340   if (message.workerError != null) {
 341     return
 342   }
 343   const eluTaskStatisticsRequirements =
 344     workerChoiceStrategiesContext?.getTaskStatisticsRequirements().elu
 345   updateMeasurementStatistics(
 346     workerUsage.elu.active,
 347     eluTaskStatisticsRequirements,
 348     message.taskPerformance?.elu?.active ?? 0
 349   )
 350   updateMeasurementStatistics(
 351     workerUsage.elu.idle,
 352     eluTaskStatisticsRequirements,
 353     message.taskPerformance?.elu?.idle ?? 0
 354   )
 355   if (eluTaskStatisticsRequirements?.aggregate === true) {
 356     if (message.taskPerformance?.elu != null) {
 357       if (workerUsage.elu.utilization != null) {
 358         workerUsage.elu.utilization =
 359           (workerUsage.elu.utilization +
 360             message.taskPerformance.elu.utilization) /
 361           2
 362       } else {
 363         workerUsage.elu.utilization = message.taskPerformance.elu.utilization
 364       }
 365     }
 366   }
 367 }
 368
 369 export const createWorker = <Worker extends IWorker>(
 370   type: WorkerType,
 371   filePath: string,
 372   opts: { env?: Record<string, unknown>, workerOptions?: WorkerOptions }
 373 ): Worker => {
 374   switch (type) {
 375     case WorkerTypes.thread:
 376       return new ThreadWorker(filePath, {
 377         env: SHARE_ENV,
 378         ...opts.workerOptions
 379       }) as unknown as Worker
 380     case WorkerTypes.cluster:
 381       return cluster.fork(opts.env) as unknown as Worker
 382     default:
 383       // eslint-disable-next-line @typescript-eslint/restrict-template-expressions
 384       throw new Error(`Unknown worker type '${type}'`)
 385   }
 386 }
 387
 388 /**
 389  * Returns the worker type of the given worker.
 390  *
 391  * @param worker - The worker to get the type of.
 392  * @returns The worker type of the given worker.
 393  * @internal
 394  */
 395 export const getWorkerType = (worker: IWorker): WorkerType | undefined => {
 396   if (worker instanceof ThreadWorker) {
 397     return WorkerTypes.thread
 398   } else if (worker instanceof ClusterWorker) {
 399     return WorkerTypes.cluster
 400   }
 401 }
 402
 403 /**
 404  * Returns the worker id of the given worker.
 405  *
 406  * @param worker - The worker to get the id of.
 407  * @returns The worker id of the given worker.
 408  * @internal
 409  */
 410 export const getWorkerId = (worker: IWorker): number | undefined => {
 411   if (worker instanceof ThreadWorker) {
 412     return worker.threadId
 413   } else if (worker instanceof ClusterWorker) {
 414     return worker.id
 415   }
 416 }
 417
 418 export const waitWorkerNodeEvents = async <
 419   Worker extends IWorker,
 420   Data = unknown
 421 >(
 422   workerNode: IWorkerNode<Worker, Data>,
 423   workerNodeEvent: string,
 424   numberOfEventsToWait: number,
 425   timeout: number
 426 ): Promise<number> => {
 427   return await new Promise<number>(resolve => {
 428     let events = 0
 429     if (numberOfEventsToWait === 0) {
 430       resolve(events)
 431       return
 432     }
 433     switch (workerNodeEvent) {
 434       case 'idle':
 435       case 'backPressure':
 436       case 'taskFinished':
 437         workerNode.on(workerNodeEvent, () => {
 438           ++events
 439           if (events === numberOfEventsToWait) {
 440             resolve(events)
 441           }
 442         })
 443         break
 444       default:
 445         throw new Error('Invalid worker node event')
 446     }
 447     if (timeout >= 0) {
 448       setTimeout(() => {
 449         resolve(events)
 450       }, timeout)
 451     }
 452   })
 453 }