Skip to main content
Version: 3.1

AutoscaledPool

Manages a pool of asynchronous resource-intensive tasks that are executed in parallel. The pool only starts new tasks if there is enough free CPU and memory available and the Javascript event loop is not blocked.

The information about the CPU and memory usage is obtained by the Snapshotter class, which makes regular snapshots of system resources that may be either local or from the Apify cloud infrastructure in case the process is running on the Apify platform. Meaningful data gathered from these snapshots is provided to AutoscaledPool by the SystemStatus class.

Before running the pool, you need to implement the following three functions: AutoscaledPoolOptions.runTaskFunction, AutoscaledPoolOptions.isTaskReadyFunction and AutoscaledPoolOptions.isFinishedFunction.

The auto-scaled pool is started by calling the AutoscaledPool.run function. The pool periodically queries the AutoscaledPoolOptions.isTaskReadyFunction function for more tasks, managing optimal concurrency, until the function resolves to false. The pool then queries the AutoscaledPoolOptions.isFinishedFunction. If it resolves to true, the run finishes after all running tasks complete. If it resolves to false, it assumes there will be more tasks available later and keeps periodically querying for tasks. If any of the tasks throws then the AutoscaledPool.run function rejects the promise with an error.

The pool evaluates whether it should start a new task every time one of the tasks finishes and also in the interval set by the options.maybeRunIntervalSecs parameter.

Example usage:

const pool = new AutoscaledPool({
maxConcurrency: 50,
runTaskFunction: async () => {
// Run some resource-intensive asynchronous operation here.
},
isTaskReadyFunction: async () => {
// Tell the pool whether more tasks are ready to be processed.
// Return true or false
},
isFinishedFunction: async () => {
// Tell the pool whether it should finish
// or wait for more tasks to become available.
// Return true or false
}
});

await pool.run();

Index

Constructors

constructor

  • new AutoscaledPool(options: AutoscaledPoolOptions, config?: Configuration): AutoscaledPool
  • Parameters

    • options: AutoscaledPoolOptions
    • config: Configuration = ...

    Returns AutoscaledPool

Accessors

currentConcurrency

  • get currentConcurrency(): number
  • Gets the the number of parallel tasks currently running in the pool.


    Returns number

desiredConcurrency

  • get desiredConcurrency(): number
  • set desiredConcurrency(value: number): void
  • Gets the desired concurrency for the pool, which is an estimated number of parallel tasks that the system can currently support.


    Returns number

  • Sets the desired concurrency for the pool, i.e. the number of tasks that should be running in parallel if there's large enough supply of tasks.


    Parameters

    • value: number

    Returns void

maxConcurrency

  • get maxConcurrency(): number
  • set maxConcurrency(value: number): void
  • Gets the maximum number of tasks running in parallel.


    Returns number

  • Sets the maximum number of tasks running in parallel.


    Parameters

    • value: number

    Returns void

minConcurrency

  • get minConcurrency(): number
  • set minConcurrency(value: number): void
  • Gets the minimum number of tasks running in parallel.


    Returns number

  • Sets the minimum number of tasks running in parallel.

    WARNING: If you set this value too high with respect to the available system memory and CPU, your code might run extremely slow or crash. If you're not sure, just keep the default value and the concurrency will scale up automatically.


    Parameters

    • value: number

    Returns void

Methods

abort

  • abort(): Promise<void>
  • Aborts the run of the auto-scaled pool and destroys it. The promise returned from the AutoscaledPool.run function will immediately resolve, no more new tasks will be spawned and all running tasks will be left in their current state.

    Due to the nature of the tasks, auto-scaled pool cannot reliably guarantee abortion of all the running tasks, therefore, no abortion is attempted and some of the tasks may finish, while others may not. Essentially, auto-scaled pool doesn't care about their state after the invocation of .abort(), but that does not mean that some parts of their asynchronous chains of commands will not execute.


    Returns Promise<void>

pause

  • pause(timeoutSecs?: number): Promise<void>
  • Prevents the auto-scaled pool from starting new tasks, but allows the running ones to finish (unlike abort, which terminates them). Used together with AutoscaledPool.resume

    The function's promise will resolve once all running tasks have completed and the pool is effectively idle. If the timeoutSecs argument is provided, the promise will reject with a timeout error after the timeoutSecs seconds.

    The promise returned from the AutoscaledPool.run function will not resolve when .pause() is invoked (unlike abort, which resolves it).


    Parameters

    • optionaltimeoutSecs: number

    Returns Promise<void>

resume

  • resume(): void
  • Resumes the operation of the autoscaled-pool by allowing more tasks to be run. Used together with AutoscaledPool.pause

    Tasks will automatically start running again in options.maybeRunIntervalSecs.


    Returns void

run

  • run(): Promise<void>
  • Runs the auto-scaled pool. Returns a promise that gets resolved or rejected once all the tasks are finished or one of them fails.


    Returns Promise<void>