toad-scheduler

In-memory TypeScript job scheduler that repeatedly executes given tasks within specified intervals of time (e. g. "each 20 seconds"). Cron syntax is also supported in case you need it.

Node.js 12+ and modern browsers are supported

Getting started

First install the package:

1npm i toad-scheduler

Next, set up your jobs:

1const { ToadScheduler, SimpleIntervalJob, Task } = require('toad-scheduler')
2
3const scheduler = new ToadScheduler()
4
5const task = new Task('simple task', () => { counter++ })
6const job = new SimpleIntervalJob({ seconds: 20, }, task)
7
8scheduler.addSimpleIntervalJob(job)
9
10// when stopping your app
11scheduler.stop()

Usage with async tasks

In order to avoid unhandled rejections, make sure to use AsyncTask if your task is asynchronous:

1const { ToadScheduler, SimpleIntervalJob, AsyncTask } = require('toad-scheduler')
2
3const scheduler = new ToadScheduler()
4
5const task = new AsyncTask(
6    'simple task', 
7    () => { return db.pollForSomeData().then((result) => { /* continue the promise chain */ }) },
8    (err: Error) => { /* handle error here */ }
9)
10const job = new SimpleIntervalJob({ seconds: 20, }, task)
11
12scheduler.addSimpleIntervalJob(job)
13
14// when stopping your app
15scheduler.stop()

Note that in order to avoid memory leaks, it is recommended to use promise chains instead of async/await inside task definition. See talk on common Promise mistakes for more details.

Asynchronous error handling

Note that your error handlers can be asynchronous and return a promise. In such case an additional catch block will be attached to them, and should there be an error while trying to resolve that promise, and logging error will be logged using the default error handler (console.error).

Preventing task run overruns

In case you want to prevent second instance of a task from being fired up while first one is still executing, you can use preventOverrun options:

1import { ToadScheduler, SimpleIntervalJob, Task } from 'toad-scheduler';
2
3const scheduler = new ToadScheduler();
4
5const task = new Task('simple task', () => {
6    // if this task runs long, second one won't be started until this one concludes
7	console.log('Task triggered');
8});
9
10const job = new SimpleIntervalJob(
11	{ seconds: 20, runImmediately: true },
12	task,
13    { 
14        id: 'id_1',
15        preventOverrun: true,
16    }
17);
18
19//create and start jobs
20scheduler.addSimpleIntervalJob(job);

Using IDs and ES6-style imports

You can attach IDs to tasks to identify them later. This is helpful in projects that run a lot of tasks and especially if you want to target some of the tasks specifically (e. g. in order to stop or restart them, or to check their status).

1import { ToadScheduler, SimpleIntervalJob, Task } from 'toad-scheduler';
2
3const scheduler = new ToadScheduler();
4
5const task = new Task('simple task', () => {
6	console.log('Task triggered');
7});
8
9const job1 = new SimpleIntervalJob(
10	{ seconds: 20, runImmediately: true },
11	task,
12    { id: 'id_1' }
13);
14
15const job2 = new SimpleIntervalJob(
16	{ seconds: 15, runImmediately: true },
17	task,
18    { id: 'id_2' }
19);
20
21//create and start jobs
22scheduler.addSimpleIntervalJob(job1);
23scheduler.addSimpleIntervalJob(job2);
24
25// stop job with ID: id_2
26scheduler.stopById('id_2');
27
28// remove job with ID: id_1
29scheduler.removeById('id_1');
30
31// check status of jobs
32console.log(scheduler.getById('id_1').getStatus()); // returns Error (job not found)
33
34console.log(scheduler.getById('id_2').getStatus()); // returns "stopped" and can be started again
35

Cron support

You can use CronJob instances for handling Cron-style scheduling:

1      const task = new AsyncTask('simple task', () => {
2        // Execute your asynchronous logic here
3      })
4      const job = new CronJob(
5        {
6          cronExpression: '*/2 * * * * *',
7        },
8        task,
9        {
10          preventOverrun: true,
11        }
12      )
13      scheduler.addCronJob(job)

Note that you need to install "croner" library for this to work. Run npm i croner in order to install this dependency.

Usage in clustered environments

toad-scheduler does not persist its state by design, and has no out-of-the-box concurrency management features. In case it is necessary to prevent parallel execution of jobs in clustered environment, it is highly recommended to use redis-semaphore in your tasks.

Here is an example:

1import { randomUUID } from 'node:crypto'
2
3import type { Redis } from 'ioredis'
4import type { LockOptions } from 'redis-semaphore'
5import { Mutex } from 'redis-semaphore'
6import { AsyncTask } from 'toad-scheduler';
7
8export type BackgroundJobConfiguration = {
9    jobId: string
10}
11
12export type LockConfiguration = {
13    lockName?: string
14    refreshInterval?: number
15    lockTimeout: number
16}
17
18// Abstract Job
19export abstract class AbstractBackgroundJob {
20    public readonly jobId: string
21    protected readonly redis: Redis
22
23    protected constructor(
24        options: BackgroundJobConfiguration,
25        redis: Redis,
26    ) {
27        this.jobId = options.jobId
28        this.redis = redis
29    }
30
31    protected abstract processInternal(executionUuid: string): Promise<void>
32
33    async process() {
34        const uuid = randomUUID()
35
36        try {
37            await this.processInternal(uuid)
38        } catch (err) {
39            console.error(logObject)
40        }
41    }
42
43    protected getJobMutex(key: string, options: LockOptions) {
44        return new Mutex(this.redis, this.getJobLockName(key), options)
45    }
46
47    protected async tryAcquireExclusiveLock(lockConfiguration: LockConfiguration) {
48        const mutex = this.getJobMutex(lockConfiguration.lockName ?? 'exclusive', {
49            acquireAttemptsLimit: 1,
50            refreshInterval: lockConfiguration.refreshInterval,
51            lockTimeout: lockConfiguration.lockTimeout,
52        })
53
54        const lock = await mutex.tryAcquire()
55        // If someone else already has this lock, skip
56        if (!lock) {
57            return
58        }
59
60        return mutex
61    }
62
63    protected getJobLockName(key: string) {
64        return `${this.jobId}:locks:${key}`
65    }
66}
67
68// Job example
69
70const LOCK_TIMEOUT_IN_MSECS = 60 * 1000
71const LOCK_REFRESH_IN_MSECS = 10 * 1000
72
73export class SampleJob extends AbstractBackgroundJob {
74  constructor(redis: Redis) {
75    super(
76      {
77        jobId: 'SampleJob',
78      },
79      redis,
80    )
81  }
82
83  protected async processInternal(executionUuid: string): Promise<void> {
84    // We only want a single instance of this job running in entire cluster, let's see if someone else is already processing it
85    const lock = await this.tryAcquireExclusiveLock({
86      lockTimeout: LOCK_TIMEOUT_IN_MSECS,
87      refreshInterval: LOCK_REFRESH_IN_MSECS,
88    })
89
90    // Job is already running, skip
91    if (!lock) {
92      this.logger.debug(`Job already running in another node, skipping (${executionUuid})`)
93      return
94    }
95
96    try {
97      // Process job logic here
98      await this.sampleJobLogic()
99    } finally {
100      await lock.release()
101    }
102  }
103
104  private async sampleJobLogic() {
105    // dummy processing logic
106    return Promise.resolve()
107  }
108
109
110// Job registration
111function createTask(job: AbstractBackgroundJob): AsyncTask {
112    return new AsyncTask(
113        job.jobId,
114        () => {
115            return job.process()
116        },
117    )
118}

API for schedule

• days?: number - how many days to wait before executing the job for the next time;
• hours?: number - how many hours to wait before executing the job for the next time;
• minutes?: number - how many minutes to wait before executing the job for the next time;
• seconds?: number - how many seconds to wait before executing the job for the next time;
• milliseconds?: number - how many milliseconds to wait before executing the job for the next time;
• runImmediately?: boolean - if set to true, in addition to being executed on a given interval, job will also be executed immediately when added or restarted.

API for jobs

• start(): void - starts, or restarts (if it's already running) the job;
• stop(): void - stops the job. Can be restarted again with start command;
• getStatus(): JobStatus - returns the status of the job, which is one of: running, stopped.

API for scheduler

• addSimpleIntervalJob(job: SimpleIntervalJob): void - registers and starts a new job;
• addLongIntervalJob(job: LongIntervalJob): void - registers and starts a new job with support for intervals longer than 24.85 days;
• addIntervalJob(job: SimpleIntervalJob | LongIntervalJob): void - registers and starts new interval-based job;
• stop(): void - stops all jobs, registered in the scheduler;
• getById(id: string): Job - returns the job with a given id.
• existsById(id: string): boolean - returns true if job with given id exists, false otherwise.
• getAllJobs(): Job[] - returns all registered jobs (only jobs with an id are in the registry).
• getAllJobsByStatus(status: JobStatus): Job[] - returns all registered jobs with a given status.
• stopById(id: string): void - stops the job with a given id.
• removeById(id: string): Job | undefined - stops the job with a given id and removes it from the scheduler. If no such job exists, returns undefined, otherwise returns the job.
• startById(id: string): void - starts, or restarts (if it's already running) the job with a given id.