egg-schedule

A schedule plugin for egg, has been built-in plugin for egg enabled by default.

It's fully extendable for a developer and provides a simple built-in TimerStrategy.

Usage

Just add your job file to {app_root}/app/schedule.

// {app_root}/app/schedule/cleandb.js
const Subscription = require('egg').Subscription;

class CleanDB extends Subscription {
  /**
   * @property {Object} schedule
   *  - {String} type - schedule type, `worker` or `all` or your custom types.
   *  - {String} [cron] - cron expression, see [below](#cron-style-scheduling)
   *  - {Object} [cronOptions] - cron options, see [cron-parser#options](https://github.com/harrisiirak/cron-parser#options)
   *  - {String | Number} [interval] - interval expression in millisecond or express explicitly like '1h'. see [below](#interval-style-scheduling)
   *  - {Boolean} [immediate] - To run a scheduler at startup
   *  - {Boolean} [disable] - whether to disable a scheduler, usually use in dynamic schedule
   *  - {Array} [env] - only enable scheduler when match env list
   */
  static get schedule() {
    return {
      type: 'worker',
      cron: '0 0 3 * * *',
      // interval: '1h',
      // immediate: true,
    };
  }

  async subscribe() {
    await this.ctx.service.db.cleandb();
  }
}

module.exports = CleanDB;

You can also use function simply like:

exports.schedule = {
  type: 'worker',
  cron: '0 0 3 * * *',
  // interval: '1h',
  // immediate: true,
};

exports.task = async function (ctx) {
  await ctx.service.db.cleandb();
};

Overview

egg-schedule supports both cron-based scheduling and interval-based scheduling.

Schedule decision is being made by agent process. agent triggers a task and sends a message to worker process. Then, one or all worker process(es) execute the task based on schedule type.

To setup a schedule task, simply create a job file in {app_root}/app/schedule. A file contains one job and exports schedule and task properties.

The rule of thumbs is one job per file.

Task

Task is a class which will be instantiated with every schedule, and a subscribe method will be invoked.

You can get anonymous context with this.ctx.

• ctx.method: SCHEDULE
• ctx.path: /__schedule?path=${schedulePath}&${schedule}.

To create a task, subscribe can be a generator function or async function. For example:

// A simple logger example
const Subscription = require('egg').Subscription;
class LoggerExample extends Subscription {
  async subscribe() {
    this.ctx.logger.info('Info about your task');
  }
}

// A real world example: wipe out your database.
// Use it with caution. :)
const Subscription = require('egg').Subscription;
class CleanDB extends Subscription {
  async subscribe() {
    await this.ctx.service.db.cleandb();
  }
}

Scheduling

schedule is an object that contains one required property, type, and optional properties, { cron, cronOptions, interval, immediate, disable, env }.

Cron-style Scheduling

Use cron-parser.

Note: cron-parser support second as optional that is not supported by linux crontab.

@hourly / @daily / @weekly / @monthly / @yearly is also supported.

*    *    *    *    *    *
┬    ┬    ┬    ┬    ┬    ┬
│    │    │    │    │    |
│    │    │    │    │    └ day of week (0 - 7) (0 or 7 is Sun)
│    │    │    │    └───── month (1 - 12)
│    │    │    └────────── day of month (1 - 31)
│    │    └─────────────── hour (0 - 23)
│    └──────────────────── minute (0 - 59)
└───────────────────────── second (0 - 59, optional)

Example:

// To execute task every 3 hours
exports.schedule = {
  type: 'worker',
  cron: '0 0 */3 * * *',
  cronOptions: {
    // tz: 'Europe/Athens',
  }
};

Interval-style Scheduling

To use setInterval, and support ms conversion style

Example:

// To execute task every 3 hours
exports.schedule = {
  type: 'worker',
  interval: '3h',
};

Notice: Egg built-in TimerStrategy will schedule each execution at a fix rate, regardless of its execution time. So you have to make sure that your actual execution time of your task/subscribe must be smaller than your delay time.

Schedule Type

Build-in support is:

• worker: will be executed in one random worker when a schedule runs.
• all: will be executed in all workers when a schedule runs.

Custom schedule:

To create a custom schedule, simply extend agent.ScheduleStrategy and register it by agent.schedule.use(type, clz). You can schedule the task to be executed by one random worker or all workers with the built-in method this.sendOne(...args) or this.sendAll(...args) which support params, it will pass to subscribe(...args) or task(ctx, ...args).

// {app_root}/agent.js
module.exports = function(agent) {
  class CustomStrategy extends agent.ScheduleStrategy {
    start() {
      // such as mq / redis subscribe
      agent.notify.subscribe('remote_task', data => {
        this.sendOne(data);
      });
    }
  }
  agent.schedule.use('custom', CustomStrategy);
};

Then you could use it to defined your job:

// {app_root}/app/schedule/other.js
const Subscription = require('egg').Subscription;
class ClusterTask extends Subscription {
  static get schedule() {
    return {
      type: 'custom',
    };
  }
  async subscribe(data) {
    console.log('got custom data:', data);
    await this.ctx.service.someTask.run();
  }
}
module.exports = ClusterTask;

Dynamic schedule

// {app_root}/app/schedule/sync.js
module.exports = app => {
  class SyncTask extends app.Subscription {
    static get schedule() {
      return {
        interval: 10000,
        type: 'worker',
        // only start task when hostname match
        disable: require('os').hostname() !== app.config.sync.hostname,
        // only start task at prod mode
        env: [ 'prod' ],
      };
    }
    async subscribe() {
      await this.ctx.sync();
    }
  }
  return SyncTask;
}

Configuration

Logging

See ${appInfo.root}/logs/{app_name}/egg-schedule.log which provided by config.customLogger.scheduleLogger.

// config/config.default.js
config.customLogger = {
  scheduleLogger: {
    // consoleLevel: 'NONE',
    // file: path.join(appInfo.root, 'logs', appInfo.name, 'egg-schedule.log'),
  },
};

Customize directory

If you want to add additional schedule directories, you can use this config.

// config/config.default.js
config.schedule = {
  directory: [
    path.join(__dirname, '../app/otherSchedule'),
  ],
};

Testing

app.runSchedule(scheduleName) is provided by egg-schedule plugin only for test purpose.

Example:

it('test a schedule task', async function () {
  // get app instance
  await app.runSchedule('clean_cache');
});

Questions & Suggestions

Please open an issue here.

License

MIT