The Scheduler API gives apps the possibility of creating tasks that run in a defined schedule. It can be a one-time event or a recurring task. It uses agenda.js as the backend, so the schedule syntax and internal processes are all according to its documentation.
See the details on the API in the Module scheduler definition. There's also an example app that can be used as a guide.
As per the Permission System, the Schedule API needs the following permission in your app's manifest (app.json
file):
{
permissions: [
{ "name": "scheduler" }
]
}
To use the Scheduler API you'll need two things:
- The functions to be run as jobs (we call them
processors
). - The schedule in which they run.
This is configured in the app's extendConfiguration
method. During the app's startup, the processors are registered to make them available for scheduling.
{% code lineNumbers="true" %}
public async extendConfiguration(configuration: IConfigurationExtend) {
configuration.scheduler.registerProcessors([
{
id: 'first',
processor: async (jobData) => console.log(`[${ Date() }] this is a task`, jobData),
},
]);
}
{% endcode %}
The processor
is an async
function and can receive arguments (jobData
). The arguments are passed during scheduling.
To trigger the registered processor as a job, you must provide:
- The
id
of the processor. - The type of job it will be. The available types are
scheduleRecurring
(the job runs in aninterval
) andscheduleOnce
(the job runs once using thewhen
keyword).
Starting a job can be done when running a slash command, for example:
{% code lineNumbers="true" %}
// slashcommand class
public async executor(context: SlashCommandContext, read: IRead, modify: IModify): Promise<void> {
// SCHEDULING A RECURRING TASK
const task = {
id: 'first',
interval: '10 seconds',
data: { test: true },
};
await modify.getScheduler().scheduleRecurring(task);
// SCHEDULING ONETIME TASK
const task = {
id: 'first',
when: '8 seconds',
};
await modify.getScheduler().scheduleOnce(task);
}
{% endcode %}
You can also trigger a job as soon as it gets registered, without the need for manual or automated triggering. When you register your processor in the extendConfiguration
method, you can pass a prop called startupSetting
in the processor's object:
{% code lineNumbers="true" %}
import { StartupType } from '@rocket.chat/apps-engine/definition/scheduler';
// ...
configuration.scheduler.registerProcessors([
{
id: 'first',
processor: async (jobData) => console.log(`[${ Date() }] this is a task`, jobData),
startupSetting: {
type: StartupType.ONETIME,
when: '20 seconds',
data: { test: true },
}
},
{
id: 'second',
processor: async (jobData) => console.log(`[${ Date() }] this is a task`, jobData),
startupSetting: {
type: StartupType.RECURRING,
interval: '20 seconds',
}
},
]);
{% endcode %}
This indicates that you want that particular processor to be scheduled as soon as it is registered. You can define the "immediate scheduling" as a recurring job (StartupType.RECURRING
) or a one-time job (StartupType.ONETIME
). You can also pass data using the data
object. It will work just like when you schedule a task using the modify
accessor.
Here, data
is not something that's passed to the processor or a function as a living object or executable code. This data
is a static piece of data that is passed to the processor's first argument.
The signature of the processor function is as follows:
{% code overflow="wrap" %}
(jobContext: IJobContext, read: IRead, modify: IModify, http: IHttp, persis: IPersistence) => Promise<void>
{% endcode %}
The first argument is the data object you're passing when actually scheduling the job; ([k: string]:any
). The rest of the arguments are passed when the function is run.
To stop a job, all you have to do is pass the ID of the job you want to stop:
const jobId = 'first';
await modify.getScheduler().cancelJob(jobId);
It will stop the running job (if any).
To stop all the current running jobs from the app:
await modify.getScheduler().cancelAllJobs();