Define work requests

The getting started guide covered how to create a simple WorkRequest and enqueue it.

In this guide you will learn how to define and customize WorkRequest objects to handle common use cases, such as how to:

• Schedule one-time and recurring work
• Set work constraints like requiring Wi-Fi or charging
• Guarantee a minimum delay in work execution
• Set retry and back-off strategies
• Pass input data to work
• Group related work together using tags

Overview

Work is defined in WorkManager via a WorkRequest. In order to schedule any work with WorkManager you must first create a WorkRequest object and then enqueue it.

val myWorkRequest = ...
WorkManager.getInstance(myContext).enqueue(myWorkRequest)

WorkRequest myWorkRequest = ...
WorkManager.getInstance(myContext).enqueue(myWorkRequest);

The WorkRequest object contains all of the information needed by WorkManager to schedule and run your work. It includes constraints which must be met for your work to run, scheduling information such as delays or repeating intervals, retry configuration, and may include input data if your work relies on it.

WorkRequest itself is an abstract base class. There are two derived implementations of this class that you can use to create the request, OneTimeWorkRequest and PeriodicWorkRequest. As their names imply, OneTimeWorkRequest is useful for scheduling non-repeating work, whilst PeriodicWorkRequest is more appropriate for scheduling work that repeats on some interval.

Schedule one-time work

For simple work, which requires no additional configuration, use the static method from:

val myWorkRequest = OneTimeWorkRequest.from(MyWork::class.java)

WorkRequest myWorkRequest = OneTimeWorkRequest.from(MyWork.class);

For more complex work, you can use a builder.

val uploadWorkRequest: WorkRequest =
   OneTimeWorkRequestBuilder<MyWork>()
       // Additional configuration
       .build()

WorkRequest uploadWorkRequest =
   new OneTimeWorkRequest.Builder(MyWork.class)
       // Additional configuration
       .build();

Schedule periodic work

Your app may at times require that certain work runs periodically. For example, you may want to periodically backup your data, download fresh content in your app, or upload logs to a server.

Here is how you use the PeriodicWorkRequest to create a WorkRequest object which executes periodically:

val saveRequest =
       PeriodicWorkRequestBuilder<SaveImageToFileWorker>(1, TimeUnit.HOURS)
    // Additional configuration
           .build()

PeriodicWorkRequest saveRequest =
       new PeriodicWorkRequest.Builder(SaveImageFileWorker.class, 1, TimeUnit.HOURS)
           // Constraints
           .build();

In this example, the work is scheduled with a one hour interval.

The interval period is defined as the minimum time between repetitions. The exact time that the worker is going to be executed depends on the constraints that you are using in your WorkRequest object and on the optimizations performed by the system.

Flexible run intervals

If the nature of your work makes it sensitive to run timing, you can configure your PeriodicWorkRequest to run within a flex period inside each interval period, as shown in Figure 1.

Figure 1. Diagram shows repeating intervals with the flexible period in which the work can run.

To define periodic work with a flex period, you pass a flexInterval along with the repeatInterval when creating the PeriodicWorkRequest. The flex period begins at repeatInterval - flexInterval, and goes to the end of the interval.

The following is an example of periodic work that can run during the last 15 minutes of every one hour period.

val myUploadWork = PeriodicWorkRequestBuilder<SaveImageToFileWorker>(
       1, TimeUnit.HOURS, // repeatInterval (the period cycle)
       15, TimeUnit.MINUTES) // flexInterval
    .build()

WorkRequest saveRequest =
       new PeriodicWorkRequest.Builder(UploadWork.class,
               1, TimeUnit.HOURS,
               15, TimeUnit.MINUTES)
           .build();

The repeat interval must be greater than or equal to PeriodicWorkRequest.MIN_PERIODIC_INTERVAL_MILLIS and the flex interval must be greater than or equal to PeriodicWorkRequest.MIN_PERIODIC_FLEX_MILLIS.

Effect of Constraints on Periodic Work

You can apply constraints to periodic work. For example, you could add a constraint to your work request such that the work only runs when the user’s device is charging. In this case, even if the defined repeat interval passes, the PeriodicWorkRequest will not run until this condition is met. This could cause a particular run of your work to be delayed, or even skipped if the conditions are not met within the run interval.

Work constraints

Constraints ensure that work is deferred until optimal conditions are met. The following constraints are available to WorkManager.

To create a set of constraints and associate it with some work, create a Constraints instance using the Contraints.Builder() and assign it to your WorkRequest.Builder().

For example, the following code builds a work request which only runs when the user’s device is both charging and on Wi-Fi:

val constraints = Constraints.Builder()
   .setRequiredNetworkType(NetworkType.UNMETERED)
   .setRequiresCharging(true)
   .build()

val myWorkRequest: WorkRequest =
   OneTimeWorkRequestBuilder<MyWork>()
       .setConstraints(constraints)
       .build()

Constraints constraints = new Constraints.Builder()
       .setRequiredNetworkType(NetworkType.UNMETERED)
       .setRequiresCharging(true)
       .build();

WorkRequest myWorkRequest =
       new OneTimeWorkRequest.Builder(MyWork.class)
               .setConstraints(constraints)
               .build();

When multiple constraints are specified, your work will run only when all the constraints are met.

In the event that a constraint becomes unmet while your work is running, WorkManager will stop your worker. The work will then be retried when all the constraints are met.

Delayed Work

In the event that your work has no constraints or that all the constraints are met when your work is enqueued, the system may choose to run the work immediately. If you do not want the work to be run immediately, you can specify your work to start after a minimum initial delay.

Here is an example of how to set your work to run at least 10 minutes after it has been enqueued.

val myWorkRequest = OneTimeWorkRequestBuilder<MyWork>()
   .setInitialDelay(10, TimeUnit.MINUTES)
   .build()

WorkRequest myWorkRequest =
      new OneTimeWorkRequest.Builder(MyWork.class)
               .setInitialDelay(10, TimeUnit.MINUTES)
               .build();

While the example illustrates how to set an initial delay for a OneTimeWorkRequest, you can also set an initial delay for a PeriodicWorkRequest. In that case, only the first run of your periodic work would be delayed.

Retry and Backoff Policy

If you require that WorkManager retry your work, you can return Result.retry() from your worker. Your work is then rescheduled according to a backoff delay and backoff policy.

• Backoff delay specifies the minimum amount of time to wait before retrying your work after the first attempt. This value can be no less than 10 seconds (or MIN_BACKOFF_MILLIS).

• Backoff policy defines how the backoff delay should increase over time for subsequent retry attempts. WorkManager supports 2 backoff policies, LINEAR and EXPONENTIAL.

Every work request has a backoff policy and backoff delay. The default policy is EXPONENTIAL with a delay of 10 seconds, but you can override this in your work request configuration.

Here is an example of customizing the backoff delay and policy.

val myWorkRequest = OneTimeWorkRequestBuilder<MyWork>()
   .setBackoffCriteria(
       BackoffPolicy.LINEAR,
       OneTimeWorkRequest.MIN_BACKOFF_MILLIS,
       TimeUnit.MILLISECONDS)
   .build()

WorkRequest myWorkRequest =
       new OneTimeWorkRequest.Builder(MyWork.class)
               .setBackoffCriteria(
                       BackoffPolicy.LINEAR,
                       OneTimeWorkRequest.MIN_BACKOFF_MILLIS,
                       TimeUnit.MILLISECONDS)
               .build();

In this example, the minimum backoff delay is set to the minimum allowed value, 10 seconds. Since the policy is LINEAR the retry interval will increase by approximately 10 seconds with each new attempt. For instance, the first run finishing with Result.retry() will be attempted again after 10 seconds, followed by 20, 30, 40, and so on, if the work continues to return Result.retry() after subsequent attempts. If the backoff policy were set to EXPONENTIAL, the retry duration sequence would be closer to 20, 40, 80, and so on.

Tag work

Every work request has a unique identifier, which can be used to identify that work later in order to cancel the work or observe its progress.

If you have a group of logically related work, you may also find it helpful to tag those work items. Tagging allows you to operate with a group of work requests together.

For example, WorkManager.cancelAllWorkByTag(String) cancels all Work Requests with a particular tag, and WorkManager.getWorkInfosByTag(String) returns a list of the WorkInfo objects which can be used to determine the current work state.

The following code shows how you can add a "cleanup" tag to your work:

val myWorkRequest = OneTimeWorkRequestBuilder<MyWork>()
   .addTag("cleanup")
   .build()

WorkRequest myWorkRequest =
       new OneTimeWorkRequest.Builder(MyWork.class)
       .addTag("cleanup")
       .build();

Finally, multiple tags can be added to a single work request. Internally these tags are stored as a set of strings. From a work request, you retrieve its set of tags via WorkRequest.getTags().

Assign input data

Your work may require input data in order to do its work. For example, work that handles uploading an image might require the URI of the image to be uploaded as input.

Input values are stored as key-value pairs in a Data object and can be set on the work request. WorkManager will deliver the input Data to your work when it executes the work. The Worker class can access the input arguments by calling Worker.getInputData(). The code below shows how you can create a Worker instance which requires input data and how to send it in your work request.

// Define the Worker requiring input
class UploadWork(appContext: Context, workerParams: WorkerParameters)
   : Worker(appContext, workerParams) {

   override fun doWork(): Result {
       val imageUriInput =
           inputData.getString("IMAGE_URI") ?: return Result.failure()

       uploadFile(imageUriInput)
       return Result.success()
   }
   ...
}

// Create a WorkRequest for your Worker and sending it input
val myUploadWork = OneTimeWorkRequestBuilder<UploadWork>()
   .setInputData(workDataOf(
       "IMAGE_URI" to "http://..."
   ))
   .build()

// Define the Worker requiring input
public class UploadWork extends Worker {

   public UploadWork(Context appContext, WorkerParameters workerParams) {
       super(appContext, workerParams);
   }

   @NonNull
   @Override
   public Result doWork() {
       String imageUriInput = getInputData().getString("IMAGE_URI");
       if(imageUriInput == null) {
           return Result.failure();
       }

       uploadFile(imageUriInput);
       return Result.success();
   }
   ...
}

// Create a WorkRequest for your Worker and sending it input
WorkRequest myUploadWork =
      new OneTimeWorkRequest.Builder(UploadWork.class)
           .setInputData(
               new Data.Builder()
                   .putString("IMAGE_URI", "http://...")
                   .build()
           )
           .build();

Similarly, the Data class can be used to output a return value. Input and output data are covered in more detail in the section input parameters and returned values.

Next Steps

In the States and observation page, you’ll learn more about work states and how to monitor the progress of your work.