If you need to perform a data transfer that may take a long time, you can create a JobScheduler job and identify it as a user-initiated data transfer (UIDT) job. UIDT jobs are intended for longer-duration data transfers that are initiated by the device user, such as downloading a file from a remote server. UIDT jobs were introduced with Android 14 (API level 34).
User-initiated data transfer jobs are started by the user. These jobs require a notification, start immediately, and may be able to run for an extended period of time as system conditions allow. You can run several user-initiated data transfer jobs concurrently.
User initiated jobs must be scheduled while the application is visible to the user (or in one of the allowed conditions). After all constraints are met, user initiated jobs can be executed by the OS, subject to system health restrictions. The system may also use the provided estimated payload size to determine how long the job executes.
安排用户发起的数据传输作业
To run a user initiated data-transfer job, do the following:
Make sure your app has declared the
JobService
and associated permissions in its manifest:<service android:name="com.example.app.CustomTransferService" android:permission="android.permission.BIND_JOB_SERVICE" android:exported="false"> ... </service>
Also, define a concrete subclass of
JobService
for your data transfer:Kotlin
class CustomTransferService : JobService() { ... }
Java
class CustomTransferService extends JobService() { .... }
Declare the
RUN_USER_INITIATED_JOBS
permission in the manifest:<manifest ...> <uses-permission android:name="android.permission.RUN_USER_INITIATED_JOBS" /> <application ...> ... </application> </manifest>
Call the
setUserInitiated()
method when building aJobInfo
object. (This method is available beginning with Android 14.) We also recommend that you offer a payload size estimate by callingsetEstimatedNetworkBytes()
while creating your job.Kotlin
val networkRequestBuilder = NetworkRequest.Builder() // Add or remove capabilities based on your requirements. // For example, this code specifies that the job won't run // unless there's a connection to the internet (not just a local // network), and the connection doesn't charge per-byte. .addCapability(NET_CAPABILITY_INTERNET) .addCapability(NET_CAPABILITY_NOT_METERED) .build() val jobInfo = JobInfo.Builder(jobId, ComponentName(mContext, CustomTransferService::class.java)) // ... .setUserInitiated(true) .setRequiredNetwork(networkRequestBuilder) // Provide your estimate of the network traffic here .setEstimatedNetworkBytes(1024 * 1024 * 1024) // ... .build()
Java
NetworkRequest networkRequest = new NetworkRequest.Builder() // Add or remove capabilities based on your requirements. // For example, this code specifies that the job won't run // unless there's a connection to the internet (not just a local // network), and the connection doesn't charge per-byte. .addCapability(NET_CAPABILITY_INTERNET) .addCapability(NET_CAPABILITY_NOT_METERED) .build(); JobInfo jobInfo = JobInfo.Builder(jobId, new ComponentName(mContext, CustomTransferService.class)) // ... .setUserInitiated(true) .setRequiredNetwork(networkRequest) // Provide your estimate of the network traffic here .setEstimatedNetworkBytes(1024 * 1024 * 1024) // ... .build();
While the job is being executed, call
setNotification()
on theJobService
object. CallingsetNotification()
makes the user aware that the job is running, both in the Task Manager and in the status bar notification area.When execution is complete, call
jobFinished()
to signal to the system that the job is complete, or that the job should be rescheduled.Kotlin
class CustomTransferService: JobService() { private val scope = CoroutineScope(Dispatchers.IO) @RequiresApi(Build.VERSION_CODES.UPSIDE_DOWN_CAKE) override fun onStartJob(params: JobParameters): Boolean { val notification = Notification.Builder(applicationContext, NOTIFICATION_CHANNEL_ID) .setContentTitle("My user-initiated data transfer job") .setSmallIcon(android.R.mipmap.myicon) .setContentText("Job is running") .build() setNotification(params, notification.id, notification, JobService.JOB_END_NOTIFICATION_POLICY_DETACH) // Execute the work associated with this job asynchronously. scope.launch { doDownload(params) } return true } private suspend fun doDownload(params: JobParameters) { // Run the relevant async download task, then call // jobFinished once the task is completed. jobFinished(params, false) } // Called when the system stops the job. override fun onStopJob(params: JobParameters?): Boolean { // Asynchronously record job-related data, such as the // stop reason. return true // or return false if job should end entirely } }
Java
class CustomTransferService extends JobService{ @RequiresApi(Build.VERSION_CODES.UPSIDE_DOWN_CAKE) @Override public boolean onStartJob(JobParameters params) { Notification notification = Notification.Builder(getBaseContext(), NOTIFICATION_CHANNEL_ID) .setContentTitle("My user-initiated data transfer job") .setSmallIcon(android.R.mipmap.myicon) .setContentText("Job is running") .build(); setNotification(params, notification.id, notification, JobService.JOB_END_NOTIFICATION_POLICY_DETACH) // Execute the work associated with this job asynchronously. new Thread(() -> doDownload(params)).start(); return true; } private void doDownload(JobParameters params) { // Run the relevant async download task, then call // jobFinished once the task is completed. jobFinished(params, false); } // Called when the system stops the job. @Override public boolean onStopJob(JobParameters params) { // Asynchronously record job-related data, such as the // stop reason. return true; // or return false if job should end entirely } }
Periodically update the notification to keep the user informed of the job's status and progress. If you cannot determine the transfer size ahead of scheduling the job, or need to update the estimated transfer size, use the new API,
updateEstimatedNetworkBytes()
to update the transfer size after it becomes known.
Recommendations
To run UIDT jobs effectively, do the following:
Clearly define network constraints and job execution constraints to specify when the job should be executed.
Execute the task asynchronously in
onStartJob()
; for example, you can do this by using a coroutine. If you don't run the task asynchronously, the work runs on the main thread and might block it, which can cause an ANR.To avoid running the job longer than necessary, call
jobFinished()
when the transfer finishes, whether it succeeds or fails. That way, the job doesn't run longer than necessary. To discover why a job was stopped, implement theonStopJob()
callback method and callJobParameters.getStopReason()
.
向后兼容性
There is currently no Jetpack library that supports UIDT jobs. For this reason, we recommend that you gate your change with code that verifies that you're running on Android 14 or higher. On lower Android versions, you can use WorkManager's foreground service implementation as a fallback approach.
Here's an example of code that checks for the appropriate system version:
Kotlin
fun beginTask() { if (Build.VERSION.SDK_INT < Build.VERSION_CODES.UPSIDE_DOWN_CAKE) { scheduleDownloadFGSWorker(context) } else { scheduleDownloadUIDTJob(context) } } private fun scheduleDownloadUIDTJob(context: Context) { // build jobInfo val jobScheduler: JobScheduler = context.getSystemService(Context.JOB_SCHEDULER_SERVICE) as JobScheduler jobScheduler.schedule(jobInfo) } private fun scheduleDownloadFGSWorker(context: Context) { val myWorkRequest = OneTimeWorkRequest.from(DownloadWorker::class.java) WorkManager.getInstance(context).enqueue(myWorkRequest) }
Java
public void beginTask() { if (Build.VERSION.SDK_INT < Build.VERSION_CODES.UPSIDE_DOWN_CAKE) { scheduleDownloadFGSWorker(context); } else { scheduleDownloadUIDTJob(context); } } private void scheduleDownloadUIDTJob(Context context) { // build jobInfo JobScheduler jobScheduler = (JobScheduler) context.getSystemService(Context.JOB_SCHEDULER_SERVICE); jobScheduler.schedule(jobInfo); } private void scheduleDownloadFGSWorker(Context context) { OneTimeWorkRequest myWorkRequest = OneTimeWorkRequest.from(DownloadWorker.class); WorkManager.getInstance(context).enqueue(myWorkRequest) }
停止 UIDT 作业
Both the user and the system can stop user-initiated transfer jobs.
由用户从任务管理器提供
The user can stop a user-initiated data transfer job that appears in the Task Manager.
At the moment that the user presses Stop, the system does the following:
- Terminates your app's process immediately, including all other jobs or foreground services running.
- Doesn't call
onStopJob()
for any running jobs. - Prevents user-visible jobs from being rescheduled.
For these reasons, it's recommended to provide controls in the notification posted for the job to allow gracefully stopping and rescheduling the job.
Note that, under special circumstances, the Stop button doesn't appear next to the job in the Task Manager, or the job isn't shown in the Task Manager at all.
由系统提供
与常规作业不同,用户发起的数据传输作业不受应用待机模式存储分区配额的影响。但是,如果出现以下任一情况,系统仍会停止作业:
- 不再满足开发者定义的约束条件。
- 系统确定该作业的运行时间超出了完成数据传输任务所需的时间。
- 系统需要优先考虑系统运行状况,并因发热程度上升而停止作业。
- 应用进程因设备内存不足而被终止。
如果系统因设备内存不足以外的原因停止作业,系统会调用 onStopJob()
,并在系统认为最佳的时间重试作业。确保您的应用可以保留数据传输状态(即使未调用 onStopJob()
),并且您的应用可以在再次调用 onStartJob()
时恢复此状态。
允许安排用户发起的数据传输作业的条件
Apps can only start a user-initiated data transfer job if the app is in the visible window, or if certain conditions are met:
- If an app can launch activities from the background, it can also launch user-initiated data transfer jobs from the background.
- If an app has an activity in the back stack of an existing task on the Recents screen, that alone doesn't allow a user-initiated data transfer job to run.
If the job is scheduled to run at a time when the necessary conditions are not
met, the job fails and returns a RESULT_FAILURE
error code.
适用于用户发起的数据传输作业的约束条件
To support jobs running at optimal points, Android offers the ability to assign constraints to each job type. These constraints are available as of Android 13.
Note: The following table only compares the constraints that vary between each job type. See JobScheduler developer page or work constraints for all constraints.
The following table shows the different job types that support a given job constraint, as well as the set of job constraints that WorkManager supports. Use the search bar before the table to filter the table by the name of a job constraint method.
These are the constraints allowed with user-initiated data transfer jobs:
setBackoffCriteria(JobInfo.BACKOFF_POLICY_EXPONENTIAL)
setClipData()
setEstimatedNetworkBytes()
setMinimumNetworkChunkBytes()
setPersisted()
setNamespace()
setRequiredNetwork()
setRequiredNetworkType()
setRequiresBatteryNotLow()
setRequiresCharging()
setRequiresStorageNotLow()
测试
The following list shows some steps on how to test your app's jobs manually:
- To get the job ID, get the value that is defined upon the job being built.
To run a job immediately, or to retry a stopped job, run the following command in a terminal window:
adb shell cmd jobscheduler run -f APP_PACKAGE_NAME JOB_ID
To simulate the system force-stopping a job (due to system health or out-of-quota conditions), run the following command in a terminal window:
adb shell cmd jobscheduler timeout TEST_APP_PACKAGE TEST_JOB_ID
另请参阅
其他资源
如需详细了解用户发起的数据转移,请参阅以下其他资源:
- UIDT 集成案例研究:Google 地图使用用户发起的数据传输 API 将下载可靠性提高了 10%