serverpod/serverpod-scheduling

Serverpod Scheduling (Future Calls)

Future calls run at a specified time, are stored in the database, and survive restarts. Each call is claimed for one execution across the cluster. Future calls require database support; execution is started by monolith/maintenance roles, not serverless role.

Defining a future call

class ExampleFutureCall extends FutureCall {
  Future<void> doWork(Session session) async {
    // Work here.
  }

  Future<void> doOtherWork(Session session, MyModel data) async {
    // Work with data.
  }
}

Run serverpod generate for the type-safe API on pod.futureCalls.

Supported parameter and return types

The same parameters as for endpoint methods are supported:

• Primitives: bool, int, double, String
• Duration, DateTime (UTC), ByteData, UuidValue, Uri, BigInt
• Generated serializable models (from .spy.yaml)
• List, Map, Set, Record — strictly typed with the above

Scheduling

// Delay
await pod.futureCalls
    .callWithDelay(const Duration(hours: 1))
    .example.doWork();

// At specific time (UTC)
await pod.futureCalls
    .callAtTime(DateTime.utc(2026, 1, 1))
    .example.doOtherWork(myModel);

// With identifier (for cancellation)
await pod.futureCalls
    .callWithDelay(const Duration(hours: 1), identifier: 'my-job-id')
    .example.doWork();

// Recurring task from `Duration` interval with an optional start time
await pod.futureCalls
    .callRecurring()
    .every(const Duration(hours: 1), start: DateTime.now())
    .example.doWork();

// Recurring task from `cron` expression
await pod.futureCalls
    .callRecurring()
    .cron("0 * * * *")
    .example.doWork();

await pod.futureCalls.cancel('my-job-id');  // Cancels all with that identifier

Handle failures inside the call and reschedule if the work must eventually succeed.

Configuration

futureCallExecutionEnabled: true  # SERVERPOD_FUTURE_CALL_EXECUTION_ENABLED
futureCall:
  concurrencyLimit: 5             # SERVERPOD_FUTURE_CALL_CONCURRENCY_LIMIT (default 1, <1 maps to unlimited and is not recommended)
  scanInterval: 2000              # SERVERPOD_FUTURE_CALL_SCAN_INTERVAL (ms, default 5000)

Keep future calls idempotent. A call should tolerate retries or restarts without duplicating irreversible side effects.