Distributed Scheduler Service

System Overview

The Scheduler service is the pulse of your application. It manages thousands of concurrent tasks, ensuring that whether a job is delayed by 5 seconds or 5 months, it executes precisely when it should, with guaranteed at-least-once delivery.

Worker Architecture

The service uses a leader-election model for the scheduler loop and a pull-based worker model for task execution, ensuring that no job is lost even if individual workers crash.

Scheduling Jobs

Interact with the scheduler via ConnectRPC. Jobs are identified by a unique key to prevent accidental duplicate scheduling.

const response = await schedulerClient.createJob({
  job: {
    key: "onboarding.reminder:user-123",
    type: "HTTP_CALLBACK",
    schedule: {
      oneOff: {
        executeAt: { seconds: Date.now() / 1000 + 3600 } // 1 hour from now
      }
    },
    callbackConfig: {
      url: "https://api.myapp.com/webhooks/onboarding",
      method: "POST"
    },
    payload: {
      userId: "user-123",
      template: "welcome_reminder"
    }
  }
});

CLI Operations

Use the lr-ctl tool to inspect and trigger jobs manually during development.

Observability & Monitoring

The Scheduler service emits real-time metrics for job performance, allowing you to track execution patterns and latency.