Hint 1 — locked

Hint 2 — locked

Hint 3 — locked

Problem

Build a distributed task scheduler that executes scheduled jobs reliably — similar to cron, but at cloud scale. Clients register tasks with a schedule (one-time or recurring) and the system executes them exactly once per trigger, even across node failures.

Functional Requirements

• Register tasks with cron expression or one-time timestamp
• Trigger task execution within 1 second of scheduled time
• Guarantee at-least-once execution (prefer exactly-once)
• Support task cancellation and rescheduling
• Provide execution history and failure logs

Non-Functional

• 10M registered tasks; 100K tasks triggered per minute
• < 1 s scheduling jitter
• No task lost or double-executed on node failure
• Horizontally scalable scheduler nodes

Prerequisites