Deterministic Job Scheduler (Rust)

A deterministic, crash-safe job scheduler and execution engine implemented in Rust.

This project demonstrates how to build infrastructure-grade background workers with:

• Explicit state management
• Bounded concurrency
• Deterministic scheduling
• Crash-safe behavior
• Clear failure modes

There is no UI and no SaaS layer.
This is a backend control-plane component.

---

What this does (plain English)

• Watches a database table for jobs
• Selects a limited number of jobs at a time
• Executes them in the background
• Records success or failure
• Repeats continuously

Think of it as a small engine that safely pulls work from a database and runs it in controlled batches.

---

Core Properties

• Deterministic scheduling

  • Jobs are selected in a predictable order
  • Concurrency is strictly bounded

• Crash-safe

  • Jobs in progress are not lost
  • Failures are explicit and recorded

• Fail-fast configuration

  • Missing critical config causes startup failure
  • No silent misconfiguration

• Infrastructure-first

  • No framework magic
  • No hidden state
  • Clear ownership boundaries

---

Architecture Overview

• Domain layer – job state, transitions, failure types
• Scheduler – pure, deterministic job selection
• Executor – bounded async execution with timeouts
• Storage – PostgreSQL-backed persistence
• Orchestrator – control loop that ties everything together

This is a single-node v1 designed to be extended, not a finished product.

---

Running locally

Requirements

• Rust (stable)
• PostgreSQL (local or Docker)

Environment variables

DATABASE_URL=postgres://postgres:postgres@localhost:5432/deterministic_scheduler
MAX_CONCURRENCY=10
SCHEDULER_TICK_MS=500
JOB_TIMEOUT_SECS=5
RUST_LOG=info