Dependencies

Runix supports process dependencies — a process can wait for other processes to be running before starting.

Configuration

processes:
  postgres:
    entrypoint: postgres
    priority: 100
    health_check:
      type: tcp
      tcp_endpoint: localhost:5432

  redis:
    entrypoint: redis-server
    priority: 90
    health_check:
      type: tcp
      tcp_endpoint: localhost:6379

  api:
    entrypoint: ./cmd/api
    depends_on: [postgres, redis]
    priority: 50

  worker:
    entrypoint: worker.py
    depends_on: [postgres, redis, api]
    priority: 10

Startup Ordering

When starting from a config file, Runix resolves dependencies using topological sort:

Start order: postgres → redis → api → worker

Resolution Algorithm

1. Build a dependency graph from depends_on fields
2. Detect circular dependencies (rejected with error)
3. Perform topological sort (Kahn's algorithm)
4. Within each level, sort by priority (higher priority starts first)
5. Start processes in order, waiting for each dependency to reach running state

Priority

Priority is used as a tiebreaker when processes have no dependency relationship:

Priority	Behavior
Higher value	Starts first
Default (0)	Starts last among peers

processes:
  critical-service:
    priority: 100    # Starts first
  standard-service:
    priority: 50     # Starts second
  low-priority:
    priority: 0      # Starts last

Circular Dependency Detection

Circular dependencies are detected at config validation time:

processes:
  a:
    depends_on: [b]
  b:
    depends_on: [a]

Error: circular dependency detected: a → b → a

Runtime Behavior

When a process has dependencies:

1. Runix checks if all dependencies are in running state
2. If not, the process waits (with context cancellation support)
3. Once all dependencies are running, the process starts
4. If a dependency crashes and doesn't recover, the dependent process stays in created state

What's Next

• Process Lifecycle — Full start sequence
• Configuration Reference — depends_on and priority fields