Skip to main content

Scaling & Availability

Replicas​

  • Instance Count controls the requested number of Swarm service tasks.
  • Scale stateless apps whose code and background work support concurrent execution.
  • Check CPU, RAM, database connection limits, and task state with docker service ps SERVICE_NAME.

Rolling deployments and health checks​

  • Stateless apps generally use Docker Swarm start-first updates.
  • Apps with mounted data use stop-first updates by default to prevent simultaneous writers.
  • A meaningful health check helps Swarm determine readiness:
HEALTHCHECK --interval=30s --timeout=5s --start-period=30s --retries=3 \
  CMD curl -fsS http://127.0.0.1:3000/health || exit 1

Use an actual readiness endpoint and a command included in your image. Capacity, dependencies and application behavior still determine whether updates are interruption-free.

Placement​

  • Swarm schedules stateless tasks across available nodes based on constraints and capacity.
  • CapRover pins persistent apps to the node hosting local data.
  • Check registry access, node availability and placement constraints when tasks stay pending.

Stateful scaling​

  • Local volumes and bind mounts are not automatically shared between nodes.
  • Avoid forcing multiple writers against the same local data. For start-first on stateful services, the application and storage must explicitly support it.
  • Migrate data before moving a persistent app. See Persistent Storage and Multi-Node Setup.