How to update running containers with zero downtime?

Short answer

Achieve zero downtime by using a reverse proxy or load balancer to route traffic to new containers while old ones finish processing. Blue-green and rolling deployments are the most common patterns.

Steps

1. Start new containers alongside old ones.
2. Health check the new containers.
3. Update the load balancer to point to the new containers.
4. Gracefully drain and stop the old containers.

Example

Using Docker Compose with a blue-green approach:

docker compose up -d --scale web=2 --no-recreate
# Update proxy to new container, then remove old

With Swarm rolling updates:

docker service update --image myapp:v2 --update-parallelism 1 --update-delay 10s web

Tips

• Use health checks before switching traffic.
• Ensure sessions are stored externally if using sticky sessions.
• Set appropriate grace periods for long-running requests.

Common issues

• Sticky sessions may drop if not handled during the switch.
• Database schema changes complicate rolling updates.
• Insufficient health check intervals can route traffic to unhealthy containers.