Update Strategy

TurboCI uses a cluster-based update strategy for zero-downtime deployments with built-in safety.

How It Works

TurboCI updates clusters one at a time, ensuring high availability throughout the deployment process:

1. Cluster-Based Updates

Services are divided into clusters. Each cluster contains multiple instances that deploy simultaneously, but clusters update one at a time.

2. Load Balancer Updates

After each cluster is updated, any assigned load balancer is also updated. This ensures traffic is always routed to healthy servers.

3. Zero Downtime

Because clusters update sequentially, you always have running instances serving traffic. Even during vertical scaling, there's zero downtime.

Example Configuration

Configure clusters and instances for your deployment strategy:

services:
  web:
    instances: 3  # 3 servers per cluster
    clusters: 2   # 2 clusters = 6 total servers
    server_type: cpx21
    # ... other config

Total servers = instances × clusters. In this example: 3 × 2 = 6 servers total.

Update Process

Example: 6 servers (2 clusters × 3 instances)

✓

Cluster 1 (Servers 1-3)

All 3 servers updated → Load balancer updated → Continue

✓

Cluster 2 (Servers 4-6)

All 3 servers updated → Load balancer updated → Complete

While Cluster 1 updates, Cluster 2 continues serving traffic. While Cluster 2 updates, Cluster 1 (now updated) serves traffic.

Health Checks

Configure health checks to verify services are working correctly after deployment:

services:
  web:
    instances: 3
    clusters: 2
    healthcheck:
      cmd: curl http://localhost:3000/api/healthcheck
      test: Server Running

The cmd runs on each server, and the test string must be found in the output for the healthcheck to pass.

Next Steps

Learn how to deploy the same application across multiple cloud providers.

Continue to Multi-Cloud Setup →