Startup Grace Period
The startup grace period controls how long the load balancer waits after a new container starts before failing health checks can cause it to be replaced. It protects slow-booting applications from being restarted mid-boot during deployments.
When You Need It
Section titled “When You Need It”Some applications take a while to become ready — for example, a CMS that runs database migrations, cache rebuilds, or configuration imports on startup. Without enough grace, the platform can see failing health checks during that boot work and replace the container before it ever finishes starting, causing a deployment loop and visible errors.
How It Works
Section titled “How It Works”- Set Startup grace period (seconds) on the environment’s configuration form (next to the scaling settings). The default is 120 seconds; the range is 0–3600.
- During the grace window, failing load-balancer health checks are ignored for newly started containers.
- The grace period is a ceiling, not a delay: a container that passes health checks after 20 seconds enters service at 20 seconds. Raising the grace period never slows down a healthy deployment.
- The value is stored with your environment configuration and re-applied on every deployment.
Choosing a Value
Section titled “Choosing a Value”| Application behavior | Suggested value |
|---|---|
| Static site / lightweight service | 120 (default) |
| Boots in under a minute | 120 (default) |
| Runs migrations or cache warms on boot (1–5 min) | 300–600 |
| Long startup jobs | Up to 3600 |
If you see deployments where new containers are repeatedly replaced before the application finishes booting, raise the grace period to comfortably exceed your worst-case boot time.