Zero-downtime deploys

Ship while users
keep clicking.

The old version keeps serving until the new one passes its health check. Traffic shifts in seconds. If the new build looks shaky, we hold it back — and so can you.

The flow

Build, verify, cut over.

01 / Build

New version comes up

A fresh instance boots alongside the live one. It doesn't take traffic — yet.

02 / Health

Health check decides

Your /health endpoint runs against the new instance. Three consecutive 200s clears it.

03 / Route

Edge shifts traffic

New connections route to the new version. The old one drains, then retires.

Visualized

A deploy looks like this.

v22
v23
buildhealth checkroute shiftdrainretire
Safety

If the build's shaky, it stays parked.

One-click rollback

Promote any previous deploy back to live in under 4 seconds. Old image is kept warm for 24 hours.

Custom health checks

HTTP, TCP, or a shell command. Threshold and grace period are tunable per environment.

Traffic splitting

Route 5% of requests to a new build before going wide. Promote when the metrics look right.

Auto-rollback on errors

If the new build's error rate exceeds the baseline for 90 seconds, traffic snaps back.

Preview environments

Every PR gets its own URL. Burns down automatically when the branch is merged or closed.

Connection draining

Long-lived requests on the old instance run to completion — even if it takes 5 minutes.

Ship your first app today.

Closed beta. Onboarding a few builders each week — most projects are running within an hour of joining.