Ship it.
Roll it back in one click.

With managed cloud hosting, Cloud will run a durable, idempotent release pipeline: it provisions what your app needs, holds a migration wall, waits for health, then flips traffic — and keeps the previous version live so a bad deploy rolls back instantly. This is coming soon; today you self-host and register your apps with the control plane.

deploy · timeline
Status
# Deploy timeline (dashboard, live)
queued        ✓
building      ✓
provisioning  ✓   neon + redis ready
migrating     ✓   additive — passed the wall
warming       ✓   health check ok
ready         ●   traffic activated, old superseded
              ↩   roll back in one click
Coming soon — with managed cloud hosting
The managed deploy pipeline arrives with managed cloud hosting and is not yet available. This page describes how it will work. Today you self-host: build and run your app image on your own infrastructure, and register the app with the control plane for observability and governance.

What a deploy actually does

A checkpointed, resumable workflow — every step is idempotent, so a mid-flight failure resumes instead of restarting.

Gate & provision

On the roadmap

Validates the release, then ensures the project’s managed database (Neon) and cache (Scaleway Redis) exist — created once, reused after.

Migration wall

On the roadmap

Classifies the schema change and halts destructive migrations, routing them into the approval flow. Additive changes pass straight through.

Health-gated activation

On the roadmap

Starts the new runtime service, waits for it to report healthy, then atomically activates it and supersedes the old one. Traffic only moves to a version that is up.

One-click rollback

On the roadmap

Every deploy chains a rollback target. Roll back to the previous image with a single click — no rebuild, just a re-activation of the last good release.

Self-healing reconciler

On the roadmap

A reconciler continuously checks live services against desired state — absent, failed or wrong image — and re-enqueues a redeploy with bounded backoff (a capability you’d otherwise build with k8s controllers).

Scale to zero

On the roadmap

Idle apps sleep to roughly zero cost and wake on the next request or a due scheduled job — the framework’s dormancy, managed for you.

Deploy from your pipeline

Push an image, Cloud takes it from there

When managed cloud hosting ships, the flow will be: your CI builds and pushes a container image, then Cloud runs the migration wall, health-checks the release and activates it, with automatic rollback on failure. Until then you self-host — build and run the image yourself and register the app with the control plane.

Related capabilities

See it running.

Open the dashboard on a live demo project and click through the panels yourself — logs, traces, workflows, migrations and more.