Deploys, logs and restarts
Read deployment status, stream build logs, cancel a stuck build, restart the app and understand what old images are kept for.
Every change to an app produces a deployment. This page covers the lifecycle controls and where to look when something goes wrong.
Deployment states
| State | What it means |
|---|---|
queued |
Waiting for a build worker |
building |
Dependencies and the image are being prepared |
deploying |
The new container is starting and being routed |
success |
The release is live |
failed |
The build or start failed; the previous release stays live |
cancelled |
Cancelled before completion |
Where to see deployments
- App → Deployments — every deployment for this app, with commit, branch, author, duration and date.
- Websites & Apps → Deployments — every app in the organization. Search by app, domain, branch or commit; filter by state.
Click a row for the deployment detail, or Logs for the streaming build log. Copy takes the log to your clipboard for a support ticket.
Cancel, redeploy, restart
- Cancel is available only while a deployment is
buildingorqueued. Once it is deploying, let it finish. - Redeploy queues a fresh build of the current app settings and source. Use it after changing an environment variable, a domain, or a registry tag.
- Restart restarts the running container without rebuilding. Use it for a wedged process, not for a code change.
- Stop takes the app offline. Start brings it back.
- Delete removes the app and its deployments permanently and frees the slot.
Stopping an app does not stop billing for the plan; the slot remains allocated.
Runtime logs
App → Logs shows what the process prints after it starts. The view auto-refreshes every 10 seconds; use Refresh for a manual pull.
Runtime logs are for diagnosis, not retention. If you need a durable audit trail, ship logs from your application to Z2 Object Storage or an external service.
Build vs runtime failures
Build fails — the log ends with a compiler, dependency or Docker step error. The previous release stays live. Fix the code and push again. Common causes are in Failed deployments.
Build succeeds, app crashes on start — the build log looks clean but the deployment never becomes healthy. Check:
- The start command is correct for the build pack.
- The app listens on
0.0.0.0and on the expected port. - Required environment variables exist at runtime.
- The health check path returns a 2xx/3xx response.
App runs but returns 502 — the container is up but the proxy cannot reach it. Almost always the port or bind address.
Old images and rollback
Settings → Advanced → Images to keep controls how many previous images are retained. Zoophost does not expose a one-click rollback button: to return to an earlier version, deploy the earlier commit again (or revert the change and push).
Keeping more images gives you more recovery options on the platform side. It also consumes storage.
Health checks
Under Settings → Health check you can enable an HTTP probe:
- Path, Port, Method, Expected code
- Interval, timeout, retries and start period
Without a health check, a container that starts but serves errors is still marked deployed. For production apps, add a cheap endpoint such as /health that checks the app's dependencies.
Scheduled jobs and workers
Scheduled jobs, background workers and queue consumers are supported by deploying them as their own app with a start command that runs the worker process. Give each its own slot and keep it separate from the web app so a restart does not interrupt both.
Next
- deployment
- logs
- build
- restart
- stop
- redeploy
- rollback
- images
Ready to use this
Git-push containers, previews and custom domains, billed monthly in INR.
- 5 websites on the smallest plan
- Free SSL and custom domains

