Deployments
Follow, cancel and redeploy deployments.
Every time a service is built or rolled out, Unbind creates a deployment. The service's Deployments tab lists them, newest first, with the commit that caused each one.
Statuses
| Status | What it means |
|---|---|
| QUEUED | Waiting for a free build slot. |
| BUILDING | The build is running. |
| LAUNCHING | The build is done and the new version is starting. |
| ACTIVE | This version is running and serving traffic. |
| FAILED | The build failed. Open Build Logs. |
| ERROR | The build worked, but the new version could not start. Open Deploy Logs for the reason. |
| CRASHING | It started and keeps stopping. Open Deploy Logs. |
| CANCELLED | Someone aborted the build. |
| REMOVED | An older deployment that was replaced, or one you removed. |
A failed deployment does not take your service down. The last working version keeps running until a new one is ready.
What starts a deployment
- A push to GitHub, for services created from a repository. See Deploy from GitHub.
- The Deploy button on the changes bar, after you change settings or variables.
- Redeploy or Rollback on an existing deployment.
- The API. See API & MCP.
Staged changes
Changing a setting or a variable does not touch the running service. The change is staged, and a bar appears at the bottom: Apply 3 changes.
- Details shows every change with its current and new value, and which services are affected.
- Deploy rolls everything out at once.
- Discard All throws the changes away.
A changed field is highlighted and shows a revert button, which puts back the value that is deployed.
Deleting a service, volume, environment or project also discards the changes staged for it.
Unbind does the least work that makes the change take effect. Each affected service shows what will happen to it:
| Action | When |
|---|---|
| Rebuild | The source or a build setting changed, like the branch or the build command. |
| Redeploy | A runtime setting changed, like replicas, ports or a volume. The current image is reused. |
| Restart | Only variables changed. |
| No rollout | The change does not need a rollout, or the service is not running. |
Because variables can reference other services, one change can affect several services. They all show up in the list before you confirm.
You can deploy changes while a build is running. They go live right away on the current image, and the build keeps them when it finishes.
Creating or deleting a service, and the actions below, happen right away. They are not staged.
How the old version stops
When a deployment replaces a version, the old replicas get a stop signal (SIGTERM) and 5 seconds to finish what they are doing. A replica that is still running after that is stopped by force.
A service with a volume gets about 2 seconds instead, because the new version can't start until the old one is gone.
A service with Rollout set to Stop first also waits for the old version to stop, but keeps the 5 seconds so it can finish its work.
Actions on a deployment
Open the menu on a deployment.
Redeploy
Runs the deployment again. Two options:
- Skip cache builds from scratch. Use it when the cache seems to be the problem.
- Skip build if possible reuses the existing image and only rolls it out again.
Rollback
Available on older deployments. It brings that version back using the image that was already built, so it is fast and needs no build.
Restart
Restarts the running version without building or changing anything.
Abort
Stops a build that is queued or running.
Remove
Takes the current deployment down. The service goes offline until you deploy again. Databases cannot be removed this way.
Logs
Each deployment has Build Logs and Deploy Logs. The second one is what your app prints while it runs. See Logs for searching.