Skip to main content
Restart a resource when its current process needs to be replaced without changing the Stackfile or deploying a new release.
For example:
Use --stack when the target is not in the current context:
The command exits after Stackdome accepts the restart request and writes its confirmation to stderr. It does not wait for the replacement process to become ready. The current command also does not emit a structured result when -o json or -o yaml is selected.

​
What restart changes

A restart asks Stackdome to replace the running instance of one resource.
  • It does not edit stackfile.yaml or the stored stack configuration.
  • It does not create a release.
  • It does not select an older release or roll back code or configuration.
  • It does not change persistent volume or addon data.
Use a deployment for configuration or source changes. Use the dashboard-only release rollback when you need to redeploy a previous released snapshot.

​
Verify the restart

Observe logs and status immediately after the request. Use separate terminals if you want both streams live:
Wait for the resource to return to its ready state and check the new startup lines for the expected initialization. Then capture a final structured status result:
Require converged_release.state to remain Released and converged_release.health to return to ok. If a newer release exists, also confirm that its ID matches converged_release.id before describing the current deployment as successful.

​
If the resource does not recover

Run the bounded checks before changing configuration:
Use the newest false or failing condition and the matching log lines to decide whether the issue is a runtime crash, readiness failure, missing dependency, or application error. If the release itself failed during an image build, switch to the build-failure guide.