Deployment and Deployment Monitoring
A deployment is a separate running instance of your application created from the code in a workspace. Unlike a workspace, which is a development environment you actively build and iterate in, a deployment is a stable instance intended for use, demonstration, or production access.
Your workspace continues to exist and can be modified independently after you deploy from it. Changes you make in a workspace after deploying do not automatically propagate to the running deployment. To push updated code to a running deployment, use Upgrade from the Deployments page or the deployment detail page.
The default quota is five deployments per user across all applications. Contact your administrator to increase this limit.
Deploy your application
You can create a deployment from two entry points. Both open the same Create deployment modal.
Deploy from the workspace
Select Deploy in the workspace header. Before deploying, confirm that:
- The application runs correctly in Preview mode without build or runtime errors.
- Any changes you intend to include are committed to the workspace's Git branch.
Deploy from the Deployments page
Select Create deployment on the Deployments page. All three fields are editable: application, workspace, and deployment name.
Use this when you need to deploy from a workspace you are not currently working in, or when you are managing deployments across multiple applications from a central location.
When not to deploy
Avoid creating a deployment when:
- The workspace has a build error or the application is not loading in Preview mode. A broken workspace produces a deployment that immediately enters a failed or error state.
- You are mid-iteration and the current code is incomplete or unstable. Continue iterating and deploy when you reach a stable checkpoint.
- You have already reached your deployment quota. Terminate a deployment you no longer need, or contact your administrator to increase the limit.
Re-deploy to push changes
To push updated code to a running deployment, return to the workspace and select Deploy from the workspace header. Choose the same environment and select Deploy. The deployment updates with the latest workspace code.
Upgrade your deployment
Upgrading a deployment updates it with the latest code from your workspace. Upgrade is available from the row actions menu on the Deployments page and from the deployment detail page when the deployment is Running or Stopped.
The deployment transitions through an Upgrading state before returning to Running.
Do not terminate and recreate a deployment to push code changes. Use Upgrade to preserve the deployment's name and history.
Stop and resume your deployment
Stop suspends a running deployment without removing it. Use Stop when a deployment is temporarily not in use to free up resources. A stopped deployment retains its configuration and can be resumed at any time.
Resume restarts a stopped deployment. The deployment transitions through a Starting state before reaching Running.
Use these patterns to manage resource usage:
- Stop deployments used for periodic demonstrations between sessions.
- Keep production deployments running continuously and stop only for planned maintenance.
- Stop before an upgrade if you want a clean transition rather than an in-place update.
Terminate a deployment
Terminate permanently removes a deployment and frees up your quota. This cannot be undone. Terminating a deployment does not affect the workspace it was created from.
Use Terminate only when you are certain you no longer need the deployment. If there is any chance you will need it again, use Stop instead.
Monitor your deployments
The Deployments page shows all your deployments across all applications with their current status, version, and environment. The Deployments tab on an individual application's detail page shows the same data scoped to a single application.
Auto-refresh keeps the status view current at 10s, 30s, 1m, or 2m intervals. Enable this during active operations (starting or stopping) to watch a deployment transition to its final state without manually refreshing.
Diagnose deployment errors
View activity and logs
The deployment detail page provides two views into deployment activity:
- Activity section (Overview tab): Shows the most recent log entries with severity levels: INFO, WARNING, and ERROR. This is the starting point for error triage.
- Logs tab: Opens the full C3 AI Studio log viewer for the deployment. Select View All in the Activity section to open it, or add it manually with the + button in the tab bar.
When a deployment enters an Error or Failed state, open the deployment detail page and check the Activity section for ERROR-level entries. Select View All to open the Logs tab and scroll through the full log history to identify the root cause.
Respond to specific states
Starting and Upgrading: A transition is in progress. These states resolve automatically. No action is required unless the state persists beyond a few minutes, which may indicate a resource or infrastructure issue.
Error: The deployment started but encountered a runtime problem. Steps to recover:
- Check the Activity section on the deployment detail page for ERROR log entries.
- Select View All to open the Logs tab and identify the specific failure.
- If the error appears transient (for example, a service connection timeout), try stopping and resuming the deployment.
- If the error points to an application code or configuration issue, return to the workspace and reproduce the problem in Preview mode. Use the AI assistant or Code mode to diagnose and fix it. Commit the fix and re-deploy from the workspace.
Failed: The deployment could not start. This is typically a configuration or packaging issue rather than a transient error. Steps to recover:
- Open the Logs tab on the deployment detail page and look for the failure reason.
- Try resuming the deployment. If it fails repeatedly, the issue is not transient.
- Return to the workspace and verify that the application builds and runs correctly in Preview mode.
- Review recent code changes in Code mode for anything that could break the startup process.
- Fix the issue, commit the change, and re-deploy from the workspace.