Manage Your Deployments
The Deployments page lists every deployment you can reach. Use the page to see which deployments are running, open a deployment, control whether a deployment stays up, and determine what to do when a deploy fails.
Select Deployments to open the page.
Filters
Four controls narrow the list:
- Search: filters by name as you type.
- All Environments: filters to staging or to production.
- All Statuses: filters to a single state.
- Auto refresh: keeps the list current, because a deployment can change state without any action from you. A refresh control updates the list immediately.
Columns
Each row is one deployment:
- Name: the deployment name. Select the name to open the deployment.
- Application: the application this deployment belongs to. Select the application to open the application.
- Version: the version this deployment is running, stamped at deploy time.
- Environment: staging or production.
- Status: the current state, with a color indicator.
- Hibernation: the nightly hibernation time, or No hibernation when the deployment is excluded from hibernation. For a deployment with hibernation paused, hover the information icon to see the date the schedule returns.
- Last Deployed: when the most recent successful deploy finished. This column tracks successes only, so a recent failure leaves the earlier time in place.
States
A deployment reports one state at a time:
- Starting: the deployment is coming up, which follows a deploy or a resume.
- Running: the application is serving, and links to the application work.
- Hibernating: the deployment is shutting down on its nightly hibernation schedule. Once it finishes, it reports Stopped.
- Stopped: the deployment is down until you resume it. A staging deployment usually reports Stopped because it hibernated on its nightly schedule. Someone can also stop a deployment by hand.
- Degraded: the deployment is running without being healthy. Open the deployment to read the reason before deploying again.
- Error: the deployment failed. Open the deployment to read the error behind the failure.
- Failed to Start: the deploy failed before C3 Code created the deployment, so nothing is running.
Every state has a route forward. Starting and Hibernating resolve automatically. Stopped exits through Resume. Degraded, Error, and Failed to Start exit through a new deploy from a running workspace.
Actions
Each deployment offers these actions:
- Stop: shuts the deployment down. The deployment stays down until you resume it, and uses no resources meanwhile.
- Resume: restarts a stopped deployment, including one that hibernated, on the version that deployment was already running. Newer code reaches the deployment through a deploy.
- Rename: changes the deployment name, and leaves the application and the staging or production setting unchanged.
- Launch App: opens the deployed application in a new tab.
- Deployment settings: opens this deployment's settings, including the hibernation controls.
- Delete deployment: removes the deployment and returns it to your quota.
Stopping, resuming, renaming, or deleting a deployment, and changing its hibernation, need admin access. The deployment's admins and C3 Code administrators have it. A deployment's admins are the person who created it and the application's admins. Platform and cluster administrators also have it. For anyone else, these controls are disabled, and hovering one shows the reason. Rename and Delete deployment stay unavailable while a deploy is in progress, or while the deployment is starting or stopping. Hover over either action to see what it waits for.
Returning to earlier code means deploying that code again, because deployments only move forward.
For the deploy procedure itself, see Deploy to Staging and Production.
Hibernation
Staging hibernates on a schedule. Production keeps running.
The Hibernation column shows the nightly hibernation time for deployments that hibernate, and No hibernation for deployments that keep running.
Change hibernation from the Hibernation card. The card appears on a deployment's Overview tab and on its Deployment settings page.
To keep a deployment running before a demonstration or a review, pause hibernation:
- Under Pause hibernation for, select 1 day, 3 days, or 7 days.
- Select Pause.
While the pause lasts, the deployment keeps running, the switch shows as off, and the card shows the date the pause ends. After that date, the nightly schedule applies again with no action from you.
To keep a deployment running with no end date, turn off the Hibernation switch. The deployment then keeps running until someone stops it.
Turn the switch back on to restore the nightly schedule. Doing so also ends a pause before its end date.
When the switch is on, the card shows the nightly time, such as 9 PM PST. That time comes from the cluster's settings, so every deployment on the cluster shares it.
Hibernation preserves the deployment. A deployment that hibernated resumes on the same version, holding the same data.
Open a deployment
Select a deployment name to open that deployment. The deployment opens with two tabs, Overview and User Management, with Overview showing. You can open further tabs for the C3 AI Studio tools that run against that deployment, including the logs.
To let people use a deployed application, add them on the deployment's User Management tab. Tools you open from your workspace act on the workspace, not on staging or production.
A stopped deployment opens with a notice on the Overview tab, and Resume takes the place of Launch App at the top of the page. Select Resume to start the deployment. The status moves to Starting, then to Running, and resuming can take a few minutes.
The Overview tab holds three sections:
- Deployment Details: the name, the version, staging or production, the current status, and when the last deploy finished. This section carries the rename control, Resume or Stop, and Delete deployment.
- Hibernation: the nightly schedule and the controls to pause or turn off hibernation. See Hibernation.
- Activity: what the deployment has logged. Activity counts Errors, Warnings, and Informational entries, then lists the most recent entries with a time and a message. The section refreshes automatically, and View all opens the full log.
When the latest deploy went live with known issues, a Latest deployment went live with known issues box lists each issue.
Read the Activity counts before calling a deployment healthy. A Running status tells you the deployment is serving, and the counts tell you how the application inside that deployment is behaving.
To open another tool, select + next to the tabs and choose a tool, or select View all to see every tool on a new tab. The tools need a running deployment. On a stopped deployment, the new tab shows Resume deployment, which works the same as Resume.
Recover from a failed or degraded deploy
When the most recent deploy failed or went live with known issues, the deployment reports what happened. What you do next depends on which case applies.
The deployment exists and the last attempt failed: what you deployed before may still be serving. Fix the cause, then deploy again from a running workspace.
The deployment reports Failed to Start: the deploy failed before C3 Code created the deployment, so there is nothing to keep. Delete the deployment, or deploy again from a running workspace.
The deploy went live with known issues: the deployment is serving, but reports as Degraded. Read the issues in the box on the Overview tab, fix them, and deploy again.
Select the warning icon next to Last Deployment to open Deployment Failure Details, which pairs the last successful deploy with the failed attempt and shows the error text. Copy the error text and give it to the agent.
Where to go next
These pages cover deploying, scheduling, and configuring a deployment: