Starting and Stopping Clusters
This page explains how to start, stop, and destroy cloud clusters. All of these actions are on the cluster's page: open Compute > Clusters and select the cluster.
Cluster Lifecycle
- Stop shuts down the controller but keeps its disk and the cluster's storage, so you keep paying for storage. You can't stop a cluster while it has active compute nodes.
- Destroy deprovisions everything in the cloud: the controller, compute nodes, disks, and ephemeral storage. The configuration stays, so you can provision it again later.
- Delete removes the configuration from your account. It's available once the cluster is destroyed.
For more detail, see What's the difference between stopping, destroying, and deleting clusters?
Starting Clusters
On the cluster's page, click ** Provision**. The cluster's Status changes to provisioning, and the cluster becomes available once provisioning finishes.
Note
It may take up to 5 minutes for a cluster to start.
Monitoring Status
While your cluster is starting, you can see its status in the provisioning logs. Go to Clusters > your cluster > Details and click the cluster's Status.
- Steps that have completed are green with a checkmark icon.
- Ongoing steps are blue with a loading icon.
- Upcoming steps are gray with a clock icon.
- If a cluster fails to start, the step at which it failed will be red with an error icon.
Stopping Clusters
To stop a cluster, navigate to its page and click ** Stop**. Please note that you cannot stop a cluster if there are any active compute nodes.
When the cluster's controller is stopped, you can start it again anytime by clicking ** Resume**.
Destroying Clusters
When you're ready to deprovision a cluster, navigate to its page and click ** Destroy**.
When you destroy a cluster, any attached nodes and ephemeral storage resources will also be terminated. Only the data in persistent storage resources will remain. It’s important to copy any data you’d like to keep to persistent storage or to a remote location.
Note
It’s essential to destroy a cluster when you’ve finished your work. Clusters that run unmonitored continue to accrue additional charges, which will be subtracted from your organization’s allotment.