Skip to main content

Deletion & Cleanup Operations

Overview​

Proper resource cleanup ensures efficient quota utilization and cost management. This guide covers deletion procedures for various resources.

Delete JupyterHub Server​

Prerequisites​

  • Back up data from JupyterHub server

Delete Process​

Step 1: Open JupyterHub and Start Deletion​

  1. Log in as a Tenant User.
  2. In the left sidebar, click AI Studio. In the new tab that opens, click JupyterHub in the sidebar.
  3. Locate the server to delete, click the three-dot menu, then click Delete.

JupyterHub server list with delete option

Step 2: Confirm Deletion​

  1. In the confirmation popup, click Delete.

Confirm JupyterHub server deletion

Step 3: Verify Deletion​

  1. Confirm the JupyterHub server no longer appears in the JupyterHub list.

JupyterHub list after deletion

Delete Volumes​

Prerequisites​

  • Back up volume data you want to keep
  • Delete all JupyterHub servers that are using the volume before you delete the Volume

Delete Process​

Step 1: Open Volumes and Start Deletion​

  1. Log in as a Tenant User.
  2. In the left sidebar, click AI Studio. In the new tab that opens, click Volumes in the sidebar.
  3. Locate the volume to delete and click the Delete icon.

Volumes list

Step 2: Confirm Deletion​

  1. In the confirmation popup, enter the volume name exactly as shown.
  2. Click Delete volume.

Confirm volume deletion

Step 3: Verify Deletion​

  1. Confirm the volume no longer appears in the Volumes list.

Volume deleted

Delete Datasets​

Prerequisites​

  • Delete all JupyterHub servers that are using the datasets before you delete the Datasets

Delete Process​

Step 1: Open Datasets and Start Deletion​

  1. Log in as a Tenant User.
  2. In the left sidebar, click AI Studio. In the new tab that opens, click Datasets in the sidebar.
  3. Locate the Datasets to delete and click the Delete icon.

Dataset list

Step 2: Confirm Deletion​

  1. In the confirmation popup, enter the Datasets name exactly as shown.
  2. Click Delete dataset.

Confirm dataset deletion

Step 3: Verify Deletion​

  1. Confirm the Datasets no longer appears in the Datasets list.

Dataset

Delete Applications​

Prerequisites​

  • Stop all running workloads
  • Back up application data
  • Remove associated endpoints

Delete Process​

Step 1: Open Cluster Workloads​

  1. Log in as Tenant Admin or NCP Admin.
  2. Go to Cluster and open the required cluster.
  3. Open the Workloads tab.

Step 2: Delete the Application​

  1. Click the delete icon for the workload.
  2. In the confirmation popup, click Delete.

Delete Application

Delete Application Pop Up

Step 3: Verify Deletion​

  1. Confirm the application is removed from the cluster workload list.

Delete Application Success

Delete Storage Classes​

Prerequisites​

  • Delete all volumes that use the storage class
  • Delete all datasets that use the storage class

Delete Process​

Step 1: Open Storage Classes​

  1. Log in as Tenant Admin or NCP Admin.
  2. Go to Cluster and open the required cluster.
  3. Open the Storage Classes tab.

Storage Classes tab

Step 2: Delete Storage Class​

  1. Click the Delete icon for the storage class to remove.
  2. In the confirmation popup, confirm and click Delete button.

Confirm storage class deletion

Step 3: Verify Deletion​

  1. Confirm the storage class no longer appears in the Storage Classes list.

Storage class deleted

Delete CSI Drivers​

Delete Process​

Step 1: Open Uninstall Dialog​

  1. Log in as Tenant Admin or NCP Admin.
  2. Navigate to BareMetal → Kubernetes, click the three-dot menu on the cluster, and select View.
  3. Click the NFS icon. The uninstall CSI driver dialog opens.

Uninstall CSI driver dialog

Step 2: Confirm Uninstall​

  1. In the confirmation popup, click Uninstall driver.

CSI driver uninstalled successfully

Delete Models​

Prerequisites​

  • Remove model endpoints
  • Complete inference requests
  • Archive model files if needed

Delete Process​

Step 1: Open Deployed Models​

  1. Log in as Tenant Admin.
  2. Go to Models → Deployed Models.
  3. Click the delete icon for the target model.

Delete Model

Step 2: Confirm Deletion​

  1. Enter the deployed model name in the confirmation popup.
  2. Click Delete.

Delete Model Pop Up

Step 3: Verify Deletion​

  1. Confirm the model entry is removed from the UI.

Delete Model Success

Delete Endpoints​

Prerequisites​

  • Update client applications to use new endpoints
  • Redirect traffic if applicable
  • Complete pending requests

Delete Process​

Step 1: Open Endpoint Delete Action​

  1. Log in as Tenant Admin.
  2. In Endpoints, open the target endpoint menu (three dots).
  3. Click Delete.

Delete Endpoint

Step 2: Confirm Deletion​

  1. Enter the endpoint name in the confirmation popup.
  2. Click Delete.

Delete Endpoint Pop Up

Step 3: Verify Deletion​

  1. Confirm the endpoint entry is removed from the UI.

Delete Endpoint Success

Delete Clusters​

Prerequisites for Deletion​

  • Remove all workloads from cluster
  • Delete deployed models
  • Stop all services
  • Back up configuration if needed

Pre-Deletion Checklist​

  • All pods evicted
  • Persistent data backed up
  • Configuration exported
  • Network policies recorded
  • Monitoring data archived

Delete Process​

Step 1: Drain Workloads​

Gracefully stop all workloads:

kubectl drain <node-name>

Or for all nodes:

kubectl get nodes
kubectl drain <node1> <node2> <node3>

Step 2: Delete Workloads​

Delete remaining workloads:

kubectl delete deployment --all
kubectl delete statefulset --all
kubectl delete pod --all

Step 3: Delete Cluster​

  1. Log in as Tenant Admin and go to Compute → Cluster.
  2. Open the cluster menu (three dots), click Delete, enter the cluster name, then click Delete.

Delete Cluster

Delete Cluster Pop Up

Step 4: Monitor and Verify Cluster Deletion​

  1. Monitor cluster status until it shows Deleting.
  2. Verify the cluster is removed from the UI after deletion completes.

Delete Cluster Process

Delete Cluster Success

Delete Slurm Cluster​

Prerequisites for Deletion​

  • Stop all running jobs
  • Ensure no active workloads

Step 1: Delete Slurm Cluster​

  1. Open the cluster menu (three dots).
  2. Click Delete for the Slurm cluster.

Slurm Delete Button

  1. Enter the cluster name in the confirmation popup.
  2. Click Delete.

Slurm Delete Popup

Step 2: Monitor Slurm Cluster Status​

  1. After deletion starts, status changes to Deleting.

Slurm Delete Status

  1. Wait for UI status refresh (typically 1–2 minutes).
  2. Verify the Slurm cluster is removed from the UI after successful deletion.

Slurm Delete Status

Delete Storage​

Prerequisites​

  • Delete all storage classes that use the storage (and any volumes, datasets, or clusters using it) before you delete the storage

Delete Process​

Step 1: Open Storage and Start Deletion​

  1. Log in as Tenant Admin.
  2. Navigate to the Storage tab.
  3. Click the three-dot menu on the storage to delete, then click Delete.

Storage tab

Step 2: Confirm Deletion​

  1. Enter the storage name in the confirmation popup and click Delete button to complete the deletion.

Confirm storage deletion

Step 3: Verify Deletion​

  1. After deletion starts, the status changes to Deleting.
  2. Wait for the UI to refresh (typically 1–2 minutes).
  3. Confirm the storage is removed from the Storage tab.

Storage list after deletion

Deallocate Virtual Machine​

Prerequisites​

  • Back up any data you need from the VM
  • Stop or migrate any running services on the VM
  • Delete any clusters or workloads that use this VM

Deallocation Process​

Step 1: Open VM menu and start deallocation​

  1. Log in as Tenant Admin.
  2. Go to Compute → Virtual Machine.
  3. Locate the VM to deallocate, click the ellipsis (three-dot) menu on the row, then click Deallocate.

VM Deallocate

Step 2: Confirm deallocation​

  1. In the confirmation popup, enter the VM name exactly as shown.
  2. Click Deallocate.

VM Deallocate

Step 3: Monitor and verify deallocation​

  1. The VM status changes to Terminating. This may take a few minutes.
  2. Wait for deallocation to complete.
  3. Verify the VM is removed from the Virtual Machine list.

VM Deallocate

Deallocate BareMetal​

Prerequisites​

  • All clusters using servers must be deleted
  • Ensure no running services
  • Back up data from servers

Deallocation Process​

Step 1: Open Bare Metal Deallocation​

  1. Log in as Tenant Admin.
  2. Ensure dependent clusters are already deleted.
  3. Go to Compute → BareMetal, open the ellipsis menu, and click Deallocate.

Deallocate BM

Step 2: Confirm Deallocation​

  1. Enter the compute name in the confirmation popup.
  2. Click Deallocate.

Deallocate BM Pop UP

Step 3: Monitor and Verify​

  1. Monitor node status as Processing.
  2. Refresh the page if status updates are delayed.
  3. Verify compute nodes are successfully deallocated.

Deallocate BM Process

Deallocate BM Success

Delete VPC and Subnet​

Delete Process​

Step 1: Delete Subnet​

  1. Go to the Network sidebar.
  2. Click the VPC eye icon to view subnets.

Network Tab

  1. Click the three-dot menu on the subnet to delete, then click Delete.

VPC View

  1. Enter the subnet name in the confirmation popup and click Delete.

Subnet Pop UP

  1. Confirm the success message.

Subnet Delete

Step 2: Delete VPC​

  1. Go to the Network sidebar.
  2. Click the three-dot menu on the VPC, then click Delete VPC.

VPC Menu

  1. Enter the VPC name in the confirmation popup and click Delete.

VPC Pop Up

  1. Confirm the success message.

VPC Delete

Delete Security Groups​

Delete Process​

Step 1: Open Delete Action​

  1. Go to the “Security Groups” sidebar.
  2. Click the security group menu icon (three dots).
  3. Click Delete.

Step 2: Confirm Deletion​

  1. Enter the security group name in the confirmation popup.
  2. Click Delete.

Step 3: Verify Deletion​

  1. Confirm the security group is removed from the list.

Delete Tenant Users​

Before Deletion​

  • Assign the user’s workloads/resources to another user
  • Back up user data if required
  • Revoke external integrations or API access

Delete Process​

Step 1: Open User Management​

  1. Log in as Tenant Admin.
  2. Open User Management from the sidebar.

Step 2: Delete the User​

  1. Locate the target user in the list.
  2. Click the Delete icon.
  3. Click Confirm in the popup.

Delete User

Delete User Pop Up

Step 3: Verify Deletion​

  1. Confirm the user is removed from the list.

Delete User Success

Cleanup Workflows​

Complete Cluster Cleanup​

Remove all resources from a cluster:

# Delete all workloads
kubectl delete all --all-namespaces

# Delete persistent volumes
kubectl delete pvc --all-namespaces

# Delete configuration
kubectl delete configmap --all-namespaces
kubectl delete secret --all-namespaces

# Delete cluster
kubectl delete cluster <cluster-name>

Full Tenant Cleanup​

Clean up entire tenant environment:

  1. Delete Applications - Remove all deployments
  2. Delete Models - Remove all models
  3. Delete Endpoints - Remove all endpoints
  4. Delete Clusters - Remove all clusters
  5. Deallocate Resources - Return BareMetal
  6. Delete Users - Remove tenant users

Resource Dependencies​

Understanding deletion order:

Endpoints → Models
         ↓
      Clusters → Applications/Workloads
         ↓
      BareMetal Resources

Delete in this order:

  1. Endpoints
  2. Models
  3. Applications
  4. Clusters
  5. BareMetal
  6. Users

Data Backup Before Deletion​

What to Back Up​

  • Model files and weights
  • Training data
  • Configuration files
  • Logs and metrics
  • Application code

Backup Commands​

# Download models
kubectl cp <pod>:/models ./local/models

# Export configurations
kubectl get all -o yaml > cluster-config.yaml

# Backup persistent data
kubectl exec <pod> -- tar czf - /data > data-backup.tar.gz

Verification After Deletion​

Verify Resource Removal​

Check resources are deleted:

# List remaining resources
kubectl get all --all-namespaces

# Check persistent volumes
kubectl get pv,pvc

# Verify cluster status
kubectl cluster-info

Check Quota Recovery​

Verify quota is returned:

  1. Navigate to Tenant Settings
  2. Check Quota Allocation
  3. Confirm freed resources appear as available

Quota Check

Recovery Options​

Deleted Resource Recovery​

  • Endpoints - Recreate by redeploying models
  • Models - Redeploy from backup
  • Applications - Redeploy from image
  • Clusters - Create new cluster
  • BareMetal - Request reallocation from admin
  • Users - Create new user account

Data Recovery​

If accidentally deleted:

  1. Contact Tenant Admin immediately
  2. Request backup restoration
  3. Provide deletion timestamp
  4. Restore from available backups

Note: Deleted resources cannot be recovered if no backups exist.

Best Practices​

Before Deletion​

  • Always backup critical data
  • Document configurations
  • Export metrics and logs
  • Verify no dependencies
  • Notify team members

Deletion Strategy​

  • Delete in proper order (dependencies first)
  • Verify deletion success
  • Monitor quota recovery
  • Document deletion reasons
  • Archive deletion records

Preventive Measures​

  • Use version control for code
  • Backup critical models regularly
  • Document architecture
  • Tag important resources
  • Set up alerts for quota usage

Next Steps​

Last updated on