Ctrlk

Upgrade Management Plane and Hosts

The upgrade of the self-hosted Private Cloud Director is a two-phase, sequential process:

  • Management Plane: Upgrading the Management Plane updates core services, management APIs, region-specific configurations, and orchestration components.

  • Hosts: Once the management plane has been upgraded, hosts in each region are being upgraded to ensure compatibility, seamless communication, and completion of the overall upgrade process

Before you begin the upgrade, ensure you meet all prerequisites.

Prerequisites for Upgrade

Before you upgrade, complete the following checks to ensure your environment is ready.

Verify Management Plane Health

Before you upgrade, verify that the Management Plane is healthy. Run airctl status and confirm that region health shows as Ready. If any regions are not ready, resolve all issues before continuing. For example, pods may be in a CrashLoopBackOff state.

Manual Backup

You must perform a Manual Backup Procedure of your management plane infrastructure before starting the upgrade. This backup is critical for recovery.

Upgrade Management Plane and Cluster

Upgrade your Private Cloud Director environment to access the latest features, security patches, and performance improvements through this two-step process.

Upgrade Management Plane

Updating the Management Plane running on the Management Cluster ensures multiple components are automatically upgraded as part of an airctl upgrade process.

Here are the components that will be affected.

  • Core services and applications.

  • Management APIs and user interfaces.

  • Region-specific configurations and services.

  • Service coordination and orchestration components.

The Impact of Management Plane Upgrade

NOTE

Temporary service interruptions may occur during the upgrade process.

  • The user interface will be unavailable during the upgrade.

  • All API services will be temporarily unavailable.

  • Create, Read, Update, and Delete operations will not be possible during this time.

  • Running workloads on VMs will not be impacted.

Step 1: Download and run the Installer Artifacts

airctl is the command-line installer for Private Cloud Director. Run the following commands only on a verified management cluster node to download airctl with the required installation artifacts.

Air-gapped environments only

If the management node does not have internet access, download the installer artifacts on a jump host with internet connectivity and transfer them to the management node.

Follow the artifact download steps from the air-gapped installation guide, then skip this step and continue using the copied artifacts.

  1. Download the Installer Script and Artifacts using the following command.

You can choose to download a specific version of the installer script and artifact.

For example, you can replace latest with a specific version, such as v-2026.1.1-4444155

Here is an example of a modified command:

NOTE

Replace <YOUR_USER_AGENT_KEY> with your user key.

  1. Make the Installer Executable using the following command.

  1. Run the Installation Script using the following command.

The command installs the specific version of airctl based on the version.txt file.

  1. Create symbolic links for airctl and its configuration file:

Proxy Configuration (Optional)

If your environment uses a network proxy, update the values /opt/pf9/airctl/conf/helm_values/bork.template.yml again, as the changes would be lost after a new build is installed. Update it again as shown below:

The no_proxy list should include the master-ips, worker-ips, external-ip4, master-vip4, and any other addresses where traffic should not be routed through a proxy server.

NOTE

The upgrade-cluster step is skipped in this release, as the Kubernetes version remains unchanged at v1.30.

Air-gapped environments only

Ensure that all required container images for the target version are already available in the private registry accessible by the cluster nodes before running the upgrade.

If not already done, download the image list for the target version and push the images to your private registry. Refer to the air-gapped installation guide for detailed steps.

Step 2: Upgrade all regions

To upgrade all regions set up on your Private Cloud Director, execute the following command.

Optionally, you can upgrade a specific region by replacing <REGION_NAME> with your target region.

Here is the modified example command.

To monitor and diagnose the upgrade logs, add --verbose.

Here is the sample command.

Verify: Upgrade Success on Management Plane

After completing the upgrade for the management plane, verify that you can continue to access the user interface, and then verify the deployment status across all regions.

Run the following command:

Here is the sample output.

After a successful upgrade and verification, Private Cloud Director it does not support rolling back to a previous version.

Upgrade the Hosts

After the management plane is successfully upgraded, the hosts running in each of your regions must be updated. The agents running on the Host manage communication between hosts and the management plane.

The Impact from Hosts Upgrade

  • No servHostavailability is expected during the host upgrade.

  • Workloads running on VMs will continue to operate without disruption.

Step 1: Record the Current Version of Packages on the Host

Before you begin the upgrade, execute the following command to record the current package versions:

Step 2: Upgrade Hosts

To upgrade the hosts in a specific region, execute the following commands on the management cluster node:

NOTE

The upgrade-hosts command times out after 1800 seconds (30 minutes) by default. You can override this default by adding the hostupgradetimeout parameter in the airctl-config.yaml file. To prevent timeouts when upgrading a large number of hosts, increase this value proportionally — for every 50 hosts, add 30 minutes (1800 seconds) to the hostupgradetimeout value.

The command triggers the creation of a host-upgrade-xxxx pod in the corresponding region namespace. You can monitor the upgrade progress or verify its success by checking the pod status and logs:

Here is a sample output:

If a host fails to upgrade, run the following command to rerun the upgrade that the user executes using IP.

Step 3: Verify Upgrade Status of the Hosts

After the upgrade, confirm that the host packages have been successfully updated by running the following command.

Compare the output with the previous recording to ensure the packages were updated as expected.

Recovery: Upgrade Failure

Step 1. Stop the Failed Management Cluster

This command shuts down all components of the existing management cluster. --verbose helps verify the detailed output of all stopped services.

Step 2. Delete the Management Cluster Configuration

This command removes all existing configuration files and metadata associated with the management cluster. Using --force enables overriding any locked or incomplete states.

Step 3: Delete the Existing Management Cluster

This command permanently deletes the management cluster resources. Ensure that you have backed up all critical data before running this command

Step 4. Create a New Management Cluster

The command creates a new management cluster with the same configuration as before

Step 5. Restore from the Backup

Replace <BACKUP_DIRECTORY>with the actual path of your stored backup. The command restores the environment from the backup.

Warning

After a management plane upgrade, a rollback to a pre-upgrade version cannot be performed if any hosts have been upgraded as well.

PreviousAir Gapped InstallationNextCluster Scaling & Other Operations

Last updated

Was this helpful?