The kinds of upgrade below use different tools.
What you are upgrading | Where you do it |
|---|---|
A single package, such as an optional application | Configuration Service, on the package card. See Managing Packages > Upgrading a Package. |
The VAM platform as a whole | The |
Talos Linux on the cluster nodes | The |
There is no separate upgrade tool and no upgrade workflow on the Choose Workflow step. A platform upgrade is a deployment run onto a cluster that already has VAM, told to keep what is there.
Take a backup before any platform upgrade, and confirm that the backup succeeded, see Backup and Restore. Restoring is the only supported way back.
What a Platform Upgrade Changes
An upgrade replaces the deployed VAM packages with newer versions and keeps existing cluster resources in place. It does not reprovision the cluster nodes and it does not change your platform settings, such as the hostname, certificates or identity provider.
Upgrading the Platform
-
Take a backup: in the Configuration Service, open Backup and confirm that the latest backup is recent and succeeded.
-
On the installer host, launch
vamctl.exe. -
Choose the Manage Existing Cluster workflow. This is the short path: it skips every cluster-provisioning step and goes straight to the deployment screens.
-
Sign in to the registry when the Registry Login step asks.
-
On Existing Clusters, select the cluster that runs VAM, select Redeploy apps on this cluster and click Continue to Redeploy Apps.
-
On Deployment Configuration, review the Infrastructure view and click Continue to Deployment Options.
Upgrade existing installation is clear by default. With it clear, the deployment uninstalls the existing VAM releases and deletes their data, including the stored assets, before it installs from scratch.
-
On Deployment Options, the Cluster Details card reports whether the installer found an existing installation on that cluster. When it has, make sure Upgrade existing installation is selected.
- Leave Force upgrade all components clear to upgrade only the components that have a newer version. This is the routine case.
- Select it to redeploy every component regardless of version. Use it to repair an installation, not for a routine upgrade.
-
Under Platform applications on the same view, expand an entry to see its Upgrade Available, Up to Date or New badge. The badges show what the run changes before you start it.
-
Optionally select Backup AIStor data before deployment, so the installer takes its own backup before it starts.
-
Click Start Deployment. Packages are upgraded in dependency order with live progress, and the sanity check runs afterwards as it does on a first install.
Stop if the Cluster Details card reports No existing VAM installation detected on this cluster. A fresh deployment will be performed. Deploying then performs a fresh install rather than an upgrade.
An upgrade always moves to the current versions in the registry. There is no version-selection step. An upgrade at an air-gapped site needs assistance from Vizrt support.
The Installation workflow reaches the same Deployment Configuration screen and works as well, but it walks the cluster-setup steps on the way, where you must choose the existing cluster rather than let the installer create one.
Typical duration is 10 to 30 minutes for a routine upgrade.
Rolling Back
If a package fails during a platform upgrade, the installer rolls that package back to its previous revision. To return the whole installation to an earlier state, restore a backup:
- To restore a Configuration Service backup, open Backup in the Configuration Service, find the backup and click Restore. See Backup and Restore.
- To restore an installer backup, launch
vamctl.exeand choose the Backup & Restore workflow. Select the cluster in Cluster context and select Restore (Local) in Backup / restore mode. Choose the archive in Select backup and click Start Restore. This restores object-storage data only.
A package upgrade started from the Configuration Service is different: if it fails, that package is rolled back automatically to its last known-good version.
Upgrading Talos and Kubernetes
The cluster nodes run Talos Linux. The installer upgrades Talos on the nodes, but nothing in the installer upgrades the Kubernetes version.
-
Launch
vamctl.exeand choose the Manage Existing Cluster workflow. -
On Existing Clusters, select the cluster. Under Talos maintenance, the installer compares the Talos version of the cluster with the version it supports.
GPU worker nodes lose their NVIDIA extensions, because Upgrade Talos installs the standard Talos image without them. Contact Vizrt support before you upgrade Talos on a cluster with GPU nodes.
-
When an upgrade is available, click Upgrade Talos. The installer upgrades the nodes one after the other.
Reinstall Talos resets every node of the cluster. It is destructive and asks you to type REINSTALL to confirm. Use it only when a node must be reset.
Contact Vizrt support before planning a Kubernetes version change on a running installation.
After an Upgrade
Re-run the checks in Post-Install Validation. Watch the cluster for the first 24 hours. Most problems surface within minutes, but slow leaks and background-job failures can take longer.