Requirements¶
Cluster¶
- A Talos Linux cluster with the Talos API reachable from the controller.
- A namespace for the controller (the examples use
system-upgrade).
Talos API access¶
The controller upgrades nodes over the Talos API, so the namespace it runs in
must be granted os:admin access. Apply this machine config to every node
(control plane and workers):
machine:
features:
kubernetesTalosAPIAccess:
enabled: true
allowedRoles:
- os:admin
allowedKubernetesNamespaces:
- system-upgrade # the namespace tuppr is installed into
Namespace must match
allowedKubernetesNamespaces must list the exact namespace tuppr runs in. If it
does not, upgrade jobs cannot authenticate to the Talos API and every upgrade
fails at the first node.
Versions¶
- tuppr upgrades Talos to exactly the version you set in a
TalosUpgrade; it does not enforce Talos's supported upgrade path. Step through minor versions yourself - see Versioning and safe upgrade paths. - The upgrade Job uses
talosctl; its version is auto-detected but can be pinned per resource (spec.talosctl.image).
Next¶
Install the chart and run your first upgrade in the Quickstart.