Tectonic and Kubernetes can be automatically updated with controls in the Tectonic Console. Clusters are attached to an update channel and are set to update either automatically, or with manual approval. This automatic update functionality is currently an experimental feature, and must be optionally enabled during Tectonic installation. Clusters installed without enabling this experimental feature will need to perform manual upgrades of Tectonic.
For Tectonic 1.5.2-tectonic.2 clusters, in order to receive the update, users need to change their channel from alpha to tectonic-1.5 first before clicking the "Check for updates" button. This can be done by running:
$ kubectl replace -f -<<EOF apiVersion: coreos.com/v1 appID: 6bc7b986-4654-4a0f-94b3-84ce6feb1db4 automaticUpdate: false channel: tectonic-1.5 kind: ChannelOperatorConfig metadata: name: default namespace: tectonic-system server: https://tectonic.update.core-os.net triggerUpdate: false triggerUpdateCheck: false updateCheckInterval: 2700 EOF
For any clusters running Tectonic v.1.5.2-tectonic.1 and older, there is no automatic upgrade path. As we move our experimental Operators to GA, we’re making upgrades between patch releases (and, eventually, minor releases) smoother and more consistent.
During an update, the latest versions of the Tectonic and Kubernetes components are downloaded. A seamless rolling update will occur to install the latest versions. A cluster admin can pause the update at any time.
To learn more about how this process is executed, read about Operators.
If you'd like to preserve and restore etcd data to the new cluster, see the etcd disaster recovery guide.