The new KEv2-driven GKE cluster provider will support creating "private
clusters", but this has a unique meaning for GKE that is different from
other Kubernetes providers that needs some extra explanation, and can
require pre-planning and/or manual steps during the cluster provisioning
process. This change documents the considerations needed for an advanced
setup like this.
Prior to 2.5.8, enabling Auto Upgrade for GKE hosted clusters was not
recommended because it could cause the state of the cluster in GKE to
become out of sync with the state of the cluster in Rancher. With 2.5.8,
this is not a concern any more because of the syncing mechanism in the
new GKE provisioner. Therefore, it's recommended to use the Auto Upgrade
feature, which defaults to enabled in both Rancher and the GKE console.
This change removes the note erroneously added in 74b31fae.
* Add notes about Shared VPCs
* Change "Node Subnet" to "Node Subnet / Subnet" - the field name
changes depending on whether you use Ip Aliases, but means the same
thing in both cases
* Clarify use of Subnetwork Name
* Clarify use of address ranges and range names
* Add private cluster-related fields
* Fix HTTP (L7) Load Balancing reference URL
* Add reference URL for Maintenance Window
* Add notes on image types
* Enhance Machine Type
* Use "node pool" instead of "node group"
* Fix "Autoscaling" under "Node Pools" - this has been fixed in the UI
* Add note to "Auto Upgrade" to match <2.5.8 tab
* Cluster Upgrades are possible for imported RKE2 clusters
* CIS Scans are available for all types of clusters in 2.5
Signed-off-by: Bastian Hofmann <bashofmann@gmail.com>
I had a really hard time installing Longhorn on my cluster simply because I didn't know there were installation requirements. I later found out about them here: https://longhorn.io/docs/1.1.0/deploy/install/#installation-requirements
But I wasted a whole day on this, so adding an extra step to installation, linking requirements, so others don't have to struggle the way I did.
Regards.