1. Compare maxUnavailable with powered off hosts before attempting to reconcile
NotReady hosts
2. Include powered off hosts as failed hosts for controlplane upgrade to return error
3. Change coredns upgrade strategy. With addons changes it was changed to have the k8s
default value for a deployment of 25% maxUnavailable and maxSurge. This commit changes it
back to maxUnavailable of 1 to avoid dns addon upgrade issues
RKE does a cluster scan to find the unreachable hosts, and if that number
is same as or exceeds maxUnavailable, upgrade won't proceed.
This commit introduces a label users can provide for their nodes so they
don't get counted as unavailable and are excluded from upgrade.
This commit also includes a couple of bug fixes
- Put Windows worker plan generating back to reduce the changing from
Windows on rancher/rancher
- Prepare for rke bootstraps Windows cluster
**Issue:**
https://github.com/rancher/rancher/issues/16460
- Add `flannel_backend_port` and `canal_flannel_backend_port` to config
the port of Flannel
- Add `flanneld_backend_vni` and `canal_flannel_backend_vni` to config
the VxLan network identify of Flannel
Allow multiple authn strategies to be defined, including new 'webhook'
strategy. Webhook strategy configuration contains the contents of the
authentication webhook file as well as the cache timeout period.
This change allows a Kubernetes API Auth service to authenticate
user requests without proxying through the Rancher server.
1. Support to configure Flannel as "host-gw" backend
2. Define the network component yaml and ingress
controller yaml only schedule to non-Windows node
3. Support to configure Docker container's port publishing
