Check the versions skew policy
- Scale down admiralty deployments and delete all virtual nodes - those prevent the upgrades
- Make sure the etcd size fits the standard limit (2GB).
nerdctl run --rm --network host -v /etc/kubernetes/pki/etcd:/etc/kubernetes/pki/etcd:ro --env ETCDCTL_API=3 k8s.gcr.io/etcd:3.5.6-0 etcdctl --endpoints=https://127.0.0.1:2379 --cacert=/etc/kubernetes/pki/etcd/ca.crt --cert=/etc/kubernetes/pki/etcd/healthcheck-client.crt --key=/etc/kubernetes/pki/etcd/healthcheck-client.key --write-out=table endpoint status
If it doesn't, perform compaction.
- Follow the upgrade guide for master.
- During the upgrade it's better to pre-pull the new images with
kubeadm config images pull
- From backup of manifests put back the CiLogon settings to the current apiserver manifest in /etc/kubernetes/manifests,
- Bring back all custom changes to controller-manager and etcd manifests in /etc/kubernetes/manifests.
- Do the last step in the upgrade manual to restart the control place again.
-- Breathe out! The master upgrade is done. --
- Do rolling upgrade of compute nodes using ansible upgrade playbook.
- Upgrade the kubernetes version in the portal dependencies libraries