[etcd] ensure etcd is properly upgraded when managed by kubeadm (#8722)

* [etcd] ensure etcd is properly upgraded when managed by kubeadm

* [CI] add periodic job to test upgrade of etcd managed by kubeadm

.gitlab-ci/packet.yml

@@ -263,6 +263,13 @@ packet_centos7-docker-weave-upgrade-ha:
   variables:
     UPGRADE_TEST: basic
 
+packet_ubuntu20-calico-etcd-kubeadm-upgrade-ha:
+  stage: deploy-part3
+  extends: .packet_periodic
+  when: on_success
+  variables:
+    UPGRADE_TEST: basic
+
 # Calico HA Wireguard
 packet_ubuntu20-calico-ha-wireguard:
   stage: deploy-part2

roles/kubernetes/control-plane/tasks/kubeadm-upgrade.yml

@@ -18,7 +18,7 @@
     --config={{ kube_config_dir }}/kubeadm-config.yaml
     --ignore-preflight-errors=all
     --allow-experimental-upgrades
-    --etcd-upgrade={{ etcd_deployment_type == "kubeadm" | bool | lower }}
+    --etcd-upgrade={{ (etcd_deployment_type == "kubeadm") | bool | lower }}
     --force
   register: kubeadm_upgrade
   # Retry is because upload config sometimes fails
@@ -39,7 +39,7 @@
     --config={{ kube_config_dir }}/kubeadm-config.yaml
     --ignore-preflight-errors=all
     --allow-experimental-upgrades
-    --etcd-upgrade={{ etcd_deployment_type == "kubeadm" | bool | lower }}
+    --etcd-upgrade={{ (etcd_deployment_type == "kubeadm") | bool | lower }}
     --force
   register: kubeadm_upgrade
   when: inventory_hostname != first_kube_control_plane

tests/files/packet_ubuntu20-calico-etcd-kubeadm-upgrade-ha.yml

+---
+# Instance settings
+cloud_image: ubuntu-2004
+mode: ha
+
+# use the legacy setting to test the upgrade
+etcd_kubeadm_enabled: true
+
+upgrade_cluster_setup: true
+
+# Currently ipvs not available on KVM: https://packages.ubuntu.com/search?suite=focal&arch=amd64&mode=exactfilename&searchon=contents&keywords=ip_vs_sh.ko
+kube_proxy_mode: iptables
+enable_nodelocaldns: False