Ensure we always fixup kube-proxy kubeconfig (#5524) (#5558)

When running with serial != 100%, like upgrade_cluster.yml, we need to apply this fixup each time
Problem was introduced in 05dc2b3

Signed-off-by: Etienne Champetier <[email protected]>
(cherry picked from commit 5e9479c)

roles/kubernetes/kubeadm/tasks/main.yml

@@ -111,8 +111,8 @@
     | sed 's#server:.*#server: https://127.0.0.1:{{ kube_apiserver_port }}#g'
     | {{ bin_dir }}/kubectl --kubeconfig {{ kube_config_dir }}/admin.conf replace -f -
   run_once: true
+  delegate_to: "{{ groups['kube-master']|first }}"
   when:
-    - inventory_hostname == groups['kube-master']|first
     - kubeadm_config_api_fqdn is not defined
     - kubeadm_discovery_address != kube_apiserver_endpoint | replace("https://", "")
     - not kube_proxy_remove
@@ -129,8 +129,8 @@
 - name: Restart all kube-proxy pods to ensure that they load the new configmap
   shell: "{{ bin_dir }}/kubectl --kubeconfig {{ kube_config_dir }}/admin.conf delete pod -n kube-system -l k8s-app=kube-proxy --force --grace-period=0"
   run_once: true
+  delegate_to: "{{ groups['kube-master']|first }}"
   when:
-    - inventory_hostname == groups['kube-master']|first
     - kubeadm_config_api_fqdn is not defined
     - kubeadm_discovery_address != kube_apiserver_endpoint | replace("https://", "")
     - not kube_proxy_remove
@@ -153,8 +153,8 @@
 - name: Delete kube-proxy daemonset if kube_proxy_remove set, e.g. kube_network_plugin providing proxy services
   shell: "{{ bin_dir }}/kubectl --kubeconfig /etc/kubernetes/admin.conf delete daemonset -n kube-system kube-proxy"
   run_once: true
+  delegate_to: "{{ groups['kube-master']|first }}"
   when:
-    - inventory_hostname == groups['kube-master']|first
     - kube_proxy_remove
     - kubeadm_discovery_address != kube_apiserver_endpoint | replace("https://", "")
   # When scaling/adding nodes in the existing k8s cluster, kube-proxy wouldn't be created, as `kubeadm init` wouldn't run.