Use a variable for standardizing kubectl invocation (#8329)

* Add kubectl variable

* Replace kubectl usage by kubectl variable in roles

* Remove redundant --kubeconfig on kubectl usage

* Replace unecessary shell usage with command

roles/upgrade/post-upgrade/tasks/main.yml

@@ -4,7 +4,7 @@
     - needs_cordoning|default(false)
     - kube_network_plugin == 'cilium'
   command: >
-    {{ bin_dir }}/kubectl --kubeconfig {{ kube_config_dir }}/admin.conf
+    {{ kubectl }}
     wait pod -n kube-system -l k8s-app=cilium
     --field-selector 'spec.nodeName=={{ kube_override_hostname|default(inventory_hostname) }}'
     --for=condition=Ready
@@ -12,7 +12,7 @@
   delegate_to: "{{ groups['kube_control_plane'][0] }}"
 
 - name: Uncordon node
-  command: "{{ bin_dir }}/kubectl --kubeconfig {{ kube_config_dir }}/admin.conf uncordon {{ kube_override_hostname|default(inventory_hostname) }}"
+  command: "{{ kubectl }} uncordon {{ kube_override_hostname|default(inventory_hostname) }}"
   delegate_to: "{{ groups['kube_control_plane'][0] }}"
   when:
     - needs_cordoning|default(false)

roles/upgrade/pre-upgrade/tasks/main.yml

@@ -17,9 +17,9 @@
 # Node Ready: type = ready, status = True
 # Node NotReady: type = ready, status = Unknown
 - name: See if node is in ready state
-  shell: >-
-    {{ bin_dir }}/kubectl get node {{ kube_override_hostname|default(inventory_hostname) }}
-    -o jsonpath='{ range .status.conditions[?(@.type == "Ready")].status }{ @ }{ end }'
+  command: >
+    {{ kubectl }} get node {{ kube_override_hostname|default(inventory_hostname) }}
+    -o jsonpath={ range .status.conditions[?(@.type == "Ready")].status }{ @ }{ end }
   register: kubectl_node_ready
   delegate_to: "{{ groups['kube_control_plane'][0] }}"
   failed_when: false
@@ -28,9 +28,9 @@
 # SchedulingDisabled: unschedulable = true
 # else unschedulable key doesn't exist
 - name: See if node is schedulable
-  shell: >-
-    {{ bin_dir }}/kubectl get node {{ kube_override_hostname|default(inventory_hostname) }}
-    -o jsonpath='{ .spec.unschedulable }'
+  command: >
+    {{ kubectl }} get node {{ kube_override_hostname|default(inventory_hostname) }}
+    -o jsonpath={ .spec.unschedulable }
   register: kubectl_node_schedulable
   delegate_to: "{{ groups['kube_control_plane'][0] }}"
   failed_when: false
@@ -48,11 +48,11 @@
 - name: Node draining
   block:
     - name: Cordon node
-      command: "{{ bin_dir }}/kubectl cordon {{ kube_override_hostname|default(inventory_hostname) }}"
+      command: "{{ kubectl }} cordon {{ kube_override_hostname|default(inventory_hostname) }}"
       delegate_to: "{{ groups['kube_control_plane'][0] }}"
 
     - name: Check kubectl version
-      command: "{{ bin_dir }}/kubectl version --client --short"
+      command: "{{ kubectl }} version --client --short"
       register: kubectl_version
       delegate_to: "{{ groups['kube_control_plane'][0] }}"
       run_once: yes
@@ -70,7 +70,7 @@
 
     - name: Drain node
       command: >-
-        {{ bin_dir }}/kubectl drain
+        {{ kubectl }} drain
         --force
         --ignore-daemonsets
         --grace-period {{ hostvars['localhost']['drain_grace_period_after_failure'] | default(drain_grace_period) }}
@@ -98,7 +98,7 @@
 
         - name: Drain node - fallback with disabled eviction
           command: >-
-            {{ bin_dir }}/kubectl drain
+            {{ kubectl }} drain
             --force
             --ignore-daemonsets
             --grace-period {{ drain_fallback_grace_period }}
@@ -117,7 +117,7 @@
 
   rescue:
     - name: Set node back to schedulable
-      command: "{{ bin_dir }}/kubectl --kubeconfig {{ kube_config_dir }}/admin.conf uncordon {{ inventory_hostname }}"
+      command: "{{ kubectl }} uncordon {{ inventory_hostname }}"
       when: upgrade_node_uncordon_after_drain_failure
     - name: Fail after rescue
       fail: