feat: split the playbook into 3 plays

ansible/proxmox/update_proxmox.yml

@@ -1,68 +1,68 @@
 ---
-- name: Rolling update Proxmox VE cluster
+- name: Check Proxmox VE cluster health
   hosts: nodes
   any_errors_fatal: true
   become: true
-  serial: 1
+  tasks:
-  vars:
-    ntfy_password: "{{ lookup('env', 'NTFY_PASSWORD') }}"
-
-  pre_tasks:
-
-    ##########################################################
-    # Verify input variables
-    ##########################################################
-
-    - name: Assert ntfy variables
-      ansible.builtin.assert:
-        that:
-          - ntfy_topic is defined
-          - ntfy_user is defined
-          - ntfy_password != ""
-        quiet: true
-      when: ntfy_url is defined
 
     ##########################################################
     # Verify cluster health
     ##########################################################
 
-    - name: Verify quorum
+    - delegate_to: "{{ groups['nodes'][0] }}"
-      ansible.builtin.command: pvecm status
-      register: quorum_status
-      changed_when: false
-      delegate_to: "{{ groups['nodes'][0] }}"
       run_once: true
+      block:
 
-    - name: Abort if cluster has no quorum
+        - name: Verify cluster quorum
-      ansible.builtin.fail:
+          ansible.builtin.command: pvecm status
-        msg: "Cluster quorum not available"
+          register: quorum_status
-      when: quorum_status.stdout is not search('Quorate:\\s*Yes')
+          changed_when: false
+          failed_when: quorum_status.stdout is not search('Quorate:\\s*Yes')
+          
+        - name: Verify Ceph health
+          ansible.builtin.command: ceph health
+          register: ceph_health
+          changed_when: false
+          failed_when: "'HEALTH_OK' not in ceph_health.stdout"
 
-    - name: Check Ceph health
+        ##########################################################
-      ansible.builtin.command: ceph health
+        # Capture current versions
-      register: ceph_health
+        ##########################################################
-      changed_when: false
-      delegate_to: "{{ groups['nodes'][0] }}"
-      run_once: true
 
-    - name: Abort if Ceph unhealthy
+        - name: Capture version before upgrade
-      ansible.builtin.fail:
-        msg: "{{ ceph_health.stdout }}"
-      when: "'HEALTH_OK' not in ceph_health.stdout"
-
-  tasks:
-
-    - block:
-
-    ##########################################################
-    # Capture current versions
-    ##########################################################
-
-        - name: Capture version after upgrade
           ansible.builtin.shell: pveversion | awk -F'/' '{print $2}'
           register: pve_current_version
           changed_when: false
 
+      rescue:
+
+        - name: Send no ready notification
+          ansible.builtin.uri:
+            url: "{{  ntfy_url }}/{{ ntfy_topic }}"
+            method: POST
+            user: "{{ ntfy_user }}"
+            password: "{{ lookup('env', 'NTFY_PASSWORD') }}"
+            force_basic_auth: true
+            body: Proxmox cluster not ready for updates
+            headers:
+              Title: "Proxmox cluster update failed"
+              Priority: "5"
+              Tags: "x"
+          delegate_to: localhost
+          become: false
+          run_once: true
+          when: ntfy_url is defined
+
+
+- name: Rolling update of Proxmox VE cluster
+  hosts: nodes
+  serial: 1
+  any_errors_fatal: true
+  become: true
+  tasks:
+
+    - block:
+
     ##########################################################
     # Enable maintenance mode
     ##########################################################
@@ -75,15 +75,6 @@
     # Wait until node empty
     ##########################################################
 
-        - name: Wait for VMs to leave node
-          ansible.builtin.shell: |
-            qm list | awk 'NR>1 && $3=="running" {count++} END {print count+0}'
-          register: vm_count
-          changed_when: false
-          until: vm_count.stdout | int == 0
-          retries: 60
-          delay: 15
-
         - name: Wait for LXCs to leave node
           ansible.builtin.shell: |
             pct list | awk 'NR>1 && $2=="running" {count++} END {print count+0}'
@@ -93,6 +84,15 @@
           retries: 60
           delay: 15
 
+        - name: Wait for VMs to leave node
+          ansible.builtin.shell: |
+            qm list | awk 'NR>1 && $3=="running" {count++} END {print count+0}'
+          register: vm_count
+          changed_when: false
+          until: vm_count.stdout | int == 0
+          retries: 60
+          delay: 15
+
     ##########################################################
     # Upgrade packages
     ##########################################################
@@ -101,7 +101,7 @@
           ansible.builtin.apt:
             update_cache: true
 
-        - name: Dist upgrade
+        - name: Update packages
           ansible.builtin.apt:
             upgrade: dist
             autoremove: true
@@ -114,6 +114,7 @@
         - name: Capture version after upgrade
           ansible.builtin.shell: pveversion | awk -F'/' '{print $2}'
           register: pve_new_version
+          delegate_to: "{{ groups['nodes'][0] }}"
           changed_when: false
 
     ##########################################################
@@ -147,7 +148,7 @@
     # Ceph validation
     ##########################################################
 
-        - name: Wait for HEALTH_OK
+        - name: Wait for Ceph to be healthy
           ansible.builtin.command: ceph health
           register: ceph_status
           changed_when: false
@@ -167,11 +168,11 @@
             url: "{{  ntfy_url }}/{{ ntfy_topic }}"
             method: POST
             user: "{{ ntfy_user }}"
-            password: "{{ ntfy_password }}"
+            password: "{{ lookup('env', 'NTFY_PASSWORD') }}"
             force_basic_auth: true
             body: Update failed on {{ inventory_hostname_short }}
             headers:
-              Title: "Proxmox cluster update failed"
+              Title: "Proxmox Update Failed"
               Priority: "5"
               Tags: "x"
           delegate_to: localhost
@@ -182,25 +183,27 @@
         - ansible.builtin.fail:
             msg: "Update aborted"
 
-  post_tasks:
+
+- name: Send notification
+  hosts: localhost
+  tasks:
 
     - name: Send success notification
       ansible.builtin.uri:
         url: "{{  ntfy_url }}/{{ ntfy_topic }}"
         method: POST
         user: "{{ ntfy_user }}"
-        password: "{{ ntfy_password }}"
+        password: "{{ lookup('env', 'NTFY_PASSWORD') }}"
         force_basic_auth: true
         body: |
           Proxmox cluster update completed successfully.
+          {% if hostvars[groups['nodes'][0]].pve_current_version.stdout == hostvars[groups['nodes'][0]].pve_new_version.stdout %}
           Old version: {{ pve_current_version.stdout }}
           New version: {{ pve_new_version.stdout }}
         headers:
-          Title: "Proxmox Update"
+          Title: "Proxmox Update Completed"
           Priority: "3"
           Tags: "white_check_mark"
       delegate_to: localhost
-      become: false
-      run_once: true
       when: ntfy_url is defined