Updated homelab hosts & structure

2026-08-14 00:17:50 +02:00
parent bb8b9602f2
commit 097ee79bed
23 changed files with 506 additions and 0 deletions
@@ -0,0 +1,93 @@
+---
+# Ensure Ceph is healthy, before starting to patch
+- name: Ensure Ceph is HEALTH_OK and no rebalancing
+  shell: |
+    ceph status | grep 'HEALTH_OK' && ! ceph status | grep -q -e 'degraded' -e 'remapped'
+  register: ceph_health
+  failed_when: ceph_health.rc != 0
+  delegate_to: "{{ play_hosts[0] }}"
+# Ensure no node is in Maintenance Mode before patching
+- name: Ensure no node is in mainteance mode
+  shell: ha-manager status | grep '^lrm ' | grep -e maintenance | wc -l
+  register: no_node_maintenance
+  failed_when: no_node_maintenance.stdout|int != 0
+  delegate_to: "{{ play_hosts[0] }}"
+# Check that all VMs are in a HA Group, so they will be migrated when the host gets set to maintenance Mode
+- name: Ensure all VMs are in a HA Group
+  shell: |
+    for res in $(pvesh get /cluster/resources -type vm --output-format json|jq '.[]|select(has("hastate")|not)|.vmid'); do ha-manager add $res ;done
+
+- block:
+    # Set the noout flag to ensure no data on ceph is moved, while a host is unavailable
+    - name: Set Ceph noout flag (only first run)
+      shell: ceph osd set noout
+    # Enable the maintenance Mode of the Host
+    - name: Enable maintenance mode and migrate VMs
+      shell:  ha-manager crm-command node-maintenance enable {{ inventory_hostname }}
+      ignore_errors: false
+
+    # Wait till all resources are migrate away 
+    - name: Wait until no VMs are running on this host
+      shell: |
+        qm list | grep running | wc -l
+      register: vm_count
+      changed_when: false
+      retries: "{{ long_task.retries }}"
+      delay: "{{ long_task.delay }}"
+      until: vm_count.stdout|int == 0
+    - name: Wait until no LXCs are running on this host
+      shell: |
+        pct list | grep running | wc -l
+      register: lxc_count
+      changed_when: false
+      retries: "{{ long_task.retries }}"
+      delay: "{{ long_task.delay }}"
+      until: lxc_count.stdout|int == 0
+
+    # Install latest Packages
+    - name: Update apt packages
+      apt:
+        update_cache: yes
+        upgrade: dist
+        autoremove: yes
+
+    # Compare if the current running kernel is the same as the latest kernel
+    - name: Get expected boot kernel version
+      shell: |
+        grep vmlinuz /boot/grub/grub.cfg | head -1 | awk '{ print $2 }' | sed -e 's%/boot/vmlinuz-%%;s%/ROOT/pve-1@%%'
+      register: boot_kernel
+      changed_when: false
+    - name: Get currently running kernel version
+      shell: uname -r
+      register: running_kernel
+      changed_when: false
+    # If a newer kernel is on the system, then reboot
+    - name: Reboot if required
+      reboot:
+        reboot_timeout: 600
+        post_reboot_delay: 60
+      when: boot_kernel.stdout.strip() != running_kernel.stdout.strip()
+
+
+    # Disable Ceph noout, as soon as the host is online
+    - name: Unset Ceph noout flag
+      shell: ceph osd unset noout
+    
+    - name: Ensure Ceph is HEALTH_OK again
+      shell: |
+        ceph status | grep 'HEALTH_OK' && ! ceph status | grep -q -e 'degraded' -e 'remapped'
+      register: ceph_health
+      retries: "{{ long_task.retries }}"
+      delay: "{{ long_task.delay }}"
+      failed_when: ceph_health.rc != 0
+      delegate_to: "{{ play_hosts[0] }}"
+
+    - name: Disable maintenance mode and migrate VMs
+      shell:  ha-manager crm-command node-maintenance disable {{ inventory_hostname }}
+      ignore_errors: false
+    - name: Ensure no node is in mainteance mode
+      shell: ha-manager status | grep '^lrm ' | grep -e maintenance | wc -l
+      register: no_node_maintenance
+      failed_when: no_node_maintenance.stdout|int != 0
+      retries: "{{ long_task.retries }}"
+      delay: "{{ long_task.delay }}"