fix(patching): add debconf non-interactive pre-seeding, cordon/drain, and storage settling to weekly OS patching

This commit is contained in:
2026-11-03 15:59:08 -05:00
parent 42f669266e
commit 4f0fa3fe55

View File

@@ -5,10 +5,46 @@
gather_facts: true
tasks:
- name: Pre-seed debconf for non-interactive kernel and package upgrades
ansible.builtin.shell: |
echo "debconf debconf/frontend select Noninteractive" | debconf-set-selections
echo "linux-base linux-base/removing-running-kernel boolean false" | debconf-set-selections
when: ansible_os_family == "Debian"
changed_when: false
- name: Ensure critical storage prerequisites are installed
ansible.builtin.apt:
name:
- open-iscsi
- nfs-common
state: present
update_cache: false
when: ansible_os_family == "Debian" and node_role is defined and node_role in ['master', 'worker']
- name: Cordon Kubernetes node if applicable
ansible.builtin.command: kubectl cordon {{ inventory_hostname }}
delegate_to: "{{ 'sninfr2lnxctl01' if cluster == 'aerodrome' else 'sninfr0lnxctl01' }}"
when: node_role is defined and node_role == 'worker'
changed_when: true
- name: Drain Kubernetes node if applicable
ansible.builtin.command: >
kubectl drain {{ inventory_hostname }}
--ignore-daemonsets
--delete-emptydir-data
--force
--timeout=180s
delegate_to: "{{ 'sninfr2lnxctl01' if cluster == 'aerodrome' else 'sninfr0lnxctl01' }}"
when: node_role is defined and node_role == 'worker'
failed_when: false
changed_when: true
- name: Update apt cache
ansible.builtin.apt:
update_cache: true
cache_valid_time: 3600
environment:
DEBIAN_FRONTEND: noninteractive
when: ansible_os_family == "Debian"
- name: Perform dist-upgrade
@@ -16,6 +52,9 @@
upgrade: dist
autoremove: true
autoclean: true
dpkg_options: 'force-confdef,force-confold'
environment:
DEBIAN_FRONTEND: noninteractive
when: ansible_os_family == "Debian"
- name: Check if system reboot is required
@@ -28,16 +67,18 @@
block:
- name: Reboot host gracefully
ansible.builtin.reboot:
msg: "Automated rolling maintenance reboot via AWX"
msg: "Automated rolling maintenance reboot via Snarfnet Automation"
reboot_timeout: 300
connect_timeout: 30
test_command: uptime
- name: Verify systemd-timesyncd is active after reboot
ansible.builtin.systemd:
name: systemd-timesyncd
state: started
ignore_errors: true
- name: Verify systemd-timesyncd or chrony is active after reboot
ansible.builtin.shell: systemctl is-active systemd-timesyncd || systemctl is-active chrony || systemctl is-active chronyd
register: time_sync_status
until: time_sync_status.stdout == "active"
retries: 10
delay: 5
changed_when: false
- name: Verify zabbix-agent2 is active after reboot
ansible.builtin.systemd:
@@ -45,3 +86,25 @@
state: started
ignore_errors: true
- name: Uncordon Kubernetes node if applicable
ansible.builtin.command: kubectl uncordon {{ inventory_hostname }}
delegate_to: "{{ 'sninfr2lnxctl01' if cluster == 'aerodrome' else 'sninfr0lnxctl01' }}"
when: node_role is defined and node_role == 'worker'
changed_when: true
- name: Wait for Kubernetes node to report Ready
ansible.builtin.command: >
kubectl get node {{ inventory_hostname }}
-o jsonpath='{.status.conditions[?(@.type=="Ready")].status}'
delegate_to: "{{ 'sninfr2lnxctl01' if cluster == 'aerodrome' else 'sninfr0lnxctl01' }}"
register: node_ready
until: node_ready.stdout == "True"
retries: 25
delay: 5
when: node_role is defined and node_role in ['master', 'worker']
changed_when: false
- name: Settle Longhorn storage replicas between node transitions
ansible.builtin.pause:
seconds: 15
when: node_role is defined and node_role in ['master', 'worker']