fix: install pinned cloud-init for only ubuntu 20.04 (#943)

ansible/group_vars/all/defaults.yaml

@@ -108,3 +108,5 @@ nvidia_driver_version: "470.199.02"
 nvidia_runfile_installer: "NVIDIA-Linux-x86_64-{{ nvidia_driver_version }}.run"
 nvidia_runfile_installer_url: "https://download.nvidia.com/XFree86/Linux-x86_64/{{ nvidia_driver_version }}/{{ nvidia_runfile_installer }}"
 suse_packagehub_product: PackageHub/{{ ansible_distribution_version }}/{{ ansible_architecture }}
+
+pinned_debs: []

ansible/roles/packages/tasks/debian.yaml

@@ -31,30 +31,17 @@
   retries: 3
   delay: 3
 
-# The latest cloud-init version '23.3.1-0ubuntu1~20.04.1 is unable to run #boothook created by CAPA
-# https://github.com/kubernetes-sigs/cluster-api-provider-aws/blob/0bf78b04b305a77aec37a68c107102231faa7a16/pkg/cloud/services/secretsmanager/secret_fetch_script.go#L20
-# This is a workaround to downgrade to older cloud-init version. 
-# Once the fix is available in cloud-init and base ubuntu AMI are built with the fixed cloud-init,
-# we can revert to using cloud-init version provided by the base AMI. 
-- name: Install cloud-init version 23.1.2-0ubuntu0~20.04.2
+- name: install pinned debs
   apt:
-    name: cloud-init=23.1.2-0ubuntu0~20.04.2
+    force_apt_get: True
+    name: "{{ item }}"
     state: present
-    force_apt_get: true
-    allow_downgrade: true
-  when: ansible_os_family == "Debian"
-
-- name: Install cloud-init packages
-  apt:
-    name: "{{ packages }}"
-    state: present
-    force_apt_get: true
-  vars:
-    packages:
-      - cloud-guest-utils
-      - cloud-initramfs-copymods
-      - cloud-initramfs-dyn-netconf
-      - cloud-initramfs-growroot
+    force: yes
+  register: apt_lock_status
+  until: apt_lock_status is not failed
+  retries: 5
+  delay: 10
+  with_items: "{{pinned_debs}}"
   when: ansible_os_family == "Debian"
 
 - name: remove version hold for kubelet and kubectl packages

images/ami/ubuntu-2004.yaml

@@ -12,3 +12,8 @@ packer:
 build_name: "ubuntu-20"
 packer_builder_type: "amazon"
 python_path: ""
+# The latest cloud-init version '23.3.1-0ubuntu1~20.04.1 is unable to run #boothook created by CAPA
+# https://github.com/kubernetes-sigs/cluster-api-provider-aws/blob/0bf78b04b305a77aec37a68c107102231faa7a16/pkg/cloud/services/secretsmanager/secret_fetch_script.go#L20
+# This is a workaround to downgrade to older cloud-init version.
+# TODO: remove it when base ubuntu AMI has fixed version of cloud-init available. https://d2iq.atlassian.net/browse/D2IQ-99637
+pinned_debs: ["cloud-init=23.1.2-0ubuntu0~20.04.2"]