fix: control plane start retries

.github/workflows/ci.yml

@@ -41,11 +41,10 @@ jobs:
           - distro: geerlingguy/docker-fedora34-ansible:latest
             scenario: highavailabilitydb
             prebuilt: 'true'
-          - distro: geerlingguy/docker-fedora33-ansible:latest
+          - distro: geerlingguy/docker-debian10-ansible:latest
             scenario: autodeploy
           - distro: xanmanning/docker-alpine-ansible:3.16
             scenario: highavailabilityetcd
-            command: "init"
             prebuilt: 'false'
           - distro: geerlingguy/docker-rockylinux8-ansible:latest
             scenario: highavailabilityetcd

tasks/ensure_cluster.yml

@@ -46,10 +46,36 @@
     dest: "{{ k3s_systemd_unit_dir }}/k3s.service"
     mode: 0644
   become: "{{ k3s_become }}"
+  when:
+    - k3s_service_handler[ansible_service_mgr] == 'systemd'
   notify:
     - "reload {{ k3s_service_handler[ansible_service_mgr] }}"
     - "restart k3s {{ k3s_service_handler[ansible_service_mgr] }}"
 
+- name: Ensure k3s service file is present
+  ansible.builtin.template:
+    src: k3s.openrc.j2
+    dest: "{{ k3s_openrc_service_dir }}/k3s"
+    mode: 0744
+  when:
+    - k3s_service_handler[ansible_service_mgr] == 'service'
+  notify:
+    - "reload {{ k3s_service_handler[ansible_service_mgr] }}"
+    - "restart k3s {{ k3s_service_handler[ansible_service_mgr] }}"
+  become: "{{ k3s_become }}"
+
+- name: Ensure k3s logrotate file is present
+  ansible.builtin.template:
+    src: k3s.logrotate.j2
+    dest: "{{ k3s_logrotate_dir }}/k3s"
+    mode: 0640
+  when:
+    - k3s_service_handler[ansible_service_mgr] == 'service'
+  notify:
+    - "reload {{ k3s_service_handler[ansible_service_mgr] }}"
+    - "restart k3s {{ k3s_service_handler[ansible_service_mgr] }}"
+  become: "{{ k3s_become }}"
+
 - name: Ensure k3s config file exists
   ansible.builtin.template:
     src: config.yaml.j2

tasks/ensure_control_plane_started_openrc.yml

@@ -1,15 +1,15 @@
 ---
 
-- name: Ensure k3s initial control plane server is started
+- name: Ensure k3s control plane server is started
   ansible.builtin.service:
     name: k3s
     state: started
     enabled: "{{ k3s_start_on_boot }}"
   register: k3s_service_start_k3s
-  failed_when:
-    - k3s_service_start_k3s is not succeeded
-    - not ansible_check_mode
-  when: (k3s_control_node and k3s_controller_list | length == 1)
-        or (k3s_primary_control_node and k3s_controller_list | length > 1)
-        or k3s_token_cluster_check.stat.exists
+  until: k3s_service_start_k3s is succeeded
+  retries: 3
+  delay: 3
+  # failed_when:
+  #   - k3s_service_start_k3s is not succeeded
+  #   - not ansible_check_mode
   become: "{{ k3s_become }}"

tasks/ensure_control_plane_started_systemd.yml

@@ -1,16 +1,16 @@
 ---
 
-- name: Ensure k3s initial control plane server is started
+- name: Ensure k3s control plane server is started
   ansible.builtin.systemd:
     name: k3s
     state: started
     enabled: "{{ k3s_start_on_boot }}"
     scope: "{{ k3s_systemd_context }}"
   register: k3s_systemd_start_k3s
+  until: k3s_systemd_start_k3s is succeeded
+  retries: 3
+  delay: 3
   failed_when:
     - k3s_systemd_start_k3s is not succeeded
     - not ansible_check_mode
-  when: (k3s_control_node and k3s_controller_list | length == 1)
-        or (k3s_primary_control_node and k3s_controller_list | length > 1)
-        or k3s_token_cluster_check.stat.exists
   become: "{{ k3s_become }}"

tasks/ensure_installed.yml

@@ -23,3 +23,6 @@
   register: k3s_token_cluster_check
 
 - include_tasks: ensure_control_plane_started_{{ ansible_service_mgr }}.yml
+  when: (k3s_control_node and k3s_controller_list | length == 1)
+        or (k3s_primary_control_node and k3s_controller_list | length > 1)
+        or k3s_token_cluster_check.stat.exists

tasks/post_checks_nodes.yml

@@ -4,8 +4,9 @@
   ansible.builtin.command:
     cmd: "{{ k3s_install_dir }}/kubectl get nodes"
   changed_when: false
-  failed_when: kubectl_get_nodes_result.stdout.find("was refused") != -1 or
-               kubectl_get_nodes_result.stdout.find("ServiceUnavailable") != -1
+  failed_when: >-
+    kubectl_get_nodes_result.stdout.find("was refused") != -1 or
+    kubectl_get_nodes_result.stdout.find("ServiceUnavailable") != -1
   register: kubectl_get_nodes_result
   until:
     - kubectl_get_nodes_result.rc == 0