update: mask systemd service units during upgrade

This prevents the packaging from restarting services before we do need
to restart them in the rolling update sequence.
We want to handle services restart at rolling_update playbook.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

infrastructure-playbooks/rolling_update.yml

@@ -96,32 +96,39 @@
       set_fact:
         mon_host: "{{ groups[mon_group_name] | difference([inventory_hostname]) | last }}"
 
+    # NOTE: we mask the service so the RPM can't restart it
+    # after the package gets upgraded
     - name: stop ceph mon - shortname
       systemd:
         name: ceph-mon@{{ ansible_hostname }}
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       ignore_errors: True
       when:
         - not containerized_deployment
 
+    # NOTE: we mask the service so the RPM can't restart it
+    # after the package gets upgraded
     - name: stop ceph mon - fqdn
       systemd:
         name: ceph-mon@{{ ansible_fqdn }}
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       ignore_errors: True
       when:
         - not containerized_deployment
 
-    - name: stop ceph mgr
+    # only mask the service for mgr because it must be upgraded
+    # after ALL monitors, even when collocated
+    - name: mask the mgr service
       systemd:
         name: ceph-mgr@{{ ansible_hostname }}
-        state: stopped
+        masked: yes
-        enabled: yes
-      ignore_errors: True # if no mgr collocated with mons
       when:
-        - not containerized_deployment
+        - inventory_hostname in groups[mgr_group_name]
+          or groups[mgr_group_name] | length == 0
 
     - import_role:
         name: ceph-defaults
@@ -190,16 +197,6 @@
       when:
         - containerized_deployment
 
-    - name: restart containerized ceph mgr
-      systemd:
-        name: ceph-mgr@{{ monitor_name }}
-        state: restarted
-        enabled: yes
-        daemon_reload: yes
-      ignore_errors: True # if no mgr collocated with mons
-      when:
-        - containerized_deployment
-
     - name: non container | waiting for the monitor to join the quorum...
       command: ceph --cluster "{{ cluster }}" -s --format json
       register: ceph_health_raw
@@ -329,7 +326,8 @@
       systemd:
         name: ceph-mgr@{{ ansible_hostname }}
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       failed_when: false
       when:
         - not containerized_deployment
@@ -351,23 +349,6 @@
     - import_role:
         name: ceph-mgr
 
-    - name: start ceph mgr
-      systemd:
-        name: ceph-mgr@{{ ansible_hostname }}
-        state: started
-        enabled: yes
-      when:
-        - not containerized_deployment
-
-    - name: restart containerized ceph mgr
-      systemd:
-        name: ceph-mgr@{{ ansible_hostname }}
-        state: restarted
-        enabled: yes
-        daemon_reload: yes
-      when:
-        - containerized_deployment
-
 
 - name: upgrade ceph osds cluster
   vars:
@@ -396,7 +377,8 @@
       systemd:
         name: ceph-osd@{{ item }}
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       with_items: "{{ osd_ids.stdout_lines }}"
       when:
         - not containerized_deployment
@@ -429,6 +411,7 @@
         name: ceph-osd@{{ item }}
         state: started
         enabled: yes
+        masked: no
       with_items: "{{ osd_ids.stdout_lines }}"
       when:
         - not containerized_deployment
@@ -438,6 +421,7 @@
         name: "{{ item }}"
         state: restarted
         enabled: yes
+        masked: no
         daemon_reload: yes
       with_items: "{{ osd_names.stdout_lines }}"
       when:
@@ -544,7 +528,8 @@
       systemd:
         name: ceph-mds@{{ ansible_hostname }}
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       when:
         - not containerized_deployment
 
@@ -570,6 +555,7 @@
         name: ceph-mds@{{ ansible_hostname }}
         state: started
         enabled: yes
+        masked: no
       when:
         - not containerized_deployment
 
@@ -578,6 +564,7 @@
         name: ceph-mds@{{ ansible_hostname }}
         state: restarted
         enabled: yes
+        masked: no
         daemon_reload: yes
       when:
         - containerized_deployment
@@ -595,7 +582,8 @@
       systemd:
         name: ceph-radosgw@rgw.{{ ansible_hostname }}
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       when:
         - not containerized_deployment
 
@@ -621,6 +609,7 @@
         name: ceph-radosgw@rgw.{{ ansible_hostname }}
         state: started
         enabled: yes
+        masked: no
       when:
         - not containerized_deployment
 
@@ -629,6 +618,7 @@
         name: ceph-radosgw@rgw.{{ ansible_hostname }}.{{ item.instance_name }}
         state: restarted
         enabled: yes
+        masked: no
         daemon_reload: yes
       with_items: "{{ rgw_instances }}"
       when:
@@ -656,7 +646,8 @@
       systemd:
         name: "ceph-rbd-mirror@rbd-mirror.{{ ansible_hostname }}"
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       failed_when: false
 
     - import_role:
@@ -681,6 +672,7 @@
         name: "ceph-rbd-mirror@rbd-mirror.{{ ansible_hostname }}"
         state: started
         enabled: yes
+        masked: no
       when:
         - not containerized_deployment
 
@@ -689,6 +681,7 @@
         name: ceph-rbd-mirror@rbd-mirror.{{ ansible_hostname }}
         state: restarted
         enabled: yes
+        masked: no
         daemon_reload: yes
       when:
         - containerized_deployment
@@ -709,7 +702,8 @@
       systemd:
         name: nfs-ganesha
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       failed_when: false
       when:
         - not containerized_deployment
@@ -736,6 +730,7 @@
         name: nfs-ganesha
         state: started
         enabled: yes
+        masked: no
       when:
         - not containerized_deployment
         - ceph_nfs_enable_service
@@ -745,6 +740,7 @@
         name: ceph-nfs@{{ ceph_nfs_service_suffix | default(ansible_hostname) }}
         state: restarted
         enabled: yes
+        masked: no
         daemon_reload: yes
       when:
         - ceph_nfs_enable_service
@@ -767,7 +763,8 @@
       systemd:
         name: rbd-target-gw
         state: stopped
-        enabled: yes
+        enabled: no
+        masked: yes
       failed_when: false
       when:
         - not containerized_deployment
@@ -794,6 +791,7 @@
         name: rbd-target-gw
         state: started
         enabled: yes
+        masked: no
       when:
         - not containerized_deployment
 

roles/ceph-iscsi-gw/tasks/container/containerized.yml

@@ -19,6 +19,7 @@
     name: "{{ item }}"
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes
   with_items:
     - tcmu-runner

roles/ceph-iscsi-gw/tasks/non-container/prerequisites.yml

@@ -66,10 +66,12 @@
   service:
     name: rbd-target-gw
     enabled: yes
+    masked: no
     state: started
 
 - name: enable the rbd-target-api service and make sure it is running
   service:
     name: rbd-target-api
     enabled: yes
+    masked: no
     state: started

roles/ceph-mds/tasks/containerized.yml

@@ -61,6 +61,7 @@
     name: ceph-mds@{{ ansible_hostname }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes
 
 - name: wait for mds socket to exist

roles/ceph-mds/tasks/non_containerized.yml

@@ -58,4 +58,5 @@
     name: ceph-mds@{{ mds_name }}
     state: started
     enabled: yes
+    masked: no
   changed_when: false

roles/ceph-mgr/tasks/start_mgr.yml

@@ -35,4 +35,5 @@
     name: ceph-mgr@{{ ansible_hostname }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes

roles/ceph-mon/tasks/start_monitor.yml

@@ -36,4 +36,5 @@
     name: ceph-mon@{{ monitor_name if not containerized_deployment else ansible_hostname }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes

roles/ceph-nfs/tasks/start_nfs.yml

@@ -83,6 +83,7 @@
     name: ceph-nfs@{{ ceph_nfs_service_suffix | default(ansible_hostname) }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes
   when:
     - containerized_deployment
@@ -93,6 +94,7 @@
     name: nfs-ganesha
     state: started
     enabled: yes
+    masked: no
   when:
     - not containerized_deployment
     - ceph_nfs_enable_service

roles/ceph-osd/tasks/start_osds.yml

@@ -76,6 +76,7 @@
     name: ceph-osd@{{ item | regex_replace('/dev/', '') if osd_scenario != 'lvm' and containerized_deployment else item }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes
   with_items: "{{ devices if osd_scenario != 'lvm' and containerized_deployment else ((ceph_osd_ids.stdout | from_json).keys() | list) if osd_scenario == 'lvm' and not containerized_deployment else osd_ids_non_container.stdout_lines }}"
 

roles/ceph-rbd-mirror/tasks/docker/start_docker_rbd_mirror.yml

@@ -16,4 +16,5 @@
     name: ceph-rbd-mirror@rbd-mirror.{{ ansible_hostname }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes

roles/ceph-rbd-mirror/tasks/start_rbd_mirror.yml

@@ -29,6 +29,7 @@
     name: "ceph-rbd-mirror.target"
     state: started
     enabled: yes
+    masked: no
   changed_when: false
 
 - name: start and add the rbd-mirror service instance
@@ -36,4 +37,5 @@
     name: "ceph-rbd-mirror@rbd-mirror.{{ ansible_hostname }}"
     state: started
     enabled: yes
+    masked: no
   changed_when: false

roles/ceph-rgw/tasks/docker/start_docker_rgw.yml

@@ -28,6 +28,7 @@
     name: "{{ item }}"
     state: stopped
     enabled: no
+    masked: yes
     daemon_reload: yes
   with_items:
     - "ceph-rgw@{{ ansible_hostname }}"
@@ -41,5 +42,6 @@
     name: ceph-radosgw@rgw.{{ ansible_hostname }}.{{ item.instance_name }}
     state: started
     enabled: yes
+    masked: no
     daemon_reload: yes
   with_items: "{{ rgw_instances }}"

roles/ceph-rgw/tasks/start_radosgw.yml

@@ -20,6 +20,7 @@
     name: ceph-radosgw@rgw.{{ ansible_hostname }}.{{ item.instance_name }}
     state: started
     enabled: yes
+    masked: no
   with_items: "{{ rgw_instances }}"
 
 - name: enable the ceph-radosgw.target service