purge: gather monitors facts in OSD purge

the OSD part of the purge delegates commands on monitor node, we need to
gather monitors facts to know the `ansible_hostname` fact that is used
in the `docker_exec_cmd` fact.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
(cherry picked from commit 1a4a6ec855fcb2943d78d2a84c29a7e8c588824f)

diff --git a/infrastructure-playbooks/purge-docker-cluster.yml b/infrastructure-playbooks/purge-docker-cluster.yml
index 4786d32a8142e5a62a38df65217b42fb5385c15d..4a2f5d98c8836192d0639e7631cb082bfe94dace 100644 (file)
--- a/infrastructure-playbooks/purge-docker-cluster.yml
+++ b/infrastructure-playbooks/purge-docker-cluster.yml
@@ -303,11 +303,18 @@
 - name: purge ceph osd cluster
 
   hosts:
-    - "{{ osd_group_name|default('osds') }}"
+    - "{{ osd_group_name | default('osds') }}"
 
   gather_facts: true
   become: true
 
+  pre_tasks:
+  - name: gather monitors facts
+    setup:
+    delegate_to: "{{ item }}"
+    delegate_facts: True
+    with_items: "{{ groups[mon_group_name] | default('mons') }}"
+
   tasks:
 
   - import_role: