Merge pull request #1531 from ceph/wip-1495

docker: Retry OSD disk prepare to workaround race condition

roles/ceph-osd/tasks/docker/start_docker_osd.yml

@@ -21,10 +21,12 @@
 # starting the next task
 - name: prepare ceph osd disk
   shell: |
+    NAME="ceph-osd-prepare-{{ ansible_hostname }}-dev{{ item.0 | regex_replace('/', '') }}"
+    docker ps -a | grep -sq $NAME && docker rm $NAME || true
     docker run --net=host \
     --pid=host \
     --privileged=true \
-    --name="ceph-osd-prepare-{{ ansible_hostname }}-dev{{ item.0 | regex_replace('/', '') }}" \
+    --name=$NAME \
     -v /etc/ceph:/etc/ceph \
     -v /var/lib/ceph/:/var/lib/ceph/ \
     -v /dev:/dev \
@@ -41,13 +43,20 @@
     - item.1.get("rc", 0) != 0
     - ceph_osd_docker_prepare_env is defined
     - not osd_containerized_deployment_with_kv
+  # https://github.com/ceph/ceph-ansible/issues/1495
+  register: prepare_osd_disk
+  until: prepare_osd_disk.rc == 0
+  retries: 5
+  delay: 3
 
 - name: prepare ceph osd disk with kv_store
   shell: |
+    NAME="ceph-osd-prepare-{{ ansible_hostname }}-dev-{{ item.0 | regex_replace('/', '') }}"
+    docker ps -a | grep -sq $NAME && docker rm $NAME || true
     docker run  --net=host   \
     --pid=host \
     --privileged=true \
-    --name="ceph-osd-prepare-{{ ansible_hostname }}-dev-{{ item.0 | regex_replace('/', '') }}" \
+    --name=$NAME \
     -v /dev:/dev \
     -v /etc/localtime:/etc/localtime:ro \
     -e "OSD_DEVICE={{ item.0 }}" \
@@ -66,6 +75,11 @@
     - item.1.get("rc", 0) != 0
     - ceph_osd_docker_prepare_env is defined
     - osd_containerized_deployment_with_kv
+  # https://github.com/ceph/ceph-ansible/issues/1495
+  register: prepare_osd_disk_kv
+  until: prepare_osd_disk_kv.rc == 0
+  retries: 5
+  delay: 3
 
 - name: generate ceph osd docker run script
   become: true