Update Ansible to 2.8.2, Ceph-Ansible to 4.0, Ceph to Nautilus

This patch updates to the current version of Ansible in the 2.8 series,
2.8.2.

Ceph-Ansible must be moved to version 4.0 to be compatible with Ansible
2.8 [1]. Older versions of Ceph-Ansible use syntax which is invalid for
Ansible 2.8.

In turn, Ceph-Ansible 4.0 can only install the Nautilus release of Ceph,
there is no overlap with the previous Ceph release as there has been
in the prior Ceph-Ansible 3.x series.

This patch updates provisioning of the OSA AIO ceph build to be
compatible with the ceph-volume tool [2], specifically ceph-volume
cannot prepare OSDs from loopback devices. The AIO boostrap now creates
LVM on top of the loopbacks which are then consumed by ceph-volume.

The configuration of radosgw is updated to account for the addition of
instance names to rados gateway configurations [3][4].

[1] http://docs.ceph.com/ceph-ansible/master/#releases
[2] http://docs.ceph.com/docs/nautilus/ceph-volume
[3] 367dce2894/roles/ceph-config/templates/ceph.conf.j2 (L115-L116)
[4] 367dce2894/roles/ceph-facts/tasks/facts.yml (L276-L280)

Change-Id: I7075be2b06e81e1008bd393cf0fc5619dad24e7e

ansible-role-requirements.yml

@@ -261,8 +261,8 @@
 - name: ceph-ansible
   scm: git
   src: https://github.com/ceph/ceph-ansible
-  version: stable-3.2
+  version: stable-4.0
-  trackbranch: stable-3.2
+  trackbranch: stable-4.0
 - name: opendaylight
   scm: git
   src: https://github.com/opendaylight/integration-packaging-ansible-opendaylight

inventory/group_vars/all/ceph.yml

@@ -25,7 +25,7 @@ ceph_repository: community
 # The _stable_release var is used by both the OSA ceph_client role and the
 # ceph-ansible roles. It is defaulted in ceph_client but set here to keep the
 # OSA/ceph-ansible integrations in sync.
-ceph_stable_release: mimic
+ceph_stable_release: nautilus
 fetch_directory: "{{ openstack_config_dir }}/ceph-fetch/"
 # tries to create /var/log/ceph as a directory and fails if the log link already
 # exists. we handle the log dir creation so this is not something we need

inventory/group_vars/ceph-rgw.yml

@@ -1,6 +1,6 @@
 ---
 ceph_conf_overrides_rgw:
-  "client.rgw.{{ hostvars[inventory_hostname]['ansible_hostname'] }}":
+  "client.rgw.{{ hostvars[inventory_hostname]['ansible_hostname'] }}.rgw0":
     # OpenStack integration with Keystone
     rgw_keystone_url: "{{ keystone_service_adminuri }}"
     rgw_keystone_api_version: 3
@@ -12,4 +12,4 @@ ceph_conf_overrides_rgw:
     rgw_keystone_implicit_tenants: 'true'
     rgw_enable_apis: swift
     rgw_swift_account_in_url: 'true'
-    rgw_swift_versioning_enabled: 'true'
+    rgw_swift_versioning_enabled: 'true'

releasenotes/notes/ansible-2_8-ceph-nautilus-18b0a5e14a125e26.yaml

@@ -0,0 +1,18 @@
+---
+features:
+  - |
+    The ansible version used by OSA is updated from the 2.7 to the 2.8 series.
+    This requires an upgrade of ceph-ansible to 4.0 and this in turn
+    requires an upgrade of ceph from Mimic to Nautilus. This version dependancy
+    applies where OSA uses ceph-ansible directly to deploy the ceph infrastructure,
+    but not when OSA is integrated with an externally provisioned ceph cluster.
+upgrade:
+  - |
+    Any ceph infrastructure components (OSDs, MONs etc) deployed using the
+    OSA/ceph-ansible tooling will be upgraded to the Ceph Nautilus release.
+    Deployers should verify that this upgrade is suitable for their environment
+    before commencing a major upgrade to Train, and consult the ceph-ansible
+    and ceph release notes for Nautilus. For integration with external ceph
+    clusters where OSA does not deploy any of the ceph cluster infrastructure,
+    overrides can be used to select the specific version of ceph repositories
+    used by the OSA ceph_client ansible role.

scripts/bootstrap-ansible.sh

@@ -22,7 +22,7 @@ set -e -u -x
 export HTTP_PROXY=${HTTP_PROXY:-""}
 export HTTPS_PROXY=${HTTPS_PROXY:-""}
 # The Ansible version used for testing
-export ANSIBLE_PACKAGE=${ANSIBLE_PACKAGE:-"ansible==2.7.10"}
+export ANSIBLE_PACKAGE=${ANSIBLE_PACKAGE:-"ansible==2.8.2"}
 export ANSIBLE_ROLE_FILE=${ANSIBLE_ROLE_FILE:-"ansible-role-requirements.yml"}
 export SSH_DIR=${SSH_DIR:-"/root/.ssh"}
 export DEBIAN_FRONTEND=${DEBIAN_FRONTEND:-"noninteractive"}

tests/roles/bootstrap-host/tasks/prepare_ceph.yml

@@ -62,13 +62,34 @@
     mode: "0777"
   with_items: "{{ ceph_create_loopback.results }}"
 
+# NOTE(jrosser) ceph-volume is unwilling to automatically create OSD
+# directly on loop devices - see http://tracker.ceph.com/issues/36603
+# Work around this with manual LVM creation and the advanced lvm OSD
+# scenario
+- name: Create LVM VG
+  lvg:
+    vg: "vg-{{ item.stdout | basename }}"
+    pvs: "{{ item.stdout }}"
+  loop: "{{ ceph_create_loopback.results }}"
+
+- name: Create LVM LV
+  lvol:
+    lv: "lv-{{ item.stdout | basename }}"
+    vg: "vg-{{ item.stdout | basename }}"
+    size: 100%FREE
+  loop: "{{ ceph_create_loopback.results }}"
+
 # TODO(logan): Move these vars to user_variables.ceph.yml.j2 once LP #1649381
 # is fixed and eliminate this task.
 - name: Write ceph cluster config
   copy:
     content: |
       ---
-      devices: {{ ceph_create_loopback.results | map(attribute='stdout') | list | to_yaml | trim }}
+      lvm_volumes:
+      {% for d in ceph_create_loopback | json_query('results[].stdout') %}
+        - data_vg: vg-{{ d | basename }}
+          data: lv-{{ d | basename }}
+      {% endfor %}
       cinder_backends:
         "RBD":
           volume_driver: cinder.volume.drivers.rbd.RBDDriver