commit 943d390dd7d7d1b65f5beabd5c5f28c9221470ff
Author: Ovidiu Poncea <ovidiu.poncea@windriver.com>
Date:   Wed Oct 7 19:18:38 2020 +0300

    Fix ceph ansible restore failure when backup is taken on controller-1
    
    Ansible playbook starts ceph through '/etc/init.d/ceph start'.
    This script has two detection mechanisms for osds:
     1. it looks in the ceph.conf for \[osd\.[0-9]*\] sections, grabs
        the osd id, mounts the corresponding devices and starts ceph-osd
        daemons;
     2. it lists all folders in /var/lib/ceph/osd/*, grabs the osd id
        and starts the corresponding daemon.
    
    When backup is taken on controller-1 it contains /etc/ceph/ceph.conf
    with osds of this node. Restore is always done from controller-0
    where ansible extracts ceph.conf from the backup to
    /etc/ceph/ceph.conf. This leads to osds from controller-1 trying to
    start on controller-0 and to ansible failure.
    
    To fix it we remove the osds configuration from ceph.conf. This works
    as we have code in the restore playbook that scans the disks for osds
    and mount them in /var/lib/ceph/osd/* allowing 'etc/init.d/ceph start'
    to initialize the correct ceph-osd daemons for controller-0.
    
    Closes-Bug: 1899444
    Change-Id: I10672613fc26807e0cf28ac8df5a08287d80c17a

diff --git a/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml b/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml
index e7cdee4..6dc75b9 100644
--- a/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml
+++ b/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml
@@ -33,10 +33,24 @@
     - name: Restore ceph.conf file
       command: >-
         tar -C / -xpf {{ restore_data_file }}
-        'etc/ceph/ceph.conf'
+        '{{ ceph_conf[1:] }}'
       args:
         warn: false
 
+    - name: Get list of OSDs defined in {{ ceph_conf }}
+      shell: grep "\[osd\.[0-9]*\]" {{ ceph_conf }} | grep -oh "osd\.[0-9]*"
+      register: ceph_conf_osd_list
+      failed_when: false
+
+    # Allow starting ceph with a ceph.conf from a backup done on controller-1
+    # w/o this it will try to initialize OSDs that are not configured on controller-0.
+    - name: Remove "[osd.*]" sections from {{ ceph_conf }}
+      ini_file:
+        path: "{{ ceph_conf }}"
+        section: "{{ item }}"
+        state: absent
+      with_items: "{{ ceph_conf_osd_list.stdout_lines }}"
+
     - name: Set initial ceph-mon name
       set_fact:
         mon_name: 'controller-0'
@@ -91,7 +105,7 @@
 
     - name: Allow Ceph to start with a single monitor on a Standard deployment
       ini_file:
-        path: "/etc/ceph/ceph.conf"
+        path: "{{ ceph_conf }}"
         section: global
         option: mon_initial_members
         value: controller-0
@@ -177,7 +191,7 @@
 
     - name: Remove Ceph option to start with a single monitor on a Standard deployment
       ini_file:
-        path: "/etc/ceph/ceph.conf"
+        path: "{{ ceph_conf }}"
         section: global
         option: mon_initial_members
         state: absent
diff --git a/playbookconfig/src/playbooks/roles/recover-ceph-data/vars/main.yml b/playbookconfig/src/playbooks/roles/recover-ceph-data/vars/main.yml
index 1d9a6ea..9b19a21 100644
--- a/playbookconfig/src/playbooks/roles/recover-ceph-data/vars/main.yml
+++ b/playbookconfig/src/playbooks/roles/recover-ceph-data/vars/main.yml
@@ -3,3 +3,4 @@ ceph_mon_manifest_apply_log: /tmp/ceph_mon_apply_manifest.log
 ceph_wait_time: 60
 drbd_cephmon_res: /etc/drbd.d/drbd-cephmon.res
 tmp_drbd_cephmon_res: /tmp/drbd-cephmon.res
+ceph_conf: /etc/ceph/ceph.conf