From 9f54b3b4a7c1a7015def3a7987e3e9e426385251 Mon Sep 17 00:00:00 2001
From: Guillaume Abrioux <gabrioux@redhat.com>
Date: Tue, 10 Jul 2018 11:56:17 +0200
Subject: [PATCH] mon: ensure socker is purged when mon is stopped

On containerized deployment, if a mon is stopped, the socket is not
purged and can cause failure when a cluster is redeployed after the
purge playbook has been run.

Typical error:

```
fatal: [osd0]: FAILED! => {}

MSG:

'dict object' has no attribute 'osd_pool_default_pg_num'
```

the fact is not set because of this previous failure earlier:

```
ok: [mon0] => {
    "changed": false,
    "cmd": "docker exec ceph-mon-mon0 ceph --cluster test daemon mon.mon0 config get osd_pool_default_pg_num",
    "delta": "0:00:00.217382",
    "end": "2018-07-09 22:25:53.155969",
    "failed_when_result": false,
    "rc": 22,
    "start": "2018-07-09 22:25:52.938587"
}

STDERR:

admin_socket: exception getting command descriptions: [Errno 111] Connection refused

MSG:

non-zero return code
```

This failure happens when the ceph-mon service is stopped, indeed, since
the socket isn't purged, it's a leftover which is confusing the process.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
---
 roles/ceph-mon/templates/ceph-mon.service.j2 | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/roles/ceph-mon/templates/ceph-mon.service.j2 b/roles/ceph-mon/templates/ceph-mon.service.j2
index b6ebf0ba1..157ab7ca0 100644
--- a/roles/ceph-mon/templates/ceph-mon.service.j2
+++ b/roles/ceph-mon/templates/ceph-mon.service.j2
@@ -63,7 +63,8 @@ ExecStart=/usr/bin/docker run --rm --name ceph-mon-%i \
   -e CEPH_DAEMON=MON \
   {{ ceph_mon_docker_extra_env }} \
   {{ceph_docker_registry }}/{{ ceph_docker_image }}:{{ ceph_docker_image_tag }}
-ExecStopPost=-/usr/bin/docker stop ceph-mon-%i
+ExecStop=-/usr/bin/docker stop ceph-mon-%i
+ExecStopPost=-/bin/rm -f /var/run/ceph/{{ cluster }}-mon.{{ monitor_name }}.asok
 Restart=always
 RestartSec=10s
 TimeoutStartSec=120