Cannot provision OSDs in Luminous and Mimic

[alfredodeza]

Bug Report

What happened: A refactor (commit 4cc1506 ) introduced changes that attempt to start an OSD by its device path, which is incorrect for non-containerized environments.

Errors show up similar to:

Unable to start service ceph-osd@sdb: Job for ceph-osd@sdb.service failed because the control process exited with error code. See "systemctl status ceph-osd@sdb.service" and "journalctl -xe" for details.

What you expected to happen: This 'start osd' with a device name in it should not happen for non-containerized environments, and a test should be run to ensure that OSDs start in supported branches (3.2) for related Ceph versions (Luminous)

How to reproduce it (minimal and precise): Try to deploy ceph-disk OSDs in Luminous using the stable-3.2 branch

Share your group_vars files, inventory

Environment:

• OS (e.g. from /etc/os-release): Both CentOS7 and Xenial
• Kernel (e.g. uname -a):
• Ansible version (e.g. ansible-playbook --version): ansible-2.6.8
• ceph-ansible version (e.g. git head or tag or stable branch): stable-3.2
• Ceph version (e.g. ceph -v): latest dev version of Luminous (same can be seen in Mimic)