A ceph-ansible site-docker.yml run [1][2] produced a deployment with 0 failures and 0 OSDs because the osd_scenario variable was not set. I think this should have failed.
As per the "make sure an osd scenario was chosen" task in check_mandatory_vars.yml [3] the deployment will fail if an osd_scenario is not chosen. However, the when condition does not do this check for a containerized deployment. Why should this check not be done for a containerized deployment? If the code to check this is somewhere else, why wasn't it executed?
Additional info:
- The logs show that both scenarios were skipped [4].
- If you search the --extra-vars input to ansible-playbook it will show no scenario was passed. This is an error on part of the process calling the playbook, but ceph-ansible should have failed that deployment request.
[1] Resulting run: http://paste.openstack.org/show/623760/
[2] Playbook command with --extra-vars http://paste.openstack.org/show/623752/
[3] https://github.com/ceph/ceph-ansible/blob/b685aceede71a2dc872ce140ac636a7301f471f6/roles/ceph-osd/tasks/check_mandatory_vars.yml
[4]
2017-10-16 10:16:00,713 p=12340 u=mistral | TASK [ceph-osd : include scenarios/collocated.yml] *****************************
2017-10-16 10:16:00,729 p=12340 u=mistral | skipping: [192.168.24.15]
2017-10-16 10:16:00,735 p=12340 u=mistral | TASK [ceph-osd : include scenarios/non-collocated.yml] *************************
2017-10-16 10:16:00,753 p=12340 u=mistral | skipping: [192.168.24.15]
(In reply to Ken Dreyer (Red Hat) from comment #9)
> John would you please confirm this is fixed in v3.1.0beta3 (or v3.1.0beta4)?
I confirm this is fixed. I tested with v3.1.0beta4.