Cannot use heterogeneous nova_backend setting

Bug #1888252 reported by Pierre Riteau
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
kolla-ansible
Triaged
Wishlist
Unassigned

Bug Description

The nova backend (e.g. default or rbd) can be set using the variables nova_backend_ceph or nova_backend. This setting applies only to compute nodes, so it should be possible to have different values for each host or group. I tried leaving nova_backend_ceph to "no" and enabling it only for a specific group, but it fails because the following task is skipped if the first compute node in the inventory doesn't use rbd:

- name: Check nova keyring file
  stat:
    path: "{{ node_custom_config }}/nova/{{ ceph_nova_keyring }}"
  delegate_to: localhost
  run_once: True
  register: nova_cephx_keyring_file
  failed_when: not nova_cephx_keyring_file.stat.exists
  when:
    - nova_backend == "rbd"
    - external_ceph_cephx_enabled | bool

This causes a later reference to nova_cephx_keyring_file.stat.path to fail.

Pierre Riteau (priteau)
description: updated
Revision history for this message
Mark Goddard (mgoddard) wrote :

Simple option: remove run_once. Performance would suffer, but we would support different keyrings per compute.

Better option: use the extract filter to pull out nova_backend from hostvars of applicable hosts.

Changed in kolla-ansible:
importance: Undecided → Wishlist
status: New → Triaged
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.