Uninitialised variables cause AIO run failure when deploying w/ ceph

[Matt Thompson]

I'm seeing the following when deploying an AIO that uses ceph as a storage backend:

fatal: [master-aio -> 172.29.238.217] => error while evaluating conditional: inventory_hostname in groups[item.0.component] and (item.0.component != 'cinder_backup' or (cinder_service_backup_program_enabled | bool and cinder_service_backup_driver == 'cinder.backup.drivers.ceph'))

The variables cinder_service_backup_program_enabled and cinder_service_backup_driver need to be added to group_vars. When running os-nova-install.yml on an AIO w/ ceph support enabled, the AIO host hosts both nova_compute and cinder_backup but does not have access to the variable defaults in os_cinder role.

[Matt Thompson (mattt416)]

https://review.openstack.org/#/c/221705/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/221705
Committed: https://git.openstack.org/cgit/stackforge/os-ansible-deployment/commit/?id=466de05c67fa4ea071909a20872c3876711a5c23
Submitter: Jenkins
Branch: master

commit 466de05c67fa4ea071909a20872c3876711a5c23
Author: Matt Thompson <email address hidden>
Date: Wed Sep 9 12:07:50 2015 +0100

    Ensure cinder-backup-related variables are defined

    This commit simply updates ceph_client/tasks/ceph_auth.yml to ensure
    that cinder_service_backup_program_enabled and
    cinder_service_backup_driver are defined before trying to evaluate
    them. On a normal deployment, this is not an issue but when running
    os-nova-install.yml on an AIO where ceph support is enabled, this
    variable evaluation fails as the AIO host is part of both cinder_backup
    and nova_compute groups, but does not have access to these variable
    defaults in os_cinder/defaults/main.yml since we're operating on the
    os_nova role.

    An alternate solution would be to add these variable defaults to
    group_vars, however as the os_nova role doesn't actually need to use
    them it's probably cleaner to just check if they're defined.

    Change-Id: I2c35b14f42b2c452aa68a611ab68c612db1b07a4
    Closes-Bug: #1493799

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: kilo
Review: https://review.openstack.org/222610

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/222610
Committed: https://git.openstack.org/cgit/openstack/openstack-ansible/commit/?id=ba4661296ce3f9703748c6a2843cd7ff63712db4
Submitter: Jenkins
Branch: kilo

commit ba4661296ce3f9703748c6a2843cd7ff63712db4
Author: Matt Thompson <email address hidden>
Date: Wed Sep 9 12:07:50 2015 +0100

    Ensure cinder-backup-related variables are defined

    This commit simply updates ceph_client/tasks/ceph_auth.yml to ensure
    that cinder_service_backup_program_enabled and
    cinder_service_backup_driver are defined before trying to evaluate
    them. On a normal deployment, this is not an issue but when running
    os-nova-install.yml on an AIO where ceph support is enabled, this
    variable evaluation fails as the AIO host is part of both cinder_backup
    and nova_compute groups, but does not have access to these variable
    defaults in os_cinder/defaults/main.yml since we're operating on the
    os_nova role.

    An alternate solution would be to add these variable defaults to
    group_vars, however as the os_nova role doesn't actually need to use
    them it's probably cleaner to just check if they're defined.

    Change-Id: I2c35b14f42b2c452aa68a611ab68c612db1b07a4
    Closes-Bug: #1493799
    (cherry picked from commit 466de05c67fa4ea071909a20872c3876711a5c23)

[Davanum Srinivas (DIMS) (dims-v)]

This issue was fixed in the openstack/openstack-ansible 11.2.11 release.

[Doug Hellmann (doug-hellmann)]

This issue was fixed in the openstack/openstack-ansible 11.2.12 release.

[Davanum Srinivas (DIMS) (dims-v)]

This issue was fixed in the openstack/openstack-ansible 11.2.14 release.