Some roles made a bad usage of the 'node_config_directory' variable.
As described here:
https://github.com/openstack/kolla-ansible/blob/master/ansible/group_vars/all.yml#L16
'node_config_directory' is the directory to store the config files on
the destination node.
This variable MUST be changed to 'node_custom_config'.
Futhermore this will unified all roles.
Closes-Bug: #1682445
Change-Id: Id8d8a1268c79befac8938c1e0396267314b40301
Nova service-list is returning empty registered service
when in database they are. Because of this, simple_cell_setup
is not executed and deploy stucks waiting for nova-compute.
This change temporaly checks the database for existing nova services
instead of openstack compute service list.
This change will need to be reverted once the command is fixed.
Change-Id: Ic508eb3ff03b5f233186353fc7697305cc792d14
If disable cinder in openstack environment.
Cinder of configuration items do not need to configure in nova.conf
Change-Id: Ie346dfe45f8c2d3fd383095d0c64f8e9421100ff
* Ironic do not support multi glance ips.
* Write the nova-compute-ironic binary log to nova-compute-ironic.log
file
Change-Id: I87359c47a5845c4d7a6ab9daaefcc94a51c92eb0
Closes-Bug: #1671989
Add support for basic multiple regions, that is to say, many OpenStack
with a shared Keystone (same users) and Horizon. The shared Keystone
and Horizon are deployed into one region, for instance RegionOne.
Services of other regions have an access to this Keystone. This
support assumes that the operator knows the name of all OpenStack
regions in advance, and considers as many Kolla runs as there are
regions.
The new variable, multiple_regions_names, contains the name of
regions. It is needed by the region that includes Keystone and
Horizon. In register.yml, it specifies to create as many Keystone
endpoints as there are regiones, so that services of other regions can
connect to Keystone. In local_settings.j2, it changes the render to
support multiple regions in Horizon. The multi-regions.rst explains
how to perform a multiple regions deployment.
Implements: blueprint multi-kolla-config
Change-Id: Icab2aebfc4de0e3bc609950956e0af397705f403
Nova external ceph task have a type which break deployment.
State module not present, the module used should be stat.
Change-Id: Ie8a0b30f44fc35a597334383a85353d324e765cd
Closes-Bug: #1671526
Add a new subcommand 'check' to kolla-ansible, used to run the
smoke/sanity checks.
Add stub files to all services that don't currently have checks.
Change-Id: I9f661c5fc51fd5b9b266f23f6c524884613dee48
Partially-implements: blueprint sanity-check-container
cell0 is already setup in Ocata, when upgrading to Pike
is not anymore necessary to create.
All nova DBs (nova_api, nova and nova_cell0) are already
created in Ocata too. Only bootstrap_service is needed
while upgrading.
Change-Id: Idc4941334faf91feee868472155a8c8ea0eba436
It is unnecessary and a waste to run nova-compute-ironic on a single
node, change the host name field in nova-compute-ironic could ensure
nova-compute and nova-compute-ironic run on the same node.
Change-Id: If2acbf25caab0570da444472003599b9e0f2a1ff
Closes-Bug: #1671103
Booting from volume require cinder's ceph client secret now. Move cinder
before nova in site.yml, because nova depends on cinder ceph client key
now.
Change-Id: I01c9ed80843d98305b8963894c4917c21a35d3ac
Closes-Bug: #1670676
Libvirt 2.5 (latest version in cloud archive) seems to have changed
behavior of dynamic_ownership.
Change-Id: Icf9592c2f82f0f4a35074b9b13a51643a32631e9
Closes-bug: #1668654
* Move the tasks to the role
* Skip the task when container is already running
Change-Id: I1990d4dd2a02efa2b3766329000aa23419e0ff17
Closes-Bug: #1670286
nova-compute will try to connect nova-conductor when upgrade.level is
auto. So nova-conductor need run more earlier. Otherwise, nova-compute
will failed to connect nova-conductor through message queue.
Closes-Bug: #1665836
Change-Id: If6296a84305bc283fb071b9650e629c6b1b71e61
In kolla-kubernetes, nova-compute-ironic is ran with a separate virtual
interface with a different hostname. So it is OK to run nova-compute and
nova-compute-ironic on the same node.
Change-Id: I3b0f12f128af8210bc97502dbc01469f619dd1ac
In ironic environment deployment, the compute nodes info will be empty
until ironic node is created. There are also some case that user just
want deploy without any nova-compute.
Also enable auto discover hosts feature. This is useful for small
environment.
Closes-Bug: #1666031
Change-Id: I6f3d1c3668452a404875aa5621ee99b2b41e28f0
