Closes-Bug: #1704915
Change-Id: If85cd7bc8547494900c8f6c32dfd1c1f2179f928

* remove ceilometer-api and ceilometer-collector service
* use ceilometer-notification to publish message to proper backend
* remove useless ceilometer_database_type and ceilometer_event_type
  variables
* sync event_definitions.yaml, event_pipeline.yaml and pipeline.yaml
  file with upstream

Change-Id: Ib39053cb5f70bd11ee61d3f26d5b28accecd7190

* use gnocchi-api script for wsgi based on this commit[0].
* logfile is deprecated. Use log-file instead[1].

[0] https://github.com/gnocchixyz/gnocchi/commit/c8717e8b71f044953aec744f44f97abad1a0c08e
[1] https://github.com/gnocchixyz/gnocchi/blob/master/gnocchi/opts.py#L77,L82

Change-Id: I8fbe8425e43d82e0f9601ba6303dd7f24142a2e6

cinder_backup_ceph is true in default when enable_ceph is true.

Change-Id: Ic3cbd041a784610323cda4419acb6702f2aa35c1
Closes-Bug: #1699658

Add VMware VMDK driver supports to cinder service.
The VMware driver for OpenStack Block Storage is recommended and
should be used for managing volumes based on vSphere data stores.
see
https://docs.openstack.org/ocata/config-reference/compute/hypervisor-vmware.html
https://docs.openstack.org/ocata/config-reference/block-storage/drivers/vmware-vmdk-driver.html#block-storage-vmdk-driver

Partially-implements: blueprint kolla-ansible-support-vsphere

Change-Id: Ic3eb7ae34c1e8584945b3d97f8b427ee67ea8fba

Base on nova release note[0], all general scheduler configuration
options have been added to the [scheduler] group.

[0] https://github.com/openstack/nova/blob/master/releasenotes/notes/add-scheduler-config-to-scheduler-group-c83bc770e67ac115.yaml

Change-Id: Ic15ecaced3c78b86d04572fbaf2f6be5640a2a25
Closes-Bug: #1680307

Either api_v1 or api_v2 must be True in [image-feature-enabled]

Change-Id: I8d25d4b43b030c58e9c365937d81c82418a47436
Closes-bug: #1703722

Change-Id: Ia92ea9114ffec1dc3108969cb48a3264a23e0eea
Closes-bug: #1703515

designate-api service is launched using dns_interface instead of api_interface.
Haproxy is using the good interface (api_interface), hence designate_api
can't be accessed.

Change-Id: I24e356c167c931a6b69a283efba12117fcff936a
Closes-Bug: #1703425

Ironic resources can be indexed by Searchlight.
This patch enable it.

Change-Id: I003a10055629189791adbb522344e6c9cec597f3

* nova_server resources resources are not actually
indexed.

* nova flavor are not indexed too, cause the notifications topic for
flavors is versioned_notifications.
We need to config notifications_topics_exchanges with value
'versioned_notifications,nova' in order to get the related versioned.

Change-Id: Ibf4ee7d7efceee5aefac5f83e22b994ae33ce047
Closes-Bug: #1703435

* "bool" filter is removed when not required
* 'not' is used instead of '== False' check

Change-Id: I85a5bb9a5ea874ac1c397cbf8de416147d2424c3

In ansible 2.3.1, conditional statement should not include jinja2
templating delimiters (e.g. {{ ).  So, this change removes the
delimiters from when statemant, and also removes unnesessary
parentheses from until statement.

Closes-Bug: 1702607

Change-Id: I071ffd5ba6175ee1bc61719a84b805c022753459

User can use custom directory to store images.
For example using a shared file system as backend.

Change-Id: Iab7a9a51e619bdbf8f7bd4626ebe4d703e6d4819

According to [1], Ansible variable names should not include hyphens.
Kolla-Ansible fails with a wrong variable name 'barbican-api' when
deploying Barbican.

This patch fixes the issue that was recently introduced in [2].

[1] http://docs.ansible.com/ansible/playbooks_variables.html#id15
[2] 08ab3d8e

Change-Id: Ib962e31ad93316e56130c9fc38dabfc918de17ce
Closes-Bug: #1703287

Ceph rgw can be used as object store instead of Swift.
This patch enable trove to use ceph rgw as object store.

Change-Id: I50b878078b7c62c1034a102d064dfa90a1357ee8

If use /run/netns, it may cause namepace stale and unable to remove. The
root cause may be that docker engine still unrelease the namespace
folder.

Change-Id: If120d54c266878990c952b60043b596b0a0788ab
Closes-Bug: #1703077

Change-Id: I32de4c37f531a3d22bf80e1eca6635631e489842
Closes-Bug: #1702842

fwaas is in pending create state when I attach it to a router.
This patch fix this.

Change-Id: I18b56ed3698e22a02a8718b39360fde76c12428b
Closes-Bug: #1689703

As described here:
https://github.com/openstack/keystone/blob/master/keystone/resource/core.py#L841
https://github.com/openstack/keystone/blob/master/keystone/conf/identity.py#L21

* default project domain name MUST be named 'Default'
* default project domain id MUST be named 'default'
* default project user name MUST be named 'Default'
* default project user id MUST be named 'default'

Change-Id: I610a0416647fdea31bb04889364da5395d8c8d74

Cinder ubuntu binary was already migrated under apache.
This change migrates rest of the distros/install types to
be executed under an apache process.

Change-Id: I8544eec5bfebf771d758f13abf4dfbc802f5d24f

This change [0] reverted designate dashboard change because
designate was not finished, we forgot to enable again.

[0] https://review.openstack.org/#/c/408714/

Change-Id: Ibaf7e5a5dc8cbef619d86a0f2b240d384984e8bd

Custom file was check on remote target instead of local.

Change-Id: I9426056e7bb284eb8b3ad539d61ecb1e1f6370e4
Closes-Bug: #1702490

* fix wrong variable usage for horizon copy policy task
* notify restart keystone container when policy is changed

Change-Id: I3545205d5d3cfcf7bf893187ca6e65bbc152bf33
Closes-Bug: #1702486

In order to speed up deployment time some "local" actions should be run
only once using 'run_once: True'.
This will decrease deployment time in case of multihost configuration.

Change-Id: I6015d772d35c15e96c52f577013b6e41197cb41a

Kolla-ansible actually bring it's own barbican-api-paste.ini file to
enable Keystone authentication, in order to fix this
https://bugs.launchpad.net/kolla/+bug/1625337

auth_token middleware is actually managed by Barbican.

Furthermore barbican-api-paste.ini brings by Kolla-ansible is outdated:
* http_proxy_to_wsgi middleware is missing

Hence this file should not be managed statically by kolla-ansible.
This patch keep custom paste file feature. Just put the file to
/etc/kolla/config/barbican/barbican-api.ini path.

Change-Id: Ia50237f7df7f89526a976575b017145c71b11ec0
Closes-bug: #1695026

Change-Id: Id3ceaa27687fda3b773873501208dacbfa3536ab
Closes-bug: #1699680

Change-Id: Ia2afe54d37732ff33216738d7478ad98f6b84cc4

As of [1], the ironic configuration for neutron, glance, swift,
inspector and service_catalog requires explicit configuration of
authentication parameters for communication with these services.

This change adds the required parameters to [neutron], [glance] and
[inspector] sections of ironic.conf. Kolla-ansible does not configure
the [swift] or [service_discovery] sections currently.

We also replace option [glance] glance_hosts with [glance]
glance_api_servers as the former is deprecated.

Since we no longer need to support generating configuration for
kolla-kubernetes[2], some related options have been cleaned.

[1]
https://github.com/openstack/ironic/commit/4f9035c24f0465be5728ed9a8a6df76cd6f46ed3
[2]
https://blueprints.launchpad.net/kolla-ansible/+spec/clean-k8s-config

Change-Id: Ifc239af5f3e44a508fedc9dea08cb06160c4f7f3
Closes-Bug: #1701713

'elasticsearch_protocol' is define in group_vars but unused,
it must be removed.

Change-Id: I15cabb9a2d73dd60d06ec2a4fcececa76e14a1db

When an external Etcd server is used, config is generated like following:

etcd:
  client_timeout: 100
  embeded: no
  servers:
    - http://10.200.200.10:2379

"embeded" config has wrong spelling, the good one is "embedded".
Hence default config is used (yes), and external Etcd server is not used.

Closes-Bug: #1697614
Change-Id: I59ac990b0b865c926b53f829bdfea186fc1e10b1

SKIP_LOG_SETUP variable was used by Heka stuff.
Heka has been removed from Kolla and Kolla-ansible.

Change-Id: I4448b26ace899eb888d34a12a85b019597e25435

rpc_workers and rpc_state_report_workers are set to 1 by default in
Neutron:
https://github.com/openstack/neutron/blob/master/neutron/conf/service.py#L30

By design neutron-server is a central RPC service for all agents (L2
agents, L3 agents, Metadata agents, LB agents, VPN agents, ...).
For a production ready cloud, these variables must be set to a greater
value than 1.

Change-Id: Ib39be32748c3ee8077941fd1659db12c9d80055c
Closes-Bug: #1671734

* add additional options called 'endpoint_type' for each of config groups
related to openstack clients used by Magnum.
* add Glance, Neutron and Nova config groups.

Change-Id: Ie74979e05c4f5763674ba2fc5b9f07bd51ad9454

Some roles have a symlink to deploy.yml file
for reconfigure. This is causing some issues.
"included task files must contain a list of tasks"

Change-Id: Ie7ade52900a61bc1c5b867fa7a8f75fc541a6426
Closes-Bug: #1694251

The TFTP server used by ironic and ironic inspector (in.tftpd) requires
files to be world readable in order for them to be accessible via
TFTP[1].

The permissions of these files were recently changed to 0600 along with
a number of other files[2].

This change reverts the permissions to 0644 for the ironic inspector PXE
configuration files.

[1] https://linux.die.net/man/8/in.tftpd (security section)
[2]
https://github.com/openstack/kolla-ansible/commit/274291463e99eab805a4265adc856c1bffafa9ad

Change-Id: Ibc281949ebf5bab1e1d2e450ec943728aa00943b
Closes-Bug: #1701695

Logrotate configuration is missing for ironic-inspector. Prior to [1],
ironic-inspector logs were stored in
kolla_logs/ironic/ironic-inspector.log. After [1], the logs
are stored in kolla_logs/ironic-inspector/ironic-inspector.log
but the logrotate config was not updated.

[1]
https://github.com/openstack/kolla-ansible/commit/07453f346094b184a209380e375f6311987675be

Change-Id: I7e78faea361cd67069c1a96aaf0a2ffffc0e8666
Closes-Bug: #1701578

RDO packages a distribution configuration file
/usr/share/nova/nova-dist.conf which contains the following setting:

firewall_driver = nova.virt.libvirt.firewall.IptablesFirewallDriver

This causes the nova-compute-ironic service to fail to start as the
nova ironic virt driver attempts and fails to create a firewall driver
using this class.

This change reinstates the explicit setting of the [DEFAULT]
firewall_driver option to the noop driver which resolves this issue.
This comes at the cost of a WARN log message due to the option being
deprecated (see
https://github.com/openstack/kolla-ansible/commit/6d831db687a35a48a41ee581b979fb43350e0c72).

Change-Id: I41bd9d0671118ff256e7ada766e8653bb4b2b376
Closes-Bug: #1701564

Add additional options called 'endpoint_type' for each of config groups
related to nova, cinder and neutron clients.

Change-Id: I24dc11502b148fbe564dc63c6f78b7dcbfe44f01

As describe here:
https://github.com/openstack/cinder/blob/master/cinder/compute/nova.py#L42

* remove 'nova_catalog_info' deprecated option
* add new [nova] section

Change-Id: Ib89a589c8eb8d81839bd4d07d7174b3272136934