e09216aa4e
In order to demonstrate that OSA can be built without any containers, and to validate on a continuous basis that it still works, a new scenario is implemented to test it. As part of this, the following is done: 1. An example environment override file is added which sets all the containers with 'is_metal: true' which disables the container creation. 2. As haproxy is not used for the scenario on an AIO, the haproxy environment configuration is broken out from openstack_user_config into its own conf.d file and the implementation of it into user space is added to all other scenarios. 3. To ensure that the pip lockdown configuration is not implemented by the pip install role when the repo server doesn't exist yet, we ensure that the var is set in the playbook that does the validation. 4. To ensure that rabbitmq is able to correctly start up we implement the same host name on the host as we do in the inventory. Without this rabbitmq fails. In order to do this successfully with ansible, the dbus package must be installed on the host, so it is added as part of the AIO bootstrapping. 5. The workflow for a deployment needs no changes because when the lxc-related playbooks execute, they simply skip over because the lxc-related groups are empty. Depends-On: https://review.openstack.org/542307 Change-Id: I67199e1f35c91c4e2c9973e011e856c6ac3fb086 |
||
---|---|---|
.. | ||
dynamic-address-fact.yml | ||
dynamic-grouping.yml | ||
haproxy-endpoint-manage.yml | ||
mysql-db-user.yml | ||
os-log-dir-setup.yml | ||
os-lxc-container-setup.yml | ||
package-cache-proxy.yml | ||
rabbitmq-vhost-user.yml | ||
restart-service.yml | ||
set-pip-upstream-url.yml | ||
set-upper-constraints.yml | ||
unbound-clients.yml |