The variable {{ node_config_directory }} is used for the configuration
directory on the remote hosts, and should not be used for paths on the
deploy host (localhost).
This changes the default value of the TLS certificate and CA file to
reference {{ CONFIG_DIR }}, in line with the directory used for
admin-openrc.sh (as of I0709482ead4b7a67e82796e17f85bde151e71bc0).
This change also introduces a variable, {{ node_config }}, that
references {{ CONFIG_DIR | default('/etc/kolla') }}, to remove
duplication.
Change-Id: Ibd82ac78630ebfff5824c329d7399e1e900c0ee0
Closes-Bug: #1804025
kolla-ansible can take globals.yml from any directory by using
--config-dir argument. So store admin credentials there as well.
Not everyone runs kolla-ansible as a root.
Change-Id: I0709482ead4b7a67e82796e17f85bde151e71bc0
Add config_owner_user and config_owner_group to group_vars/all,
which is user and group of Kolla configuration files in /etc/kolla.
Add become to post-deploy playbook.
Add become to only neccesary tasks in roles:
- certificate
- common
- destroy
- haproxy
- mariadb
- memcached
- rabbitmq
Change-Id: I2aba745a6e3928c52642f64551470fd08cbfd058
Partial-Implements: blueprint ansible-specific-task-become
The admin-openrc.sh should copy to where the python-openstackclient was installed.
whatever multinode and all-in-one, the place where python-openstackclient was
installed is localhost, So admin-openrc.sh should copy to localhost.
The purpose of "connection: local" in ansible playbook is that make sure this script
can copy to localhost.In all-in-one, Writting as this is ok, it will copy to localhost,
but in multinode, this will make a bug, add ansible_connection=ssh in inventory file
the admin-openrc.sh will not copy to the localhost,the "connection:local" in post-deploy.yml
will be covered by "ansible_connection=ssh" in inventory file, then the script will be copied
to target node. So we should modify the hosts to localhost to avoid this bug.
Change-Id: I054717cc2b4adc600808282034a10a58c1184a38
Closes-Bug: #1666808
