This guide provides step by step instructions to use Kolla-ansible to deploy OpenStack controller services on x86 server and OpenStack compute of KVM on LinuxONE.
The following three classes of nodes are defined in the architecture: deployer, controller, and compute.
deployer node: This is used for running local docker registry, package repo, and also where the kolla-ansible tool would be installed, configured and run.- System requirements:
- Ubuntu 16.04/x86_64 platform
- 1 network interfaces
- 8GB main memory
- 400GB disk space
controller node: This is where the OpenStack controller services will be deployed onto. System requirements:- Ubuntu 16.04/x86_64 platform
- 2 network interfaces
- 8GB main memory
- 40GB disk space
compute node: This is where the OpenStack compute services will be deployed onto. System requirements:- Ubuntu 16.04/LinuxONE platform
- 1 network interface
- 8GB main memory
- 10GB disk space
Note
- Root access to all the nodes is required.
- The deployer node can be the same node with controller node.
The deployer node is supposed to run the local package repository server which is used by the kolla-ansible deploy process to install packages on target nodes including controller and compute nodes, run the local pypi server which serves python packages, run the local docker image repository to replace the need to pull images from Docker Hub. With these local servers, this omits the needs of network connection of the deploy process.
These functions require several files to be prepared and put onto the deployer node. The files should be organized as following:
root@deployer:/data/OpenStackCE# tree -L 1 . |-- deployment-docker-images.tar |-- docker-ce |-- docker-registry |-- kolla-ansible |-- pypi `-- ubuntu-mirror
deployment-docker-images.tar- Contains the three docker images: pypiserver, nginx, and registry which will be used to run the local pypi server, package repository and docker image registry respectively.docker-ce- Contains docker-ce deb package and its dependency used to install docker-ce on the deployer node.docker-registry- Contains the built docker images used to deploy OpenStack services, including both the x86_64 arch images for controller node and s390x arch images for compute node.pypi- Contains the needed python packages used by kolla-ansible to deploy OpenStack services.ubuntu-mirrir- Contains the ubuntu mirror and docker mirror which provides all the deb packages used by kolla-ansible in the deploy process.
These files need to be put into a single folder, generally we recommend you to put them into /data/OpenStackCE which is the default folder used by the setup tool to find the required files. Specify the folder base when running the setup tool if you use other folder name.
The following guides will use the /data/OpenStackCE as the folder name for the required files.
The /data/OpenStackCE/kolla-ansible holds the tool used to setup the deployer. You can check the tool usage by running:
/data/OpenStackCE/kolla-ansible/tools/setup-deployment-server -h
Then run the tool with your arguments if required:
/data/OpenStackCE/kolla-ansible/tools/setup-deployment-server
This tool would automatically do the following jobs:
- install and run docker daemon
- run local docker registry
- run local package repository server
- run local pypi server
- configure deployer node to use local repository
- install kolla-ansible and its dependencies
- collect and generate configuration templates into
etc/kolla
The tool would print actions taken by each step and the corresponding result. Please make sure each step finishes successfully.
When the tool finishes successfully, you can see there are three containers running as following:
root@deployer:/data/OpenStackCE# docker ps CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 51417101e46f pypiserver:master "pypi-server -p 8080…" About an hour ago Up About an hour 0.0.0.0:8080->8080/tcp pypiserver 85de3de2fedb registry:2 "/entrypoint.sh /etc…" About an hour ago Up About an hour 0.0.0.0:5000->5000/tcp registry 430bee95f31b nginx:1.15.3 "nginx -g 'daemon of…" About an hour ago Up About an hour 0.0.0.0:8000->80/tcp nginx
Also, you can see the deployer node has been setup to use local PYPI server and package repository:
root@deployer:/data/OpenStackCE# cat /etc/apt/sources.list deb [arch=amd64] http://DEPLOYER_IP:8000/archive.ubuntu.com/ubuntu/ xenial main restricted universe multiverse deb [arch=amd64] http://DEPLOYER_IP:8000/archive.ubuntu.com/ubuntu/ xenial-updates main restricted universe multiverse deb [arch=amd64] http://DEPLOYER_IP:8000/archive.ubuntu.com/ubuntu/ xenial-backports main restricted universe multiverse deb [arch=amd64] http://DEPLOYER_IP:8000/security.ubuntu.com/ubuntu xenial-security main restricted
The configuration steps includes inventory file, global settings, passwords used by OpenStack services, and OpenStack service configurations. And these steps all need to be done on the deployer node.
Inventory is an ansible file for user to specify target nodes, roles and access credentials.
The deployer setup tool would automatically generate the configuration example files for you under the /etc/kolla
folder:
root@deployer:~# ls /etc/kolla all-in-one globals.yml multinode openstack-ce-presetups.sh passwords.yml
Since we have separate host for controller node and compute nodes, so we need to edit the multinode inventory file.
Edit the first section of multinode with connection details of your environment, for example:
# For other connection method, please refer to the ansible document. [control] testcont network_interface=eth0 ansible_connection=ssh ansible_user=root ansible_ssh_pass=PASSWD # when you specify group_name:children, it will use contents of group specified. [network:children] control # inner-compute is the groups of compute nodes which do not have # external reachability [inner-compute] # external-compute is the groups of compute nodes which can reach # outside [external-compute] testcomp network_interface=enc123 ansible_connection=ssh ansible_user=root ansible_ssh_pass=PASSWD [compute:children] inner-compute external-compute [monitoring] [storage] [deployment] localhost ansible_connection=local
Please customize the example contents to suit your own environment:
- The
network_interfacevalue under thecontrolandexternal-computesection should be the name of interface which is used for the deployer to communicate with the controller or compute node respectively. - More than one compute node can be specified in the inventory file.
- The
testcontandtestcompshould be the hostname of the target controller node and compute node respectively. And the hostname must be resolvable on the deployer node, otherwise please specify the IP address and hostname pair in the /etc/hosts on the deployer node.
To learn more about inventory files, check Ansible documentation.
The /etc/kolla/passwords.yml file contains all the passwords that can be specified and will be used by the kolla-ansible
deploy process. Initially all passwords are blank in this file and can be filled either manually or by running random password
generator:
kolla-genpwd
This tool would fill the /etc/kolla/passwords.yml file with randomly generated passwords. You can further update specific
passwords as you need.
/etc/kolla/globals.yml is the main configuration file used by Kolla-ansible. The deployer setup tool has automatically setup
most of the required options for you, including:
--- kolla_base_distro: "ubuntu" kolla_install_type: "binary" openstack_release: "queens" node_custom_config: "/etc/kolla/config" docker_registry: "DEPLOYER_IP:5000" docker_namespace: "linuxone" local_docker_apt_url: "http://DEPLOYER_IP:8000/download.docker.com/linux/ubuntu" enable_fluentd: "no" enable_haproxy: "no" enable_heat: "no"
There are other options that are required to be specified as fit to your environment:
# Set the kolla_internal_vip_address value to the IP address of your "network_interface" as set in the [control] section of # the inventory file. kolla_internal_vip_address: "YOURIP" # Set the neutron_external_interface to the interface given to neutron as its external network port. This interface should be # active without IP address. neutron_external_interface: "INTERFACE" # Optional but suggest to enable the following option for further debug convenience. openstack_logging_debug: "True"
For deploy to kvm compute node on LinuxONE, some configurations are required for nova-compute service.
Create the /etc/kolla/config/nova/nova-compute.conf file as done by following steps:
root@deployment-server:~# mkdir -p /etc/kolla/config/nova root@deployment-server:~# vi /etc/kolla/config/nova/nova-compute.conf root@deployment-server:~# root@deployment-server:~# cat /etc/kolla/config/nova/nova-compute.conf [DEFAULT] enable_apis=osapi_compute,metadata compute_driver=libvirt.LibvirtDriver config_drive_format=iso9660 force_config_drive=True firewall_driver=nova.virt.firewall.NoopFirewallDriver pointer_model=ps2mouse [vnc] enabled=False [libvirt] virt_type=kvm cpu_mode=none inject_partition=-2
Other Configurations:
Kolla-ansible allows the operator to override configuration of services. Kolla-ansible will
look for a file in /etc/kolla/config/<< service name >>/<< config file >>.
This can be done per-project, per-service or per-service-on-specified-host.
After configuration is set, we can proceed to the deployment phase.
Bootstrap servers to setup basic host-level dependencies:
kolla-ansible -i /etc/kolla/multinode bootstrap-servers
Do pre-deployment checks for hosts:
kolla-ansible -i /etc/kolla/multinode prechecks
Proceed to actual OpenStack deployment:
kolla-ansible -i /etc/kolla/multinode deploy
When this playbook finishes successfully, OpenStack should be up, running and functional!
OpenStack requires an openrc file where credentials for admin user etc are set. To generate this file run
kolla-ansible post-deploy . /etc/kolla/admin-openrc.sh
Install basic OpenStack CLI clients:
pip install python-openstackclient python-glanceclient python-neutronclient