This project provides recipes for provisioning OpenStack clouds and application clusters on OpenPOWER servers:
- Private cloud w/ and w/o Swift Object Storage - Database as a Service (Trove) - Standalone Swift clusters
Each recipe is composed of the following solution specific files:
- README-<solution>.rst - bom/<solution>.pdf - configs/<solution>.yml
The README-<solution>.rst file provides instructions to install and configure the given solution. The starting point here is a set of bare metal servers and switches that have been appropriately racked and cabled according to the reference bom file. The operating system (and solution stack) will be installed by the toolkit.
Each solution is comprised of a unique set of hardware components that are described in a solution specific Bill of Materials file -- ie. bom/<solution>.pdf. This file provides information such as model numbers and feature codes to simplify the ordering process and it provides racking and cabling rules for the preferred layout of servers, switches, and cables.
The configuration file describes the software and hardware configuration of the solution. It maps each server to a solution specific node type which is referred to in the file as a node template and each node template to a managed set of operating system resources including networks, VLANs, VXLANS, disks, and users. It also provides mappings between networks, network adapters, and switch ports. Examples of node templates are controller, compute, and ceph-osd.
The configuration file comes pre-configured with a specific number of servers, disks, switches, networks, etc. that may need to be adjusted to reflect the actual configuration. The user is expected to make these adjustments and specify site specific parameters such as external ip addresses, so that the cluster is properly integrated into the data center. The configuration file includes comments to facilitate making these changes. Information is also provided in README files of related projects.
Note
Private cloud is described in README-dbaas.rst
There is not necessarily a one to one to one relationship between the three files listed above. The README file is the master document and should list the specific bom(s) and configuration file(s) that apply.
The toolkit runs on an Ubuntu 16.04 OpenPOWER server or VM that is connected to the internet and management switch in the cluster to be configured.
Get a local copy of this github project:
$ git clone git://github.com/open-power-ref-design/openstack-recipes $ cd openstack-recipes $ git checkout v1.3.0
Choose one of the following solutions to deploy:
The private compute cloud solution is described in the DBaaS BOM below.
Read the associated Bill of Materials (BOM) file as noted in the README, one of:
The private compute cloud BOM is included in the DBaaS BOM.
Rack and cable hardware as indicated in the selected BOM
Select a solution specific configuration file to use as a starting point for your configuration, one of:
- configs/private-compute-cloud.yml - configs/dbaas.yml - configs/swift-large.yml - configs/swift-medium.yml - configs/swift-small.yml - configs/swift-small-with-compute-cloud.yml
Place the selected configuration file to be activated:
$ git clone git://github.com/open-power-ref-design/cluster-genesis $ pushd cluster-genesis $ git checkout 1.0.1 $ cp ../configs/<selected filename> config.yml $ mkdir -p domain/scripts/ $ cp ../scripts/validate_config.py domain/scripts/ $ popd
Edit the placed config.yml file to reflect the actual configuration
This file tells the toolkit which servers to install and how they should be configured. Servers are indexed in the config file based by switch port. Switches are then programmed to allow tagged VLAN and VXLAN messages between named networked interfaces.
Instructions for editing the config file are provided in the file itself. Additional information may be found in the TBD-cluster-genesis-user-guide. Depending on the solution, additional settings may be required. These settings if required are located in the solution specific README file within this project.
Within each solution's config file is a bootstrap section that contains the GIT_BRANCH and SOLUTION_BRANCH variables. The GIT_BRANCH variable should be set to the branch or tag level of the code to load for os-services. The SOLUTION_BRANCH should be set to the level of openstack-recipes to pull the bootstrap-solution.sh script from.
Validate the placed config.yml file by running the following command:
$ ./scripts/validate_config.py --file ../cluster-genesis/config.yml
Launch the toolkit
In general, the toolkit is divided into three phases:
Load Linux and the solution specific installer
In this phase, Linux is loaded on each server. On one of these servers as indicated in the config.yml file, solution specific install tools are installed. These tools need to be manually configured to achieve the desired result in the solution, which is undertaken in the next phase.
In general, follow the instructions provided on the console. At the end of this phase, upon successful completion, the installation process is continued on another server. Log into the indicated server. The user credentials for this server are provided in placed config.yml file in the following fields:
- userid-default - password-default
Configure solution specific installer
The solution is composed by multiple projects which are identified below in the Related Projects section. Each of these projects may require customization. Consult the README file in each project to identify the settings that should be applied.
Please note that solution specific user credentials, user id and default passwords, are documented in these README files. For example, the user account kibana and password.
Install the solution
In this phase, the solution is installed, configured, and activated.
OpenStack based solutions utilize the following projects: