A Test Kitchen Driver for OpenStack.
This driver uses the fog gem to provision and destroy nova instances. Use an OpenStack cloud for your infrastructure testing!
Shamelessly copied from Fletcher Nichol's awesome work on an EC2 driver, and Adam Leff's amazing work on an VRO driver.
There are no external system requirements for this driver. However you will need access to an OpenStack cloud.
Add this line to your application's Gemfile:
gem 'kitchen-openstack'
And then execute:
$ bundle
Or install it yourself as:
$ gem install kitchen-openstack
Or if using chefdk install with:
$ chef gem install kitchen-openstack
driver:
name: openstack
openstack_username: [YOUR OPENSTACK USERNAME]
openstack_api_key: [YOUR OPENSTACK API KEY] # AKA your OpenStack Password
openstack_auth_url: [YOUR OPENSTACK AUTH URL]
require_chef_omnibus: [e.g. 'true' or a version number if you need Chef]
image_ref: [SERVER IMAGE ID]
flavor_ref: [SERVER FLAVOR ID]
transport:
username: ubuntu # For a Ubuntu Box
The image_ref
and flavor_ref
options can be specified as an exact id,
an exact name, or as a regular expression matching the name of the image or flavor.
All of Fog's openstack
options (openstack_domain_name
, openstack_project_name
,
...) are supported. This includes support for the OpenStack Identity v3 API.
Required Tell test-kitchen what driver to use. ;)
Required Your OpenStack username.
Required Your OpenStack API Key, aka your OpenStack password.
Required Your OpenStack auth url.
Required Set to true
otherwise the specific version of Chef omnibus you want installed.
Required Server Image ID.
Required Server Flavor ID.
If a server_name_prefix
is specified then this prefix will be used when
generating random names of the form <NAME PREFIX>-<RANDOM STRING>
e.g.
myproject-asdfghjk
. If both server_name_prefix
and server_name
are
specified then the server_name
takes precedence.
If you want to have a static prefix for a random server name.
The path to your private ssh key.
The path to your public ssh key.
If a key_name
is provided it will be used instead of any
public_key_path
that is specified.
If a key_name
is provided without any private_key_path
, unexpected
behavior may result if your local RSA/DSA private key doesn't match that
OpenStack key. If you do key injection via cloud-init
like this issue:
#77. The
key_name
should be a blank string if you need to skip it. Example:
driver:
[-- snip --]
key_name: ""
user_data: cloud_init
Set the SSH port for the remote access.
Your OpenStack tenant id.
Your OpenStack region id.
Your OpenStack availablity zone.
Your OpenStack compute service name.
Your OpenStack network name used to connect to, if you have only private network connections you want declare this.
When OpenStack downloads the image into cache, it takes extra time to provision. Timeout controls maximum amount of time to wait for machine to move from the Build/Spawn phase to Active.
server_wait
is a workaround to deal with how some VMs with cloud-init
.
Some clouds need this some, most OpenStack instances don't. This is a stop gap
wait makes sure that the machine is in a good state to work with. Ideally the
transport layer in Test-Kitchen will have a more intelligent way to deal with this.
There will be a dot that appears every 10 seconds as the timer counts down.
You may want to add this for WinRM instances due to the multiple restarts that
happen on creation and boot. A good default is 300
seconds to make sure it's
in a good state.
The default is 0
.
A list of security_groups
to join:
security_groups:
- [A LIST OF...]
- [...SECURITY GROUPS TO JOIN]
If your vms have cloud-init
enabled you can use the user_data
in your
kitchen.yml to inject commands at boot time.
driver_config:
user_data: userdata.txt
Then create a userdata.txt
in the same directory as your .kitchen.yml,
for example:
#!/bin/sh
echo "do whatever you want to pre-configure your machine"
Deprecated A list of network names or ids to create instances with.
network_ref:
- [OPENSTACK NETWORK NAMES OR...]
- [...ID TO CREATE INSTANCE WITH]
Deprecated You should be using transport now. This will skip the ssh check to automatically connect.
The default is false
.
Deprecated You should be using transport now. This will sleep for so many seconds. no_ssh_tcp_check
needs
to be set to true
.
Makes a new volume when set to true
.
The default is false
.
When set, will make a volume from that snapshot id.
When set, will attach the volume id.
Set this to vda
unless you really know what you are doing.
The block storage availability zone.
The default is nova
.
The volume type, this is optional.
This will delete the volume on the instance when destroy
happens, if set to true.
Otherwise set this to false
.
Timeout to wait for volume to become available. If a large volume is provisioned, it might take time to provision it on the backend. Maximum amount of time to wait for volume to be created and be available.
block_device_mapping:
make_volume: true
snapshot_id: 00000-111111-0000222-000
device_name: vda
availability_zone: nova
delete_on_termination: false
creation_timeout: 120
A specific floating_ip
can be provided to bind a floating IP to the node.
Any floating IP will be the IP used for Test Kitchen's Remote calls to the node.
A floating_ip_pool
can be provided to acquire the first free floating ip from
the pool to attach to the instance. It will be the IP used for Test Kitchen's
Remote calls to the node.
In some complex network scenarios you can have several IP addresses designated
as public or private. Use public_ip_order
or private_ip_order
to control
which one to use for further SSH connection. Default is 0 (first one)
For example if you have openstack istance that has network with several IPs assigned like
+--------------------------------------+------------+--------+------------+-------------+----------------------------------+
| ID | Name | Status | Task State | Power State | Networks |
+--------------------------------------+------------+--------+------------+-------------+----------------------------------+
| 31c98de4-026f-4d12-b03f-a8a35c6e730b | kitchen | ACTIVE | None | Running | test=10.0.0.1, 10.0.1.1 |
to use second 10.0.1.1
IP address you need to specify
private_ip_order: 1
assuming that test network is configured as private.
The network_ref
option can be specified as an exact id, an exact name,
or as a regular expression matching the name of the network. You can pass one
network_ref: MYNET1
or many networks
network_ref:
- MYNET1
- MYNET2
The openstack_network_name
is used to select IP address for SSH connection.
It's recommended to specify this option in case of multiple networks used for
instance to provide more control over network connectivity.
Please note that network_ref
relies on Network Services (Fog::Network
) and
it can be unavailable in your OpenStack installation.
disable_ssl_validation: true
Only disable SSL cert validation if you absolutely know what you are doing, but are stuck with an OpenStack deployment without valid SSL certs.
The following could be used in a .kitchen.yml
or in a .kitchen.local.yml
to override default configuration.
---
driver:
name: openstack
openstack_username: [YOUR OPENSTACK USERNAME]
openstack_api_key: [YOUR OPENSTACK API KEY] # AKA your OPENSTACK PASSWORD
openstack_auth_url: [YOUR OPENSTACK AUTH URL]
require_chef_omnibus: [e.g. 'true' or a version number if you need Chef]
image_ref: [SERVER IMAGE ID]
flavor_ref: [SERVER FLAVOR ID]
transport:
ssh_key: /path/to/id_rsa # probably the same as private_key_path
connection_timeout: 10
connection_retries: 5
username: ubuntu
password: mysecreatpassword
platforms:
- name: ubuntu-14.04
- name: ubuntu-15.04
- name: centos-7
transport:
username: centos
- name: windows-2012r2
transport:
password: myadministratorpassword
suites:
# ...
- Source hosted at GitHub
- Report issues/questions/feature requests on GitHub Issues
Pull requests are very welcome! Make sure your patches are well tested. Ideally create a topic branch for every separate change you make. For example:
- Fork the repo
- Create your feature branch (
git checkout -b my-new-feature
) - Run the tests and rubocop,
bundle exec rake spec
andbundle exec rake rubocop
- Commit your changes (
git commit -am 'Added some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create new Pull Request
Created by Jonathan Hartman ([email protected]) and maintained by JJ Asghar ([email protected])
Apache 2.0 (see LICENSE)