An SDK like wrapper around the openstackclient to reduce the need to write boilerplate subprocess code.
Ospclientsdk can be easily installed with a one line command. It is recommended (as best practice) to create a virtual environment and install ospclientsdk. Please see the commands below to install. Note this is only supported in python 3.6 and higher.
# install python virtualenv
$ pip install python-virtualenv
# create virtualenv
$ virtualenv ospclientsdk
# activate virtualenv
$ source ospclientsdk/bin/activate
# install blaster
$ (ospclientsdk) pip install ospclientsdk
Let's dive into how to use this sdk.
When working with the sdk the main interface you will be interacting with is the ClientShell class. This interface can be initialized with with or without credentials
from ospclientsdk import ClientShell
import yaml
"""
You can provide a string path to a clouds.yaml file. If a cloud isn't specified
the shell will default to using one called 'openstack'
"""
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
"""
You can also preload the clouds.yaml file or build the dictionary by hand
and provide it that way.
"""
creds = {}
with open('clouds.yaml') as f:
creds = yaml.safe_load(f)
shell = ClientShell(cloud_dict=creds, cloud='test')
"""
You can also load the shell with credentials after it has been initialized
"""
shell = ClientShell()
shell.load_cloud_config(cloud_dict=creds, cloud='test')Once the shell has been intialized you can see what command groups and commands are available in a couple of ways.
Note when the commands are loaded the white spaces are replaced with underscores ('_') to make it easier to parse and find what you're looking for. i.e. server create to server_create
from ospclientsdk import ClientShell
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
"""
You can see a simplied version of all available command groups and their respective commands.
This will return a dictionary. Where the key is the name of the of the command group and
the values are a list of commands supported by the command group.
"""
cmd_groups = shell.all_osp_cmd_groups
cmd_list = cmd_groups.get('data_processing', {})
"""
If you already know the name of the command group you want and would like to see the
commands available for it. You can access one of the many command group properties.
"""
cmd_list = shell.data_processing_commands
"""
If you would like to check which are the available commands irrespective of command group
you can do so as well
"""
all_cmds = shell.all_osp_commandsOnce you have an idea of what commands you want to execute you have a few different ways of executing them. Before we dive into those, let's talk a little bit about generating the options and their arguments.
The options and argument values for the command actions are always defined as dictionaries. Let's touch on a couple examples
Any time an option takes an argument
# This
command create --option arg <res_name>
# Can be defined like
{'option': 'arg', 'res': 'res_name'}Any time an option takes an argument and can be specified multiple times
# This
command create --option arg1 --option arg2 <res_name>
# Can be defined like
{'option': ['arg1', 'arg2'], 'res': 'res_name'}Any time an option takes an argument with the value in k=v
# This
command create --option arg1=val1
# Can be defined like
{'option': [{'arg1': 'val1'}], 'res': 'res_name'} Any time an option takes an argument with the value in k=v and can be specified multiple times.
# This
command create --option arg1=val1 --option arg2=val2 <res_name>
# Can be defined like
{'option': [{'arg1': 'val1'}, {'arg2': 'val2'}], 'res': 'res_name'}Any time an option takes an argument but the value can be a comma separated list of k=v
# This
command create --option arg1=val1,arg2=val2,arg3=val3 <res_name>
# Can be defined like
{'option': [{'arg1': 'val1', 'arg2': 'val2', 'arg3': 'val3'}], 'res': 'res_name'}Any time an option takes no argument and actions like a boolean flag
# This
command create --option <res_name>
# Can be defined like
{'option': True, 'res': 'res_name'}In the case of add command actions, which not only require the name or id of the resource it needs the name or id of the target resource you want to add
# This
command add --option arg1 <res_name> <tgt_res>
# Can be defined like
{'option': 'arg1', 'res': 'res_name', 'tgt_res': 'tgt_res'}Note rather than name you could supply id and rather than tgt_name you can supply tgt_id
In the case of delete command actions, you can specify multiple resources
# This
command delete <res_name_1> <res_name_2>
# Can be defined like
{'res': ['res_name_1', 'res_name_2']}For a full list of commands and options refer to the openstackclient documentation.
The first and recommended way is to use the high level APIs the shell provides through a series of proxy instances. Each proxy object reflects a command group and provides APIs that map to the corresponding CLI commands.
from ospclientsdk import ClientShell
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
# Build a dictionary of the required arguements/options and their parameters.
# Each key is command option and the value is the parameter.
server_params = dict(name="ccit_test_client",
image="rhel-7.5-server-x86_64-released",
flavor="m1.small",
key_name="db2-test",
network=["test_private_network"],
wait=True
)
# To create a server using the above define dict I simply access
# the compute instance and us the server_create function
results = shell.compute.server_create(server_params)from ospclientsdk import ClientShell
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
# Build a dictionary of the required arguements/options and their parameters.
# Each key is command option and the value is the parameter.
trunk_port_params = dict(name="test_compute_trunk",
parent_port="test_port",
subport=[dict(port="dummy_port",
segmentation_type="vlan",
segmentation_id=2007
)]
)
# To create a network trunk port using the above defined dict I simply access
# the neutronclient instance and us the network_trunk_create function
results = shell.neutronclient.network_trunk_create(trunk_port_params)You can use the run_command to to supply the command (using the underscore version of the command) and a dictionary options and it's parameters.
This can come in handy in scenarios where it's quite possible that other openstackclient plugins have been installed after the clientshell has been initialized. But you still want to take advantage of supplying the command options and arguments as a dictionary.
from ospclientsdk import ClientShell
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
# Build a dictionary of the required arguements/options and their parameters.
# Each key is command option and the value is the parameter.
trunk_port_params = dict(name="test_compute_trunk",
parent_port="test_port",
subport=[dict(port="dummy_port",
segmentation_type="vlan",
segmentation_id=2007
)]
)
# To create a network trunk port using the above defined dict I
# use I pass in the command as a string and the options as a dictionaru
results = shell.run_command('network_trunk_create', trunk_port_params)You can use the function run_raw_command to explicitly run full commands. This will allow you to pass in a string directly.
This commmand might be useful if there are problems with the clientshell normalizing the options or if you want to leverage the underlying shell to chain together commands and do some complex processing without having to setup subprocess boilerplate code yourself.
Note since we aren't doing any string checking make sure you aren't passing in anything unsafe for the underlying shell to execute. Also make sure to escape your command properly.
from ospclientsdk import ClientShell
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
# Pass in a string representing a string command
results = shell.run_raw_command('openstack server show -f json')When the ospclientsdk executes it will return back to you a dictionary. It will contain the following keys:
- rc
- stdout
- stderr
You can use the rc value to determine if the command passed (0) or failed (1) and how to respond to either scenario. The stdout key will always be deserialized into a dictionary so that it can be easily manipulated and accessed. The stderr will typically always be a text string.
There are use cases where it's useful to run the commands remotely on a test driver or bastion host.
The SDK provides a remote_shell context that can forward all CLI commands to the desired host without
a lot of extra boilerplate code. Before using this context first install the necessary ssh package:
$ pip install ospclientsdk['remote_shell']Next it's recommended you will load your ssh agent up with a key appropriate to the target client host.
Then you can start using it. Below are some examples
from ospclientsdk import ClientShell, remote_shell
shell = ClientShell(cloud_file='clouds.yaml', cloud='test')
# Build a dictionary of the required arguements/options and their parameters.
# Each key is command option and the value is the parameter.
server_params = dict(res="ccit_test_client",
image="rhel-7.5-server-x86_64-released",
flavor="m1.small",
key_name="db2-test",
network=["test_private_network"],
wait=True
)
# Here is an example where we log into a remote host using user 'cloud-user' and a key file
# using the high level api
with remote_shell(hostname="10.x.x.x", username="cloud-user", key_file="keys/test-key"):
results = shell.compute.server_create(server_params)
# Here is an example running raw commands to a remote host using the FQDN and 'fedora' user
# but we've specified a client path for where to look for the 'openstack' binary since it is
# not in the $PATH on the remote host. This comes in useful if you've installed the client
# in a virtualenv on the remote host.
with remote_shell(hostname="host.example.com", username="fedora", key_file="keys/test-key", client_path="/home/fedora/cbn/bin"):
results = shell.run_raw_command('openstack server show -f json')
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent