Python client for UpCloud's API
Python

README.md

UpCloud's Python API Client

OOP-based api client for UpCloud's API. Features most of the API's functionality and some convenience functions that combine several API endpoints and logic.

NOTE: This Python client is still evolving. Please test all of your use cases thoroughly before actual production use. Using a separate UpCloud account for testing / developing the client is recommended.

Installation

pip install upcloud-api

Alternatively, if you want the newest master or a devel branch - clone the project and run:

python setup.py install

!! SSL security update for python 2 !!

Supported versions as of 0.3.3 (offline tests pass with tox):

  • python 2.6
  • python 2.7
  • python 3.2 removed due to python2/3 support
  • python 3.3
  • python 3.4
  • python 3.5
  • pypi3 2.4.0

Features

  • OOP based management of Servers, Storages and IP-addresses with full CRUD.
    • since 0.2: manage both IPv4 and IPv6 addresses
    • since 0.1.1: can use custom storage templates in addition to public templates
  • Clear way to define your infrastructure, emphasis on clear and easy syntax
  • Access all the data of the objects ( e.g. ssh credentials )
  • Scale horizontally by creating / destroying servers
  • Scale vertically by changing the RAM, CPU, storage specs of any server
  • Manage firewall (on/off and individual rules)
    • since 0.2: full management of firewall rules

TODO:

  • Cloning of storages
  • Full management of special storage types:
    • CDROMs, custom OS templates
    • (custom templates can already be cloned to a disk via UUID)
  • Full management of backups (instant and scheduled)

Changelog:

Documentation:

Examples

Note that some operations are not instant, for example a server is not fully shut down when the API responds. You must take this into account in your automations.

Defining and creating Servers

import upcloud_api
from upcloud_api import Server, Storage, ZONE, login_user_block

manager = upcloud_api.CloudManager('api_user', 'password')
manager.authenticate()


login_user = login_user_block(
    username='theuser',
    ssh_keys=['ssh-rsa AAAAB3NzaC1yc2EAA[...]ptshi44x user@some.host'],
    create_password=False
)

cluster = {
    'web1': Server(
        core_number=1, # CPU cores
        memory_amount=512, # RAM in MB
        hostname='web1.example.com',
        zone=ZONE.London, # ZONE.Helsinki and ZONE.Chicago available also
        storage_devices=[
            # OS: Ubuntu 14.04 from template
            # default tier: maxIOPS, the 100k IOPS storage backend
            Storage(os='Ubuntu 14.04', size=10),
            # secondary storage, hdd for reduced cost
            Storage(size=100, tier='hdd')
        ],
        login_user=login_user  # user and ssh-keys
    ),
    'web2': Server(
        core_number=1,
        memory_amount=512,
        hostname='web2.example.com',
        zone=ZONE.London,
        storage_devices=[
            Storage(os='Ubuntu 14.04', size=10),
            Storage(size=100, tier='hdd'),
        ],
        login_user=login_user
    ),
    'db': Server(
        plan='2xCPU-2GB', # use a preconfigured plan, instead of custom
        hostname='db.example.com',
        zone=ZONE.London,
        storage_devices=[
            Storage(os='Ubuntu 14.04', size=10),
            Storage(size=100),
        ],
        login_user=login_user
    ),
    'lb': Server(
        core_number=2,
        memory_amount=1024,
        hostname='balancer.example.com',
        zone=ZONE.London,
        storage_devices=[
            Storage(os='Ubuntu 14.04', size=10)
        ],
        login_user=login_user
    )
}

for server in cluster:
    manager.create_server(cluster[server]) # automatically populates the Server objects with data from API

New in 0.3.0: servers can now be defined as dicts without using Server or Storage classes. The syntax/attributes are exactly like above and under the hood they are converted to Server and Storage classes. This feature is mainly for easier usage of the module from Ansible, but may provide useful elsewhere.

Stop / Start / Destroy Servers

for server in cluster:
    server.shutdown()
    # OR:
    server.start()
    # OR:
    server.destroy()
    for storage in server.storage_devices:
      storage.destroy()

New in 0.3.0: as the success of server.start() or server.destroy() and storage.destroy() depend on the Server's state, new helpers have been added. The helpers may be called regardless of the server's current state.

# makes sure that the server is stopped (blocking wait) and then destroys the server and its storages
server.stop_and_destroy()

# makes sure that the server is started (blocking wait)
server.ensure_started()

Upgrade a Server

server = cluster['web1']
server.shutdown()
server.core_number = 4
server.memory_amount = 4096
server.save()
server.start()

Easy access to servers and their information:

New in 0.3.0.

# returns a public IPv4 (preferred) IPv6 (no public IPv4 was attached) address
server.get_public_ip()

# returns a JSON serializable dict with the server's information (storages and ip-addresses included)
server.to_dict()

GET resources:

servers     = manager.get_servers()
server1     = manager.get_server(UUID) # e.g servers[0].uuid
storages    = manager.get_storages()
storage1    = manager.get_storage(UUID) # e.g sever1.storage_devices[0].uuid
ip_addrs    = manager.get_ips()
ip_addr     = manager.get_ip(address) # e.g server1.ip_addresses[0].address

Tests

Set up environment and install dependencies:

# run at project root, python3 and virtualenv must be installed
virtualenv ENV
source ENV/bin/activate
pip install -r requirements.txt

Install the package in editable mode, as mentioned in https://pytest.org/latest/goodpractises.html

# run at project root
pip install -e .

Tests located in project_root/tests/ directory. Run with:

py.test tests/

To test against all supported python versions, run:

tox

The project also supplies a small test suite to test against the live API at test/live_test.py. This suite is NOT run with py.test as it will permanently remove all resources related to an account. It should only be run with a throwaway dev-only account when preparing for a new release. It is not shipped with PyPI releases. See source code on how to run the live tests.

Bugs, Issues, Problems, Ideas

Feel free to open a new issue : )

Documentation

Documentation available here