Client

A python wrapper for the fleet v1 API.

>>> import fleet.v1 as fleet

# connect over tcp
>>> fleet_client = fleet.Client('http://127.0.0.1:49153')

# or over a unix domain socket
>>> fleet_client = fleet.Client('http+unix://%2Fvar%2Frun%2Ffleet.sock')

# via an ssh tunnel
>>> fleet_client = fleet.Client('http://127.0.0.1:49153', ssh_tunnel='198.51.100.23:22')

Client(self, endpoint, http=None, ssh_tunnel=None, ssh_username='core', ssh_timeout=10, ssh_known_hosts_file='~/.fleetctl/known_hosts', ssh_strict_host_key_checking=True, ssh_raw_transport=None)

Connect to the fleet API and generate a client based on it's discovery document.

Arguments

  • endpoint (str): A URL where the fleet API can be reached. Supported schemes are:

    • http: A HTTP connection over a TCP socket. http://127.0.0.1:49153
    • http+unix: A HTTP connection over a unix domain socket. You must escape the path (/ = %2F). http+unix://%2Fvar%2Frun%2Ffleet.sock

  • http (httplib2.Http): An instance of httplib2.Http (or something that acts like it) that HTTP requests will be made through. You do not need to pass this unless you need to configure specific options for your http client, or want to pass in a mock for testing.

  • ssh_tunnel (str '\<host>[:\<port>]'): Establish an SSH tunnel through the provided address for communication with fleet. Defaults to None. If specified, the following other options adjust it's behaivor:

    • ssh_username (str): Username to use when connecting to SSH, defaults to 'core'.
    • ssh_timeout (float): Amount of time in seconds to allow for SSH connection initialization before failing, defaults to 10.
    • ssh_known_hosts_file (str): File used to store remote machine fingerprints, defaults to '~/.fleetctl/known_hosts'. Ignored if ssh_strict_host_key_checking is False
    • ssh_strict_host_key_checking (bool): Verify host keys presented by remote machines before initiating SSH connections, defaults to True.

  • ssh_raw_transport (paramiko.transport.Transport): An active Transport on which open_channel() will be called to establish connections. See Advanced SSH Tunneling for more information.

Raises

  • ValueError: The endpoint provided was not accessible.

Advanced SSH Tunneling

If your ssh connection requires complex configuration, you can configure and connect() your own paramiko.client.Client and pass the result of get_transport() as ssh_raw_transport

If ssh_raw_transport is set all other ssh options are ignored, it's assumed the caller will have fully configured and connected their ssh transport before invoking us.

# example of configuring and connecting your own ssh client
# this contrived example uses a specific key, 
# and disables the use of the agent

import fleet.v1 as fleet
import paramiko

# configure our client
ssh_client = paramiko.SSHClient()
ssh_client.set_missing_host_key_policy(paramiko.AutoAddPolicy())

# connect to the host with our custom configuration
ssh_client.connect(
    hostname='198.51.100.23',
    username='core', 
    key_filename='/tmp/the.key',
    allow_agent=False,
    look_for_keys=False
)

# pass the transport to the client
fleet_client = fleet.Client(
    'http://127.0.0.1:49153', 
    ssh_raw_transport=ssh_client.get_transport()
)

Methods

create_unit()

Create a new Unit in the cluster

>>> fleet_client.create_unit('foo.service', fleet.Unit(from_file='foo.service'))
<Unit: {u'desiredState': u'launched', u'name': u'foo.service', u'currentState': u'inactive', u'options': [{u'section': u'Service', u'name': u'ExecStart', u'value': u'/usr/bin/sleep 1d'}]}>

create_unit(self, name, unit)

  • name (str): The name of the unit to create
  • unit (Unit): The unit to submit to fleet

Returns

  • Unit: The unit that was created

Raises

  • APIError: Fleet returned a response code >= 400

set_unit_desired_state()

Update the desired state of a unit running in the cluster.

>>> unit = fleet_client.create_unit('foo.service', fleet.Unit(from_file='foo.service'))
>>> unit.name
u'foo.service'
>>> unit.desiredState
u'launched'

# reference the object directly
>>> fleet_client.set_unit_desired_state(unit, 'inactive')
<Unit: {u'desiredState': u'inactive', u'name': u'foo.service', u'currentState': u'inactive', u'options': [{u'section': u'Service', u'name': u'ExecStart', u'value': u'/usr/bin/sleep 1d'}]}>

# or it's name
>>> fleet_client.set_unit_desired_state('foo.service', 'inactive')
<Unit: {u'desiredState': u'inactive', u'name': u'foo.service', u'currentState': u'inactive', u'options': [{u'section': u'Service', u'name': u'ExecStart', u'value': u'/usr/bin/sleep 1d'}]}>

# APIError raised for an invalid unit name
>>> fleet_client.set_unit_desired_state('invalid-service', 'inactive')
fleet.v1.errors.APIError: unit does not exist and options field empty (409)

# ValueError raised for an invalid state
>>> fleet_client.set_unit_desired_state('foo.service', 'invalid-state')
ValueError: state must be one of: ['inactive', 'loaded', 'launched']

set_unit_desired_state(self, unit, desired_state)

  • unit (str, Unit): The Unit, or name of the unit to delete
  • desired_state (str): State the user wishes the Unit to be in ("inactive", "loaded", or "launched")

Returns

  • Unit: The updated unit

Raises

  • APIError: Fleet returned a response code >= 400
  • ValueError: An invalid value was provided for desired_state

destroy_unit()

Delete a unit from the cluster

# delete by passing the unit
>>> fleet_client.destroy_unit(unit)
True

# or it's name
>>> fleet_client.destroy_unit('foo.service')
True

# APIError is raised if the unit does not exist
>>> fleet_client.destroy_unit('invalid-service')
fleet.v1.errors.APIError: unit does not exist (404)

destroy_unit(self, unit)

  • unit (str, Unit): The Unit, or name of the unit to delete

Returns

  • True: The unit was deleted

Raises

  • APIError: Fleet returned a response code >= 400

list_units()

Returns a generator that yields each Unit in the cluster

>>> for unit in fleet_client.list_units():
...     unit
... 
<Unit: {u'machineID': u'2901a44df0834bef935e24a0ddddcc23', u'desiredState': u'launched', u'options': [{u'section': u'Service', u'name': u'ExecStart', u'value': u'/usr/bin/sleep 1d'}], u'currentState': u'launched', u'name': u'foo.service'}>

list_units(self):

Yields:

  • Unit: The next Unit in the cluster

Raises

  • APIError: Fleet returned a response code >= 400

get_unit()

Retreive a specific Unit from the cluster by name.

# get a service by name
>>> fleet_client.get_unit('foo.service')
<Unit: {u'machineID': u'2901a44df0834bef935e24a0ddddcc23', u'desiredState': u'launched', u'options': [{u'section': u'Service', u'name': u'ExecStart', u'value': u'/usr/bin/sleep 1d'}], u'currentState': u'launched', u'name': u'foo.service'}>

# APIError raised for invalid service names
>>> fleet_client.get_unit('invalid-service')
fleet.v1.errors.APIError: unit does not exist (404)

get_unit(self, name)

  • name (str): If specified, only this unit name is returned

Returns

  • Unit: The unit identified by name

Raises

  • APIError: Fleet returned a response code >= 400

list_unit_states()

Returns a generator that yields the current UnitState for each unit in the cluster

>>> for unit_state in fleet_client.list_unit_states():
...     unit_state
... 
<UnitState: {"hash": "dd401fa78c2de99a9c4045cbb4b285679067acf6", "name": "foo.service", "machineID": "2901a44df0834bef935e24a0ddddcc23", "systemdSubState": "running", "systemdActiveState": "active", "systemdLoadState": "loaded"}>

list_unit_states(self, machine_id = None, unit_name = None)

  • machine_id (str): filter all UnitState objects to those originating from a specific machine
  • unit_name (str): filter all UnitState objects to those related to a specific unit

Yields

  • UnitState: The next UnitState in the cluster

Raises

  • APIError: Fleet returned a response code >= 400

list_machines()

Return a generator that yields each Machine in the cluster

>>> for machine in fleet_client.list_machines():
...     machine
... 
<Machine: {"primaryIP": "198.51.100.23", "id": "2901a44df0834bef935e24a0ddddcc23", "metadata": {}}>

list_machines(self)

Yields

  • Machine: The next machine in the cluster

Raises

  • APIError: Fleet returned a response code >= 400