Rough implementation of a data-source which uses vSphere / VMware VMX guestinfo for cloud-init.

DataSourceVMware.py

''' Provides a VMware VMX backed Data Source for Cloud-Init. '''

import yaml

from time import sleep
from base64 import b64decode
from subprocess import check_output
from subprocess import CalledProcessError

from cloudinit import log as logging
from cloudinit import sources
from cloudinit import util

# Define how long to wait for a value to appear in VMX, and how often to 
# wait between polls if it is not yet available.
VMX_TIMEOUT = 300
VMX_INTERVAL = 10
VMX_COMMAND = 'vmtoolsd'

LOG = logging.getLogger(__name__)


class DataSourceVMware(sources.DataSource):
    ''' Implements a VMware VMX backed data source. '''
    dsname = 'VMware'

    def _get_vmx_value(self, key, command=VMX_COMMAND):
        ''' Attempt to get a given key from VMX using vmtoolsd. '''
        vmx_timer = 0
        while vmx_timer <= VMX_TIMEOUT:
            try:
                result = check_output(
                    [command, '--cmd', '"info-get guestinfo.{}"'.format(key)]
                )
            except CalledProcessError:
                vmx_timer += VMX_INTERVAL
                sleep(VMX_INTERVAL)

        # If a result failed to be retrieved in the allotted time period, the
        # following operation will raise a NameError. If it did return but was
        # blank, just return a blank value.
        try:
            if result and not raw:
                return b64decode(result)
            else:
                return result
        except TypeError:
            raise AttributeError(
                'Decode failed, guestinfo.{} from VMX may be corrupt!'.format(
                    key
                )
            )
        except NameError:
            raise AttributeError(
                'guestinfo.{} was unable to be read from VMX in time!'.format(
                    key
                )
            )

    def get_data(self):
        ''' Attempt to pull data from VMX. '''
        LOG.debug('Attempting to get cloud-init userdata from VMX channel')
        self.userdata_raw = self._get_vmx_value('userdata')

        LOG.debug('Attempting to get cloud-init metadata from VMX channel')
        self.metadata = yaml.load(self._get_vmx_value('metadata'))

        LOG.debug('Attmpting to get cloud-init network config from VMX channel')
        self._network_config = yaml.load(self.get_vmx_value('networkconfig'))

    def get_instance_id(self):
        ''' Fetch the instance-id from the captured metadata. '''
        return self.metadata['instance-id']

    def get_hostname(self, fqdn=False):
        ''' Fetch the hostname from the captured metadata. '''
        if fqdn:
            return self.metadata['hostname']
        else:
            return self.metadata['hostname'].split('.')[0]

    @property
    def network_config(self):
        ''' Provides network configuration from the fetched data. '''
        return self._network_config

    @property
    def availability_zone(self):
        ''' Provides availability zone information from the fetched data. '''
        return None

    @property
    def region(self):
        ''' Provides region from information the fetched data. '''
        return None

datasources = [
    (DataSourceVMware, (sources.DEP_FILESYSTEM,)),
]

def get_datasource_list(depends):
    return sources.list_from_depends(depends, datasources)

For what it's worth, this is a rough semi-working implementation. Everything needed to make this work is here, but it needs quite a bit of modification to drop into cloud-init and use for machine provisioning :)