rsnyman / wrapanapi

A base system for provider management

wrapanapi

Introduction

wrapanapi is a simple virtualization client with support (in varying degrees) for the following

• Red Hat Enterprize Virtualization (RHEV)
• Red Hat Openstack (RHOS)
• Openshift
• VMware vCenter
• Microsoft SCVMM
• Microsoft Azure
• Google Compute Engine
• Kubernetes
• Hawkular
• Amazon EC2

It can be used to control virtualization clients and offers operations such as

• list_vm (returns a list of vm_names)
• list_template (returns a list of templates/images)
• start_vm (starts a vm)
• stop_vm (stops a vm)

Though conceptually names differ between cloud and infrastructure virtualization technologies (e.g. instance/vm) it was decided to stick to one representation in the codebase and interface to give a singlar API across all systems.

Usage

Each management system is invoked usually with a hostname and some credentials

from wrapanapi.virtualcenter import VMWareSystem

system = VMWareSystem(hostname='10.0.0.0', username="root", password="password")
system.list_vm()

Adding a new Management System

A management system should extend the Base class, and supply "Not Implemented" raises for items which it doesn't support. This behaviour may change in the future as more and more diverse management systems.

from base import WrapanapiAPIBase

class RHEVMSystem(WrapanapiAPIBase):

  _stats_available = {
    'num_vm': lambda self: self.api.get_summary().get_vms().total,
    'num_host': lambda self: len(self.list_host()),
    'num_cluster': lambda self: len(self.list_cluster()),
    'num_template': lambda self: len(self.list_template()),
    'num_datastore': lambda self: len(self.list_datastore()),
  }

  def __init__(self, hostname, username, password, **kwargs):
    super(RHEVMSystem, self).__init__(kwargs)

The call to super is necessary to set up the logger if noe has not been passed in with the logger keyword.

The developer can then add their own methods to interact with their own management system. Commonly accessible statistics are generally all named the same across management systems. In this way we can treat multiple management systems the same and use an identical method to check the number of vms on a RHEV system, to a VMware system.

Exceptions currently sit in a single module, this will probably change later with each management system having it's own package and exceptions stored there.

This module was originally developed for assisting in the ManageIQ testing team.

Contributing

The guidelines to follow for this project can be found in the cfme dev_guide.