A pythonic wrapper for Proxmox REST API

What does it do and what's different?

Proxmoxia is a wrapper around the Proxmox REST API . It is intended to be used by administrators and users who need to access information about their proxmox cluster which they might otherwise have read/scraped from the web UI.

Rather than writing wrappers for each individual end point and method combination in the API we've instead created a dynamic method call which responds to the attribute you've attempted to reach.

We'll outline how this works later in the doc.

First Steps

Connecting to Proxmox

The first thing to do is import the proxmox library and create a Connection to the server. This connection will be used to authenticate ourselves and can be any of the nodes in the cluster:

import proxmox

PROXMOX_HOST = "proxmox-1"  # can also be an IP or the FQDN of the host

connection = proxmox.Connector(PROXMOX_HOST, PROXMOX_PORT)

Once you have connected you should use the connection to retrieve an AuthenticationToken:

auth_token = connection.get_auth_token('user@pam', 'strawberries')

This is stored on your connection object for later use, but also returned for you to inspect.


Due to the __repr__ method which is on the AuthToken object returned, unless you assign it to a variable it will be printed out into your stdout.

Simple query based access

Queries are exposed via the access methods get, post, put and delete. The filter path is relative to the api root passed to the connection constructor:

for node in connection.get('nodes'):
    for vm in connection.get('nodes/%s/openvz' % node['node']):
        print "%s. %s => %s" % (vm['vmid'], vm['name'], vm['status'])

>>> 141. puppet-2.london.baseblack.com => running
    101. munki.london.baseblack.com => running
    102. redmine.london.baseblack.com => running
    140. dns-1.london.baseblack.com => running
    126. ns-3.london.baseblack.com => running
    113. rabbitmq.london.baseblack.com => running

Attribute based access

Having to continually build up your url in bits is a little ass. So we also have an attribute call based mechanism for reaching endpoints.

When an attribute is accessed on a proxmox.Proxmox or proxmox.Node object it will generate a new AttrMethod based object. These methods are recursive so p.cluster.config() will generate a valid GET request.


proxmox.Node is a convience class provided to wrap up proxmox.Proxmox(c).nodes(nodename). It may also be used to extend the class with an execution method. (_TBD_).

Get/Post/Put and Delete requests

The default request made when an attribute is called is a GET request. So p.nodes() will create a GET attrMethod and call it with no arguments. The url it will generate will be:


Another way of specifying a Get request should be made is to call .get() on the attribute you want. i.e: p.nodes.get().

To request a Post/Put or Delete, append the call instead with the matching name:


Named attributes

Some attributes such as a node name or virtual machine id which you do not know before you start writing your code are harder to address as a simple attribute. You could use an inline eval, but they stick out like a sore thumb in this context.

So to handle these, each AttrMethod accepts a non-keyword argument which it will use to generate a the new AttrMethod for.

For example these lines are functionally identical:

result = node.openvz.108.status.current()

vmid = 108
result = eval('node.openvz.%d.status.current()' % vmid)
result = node.openvz(vmid).status.current()

Request Parameters + Arguments

Some of the API endpoints require a number of keyword arguments. For these address each as a named parameter in your attribute call. In this example node.rrd(ds="cpu", timeframe="hour") will generate a Get request encoding the arguments into a url like this:


node.rrd.post(ds="cpu", timeframe="hour") would generate a Post request with a url/post fields like this:

url = http://SERVER:PORT/api2/json/nodes/proxmox-7/rrd?
post_fields = ds=cpu&timeframe=hour

Put/Delete requests are the same as posts.


  • Create Proxmox and Node access objects
p = p = proxmox.Proxmox(connection)
node = proxmox.Node(connection, 'proxmox-7')
  • Connect to a access end points on a node
print node.rrd(ds='cpu',timeframe='hour')
>>>{u'filename': u'/var/lib/rrdcached/db/pve2-node/proxmox-7.png'}
  • Requests the status on a vm with id number 108
vmid = 108
print node.openvz(vmid).status.current()
>>> {u'status': u'stopped', u'uptime': 0, u'disk': 0, u'maxswap': 536870912, u'name': u'packages.london.baseblack.com', u'diskread': 0, u'diskwrite': 0, u'ip': u'', u'netin': 0, u'cpus': 1, u'mem': 0, u'failcnt': 0, u'swap': 0, u'nproc': 0, u'netout': 0, u'ha': 0, u'type': u'openvz', u'cpu': 0, u'maxdisk': 4294967296, u'maxmem': 536870912}
  • Find a vm template filepath and use this as the ostemplate for a new vm you create. And then start it up once it has finished being created:
# find the path for the template you want to use.
for template in node.storage('virtual-nfs').content(content='vztmpl'):
    if re.match('.*ubuntu-12.04-bb-20121010b_amd64.tar.gz$', template['volid']):
        volume = node.storage('virtual-nfs').content(template['volid']).get()

# create the container, giving it some sensible settings
taskid = node.openvz.post( ostemplate=volume.get('path'),

# keep an eye on task and see when its completed
while node.tasks(taskid).status()['status'] == 'running':

# print out the logs
for line in node.tasks(taskid).log():
    print line['t']

    # start up the container
    raise Exception('Unable to start container')
  • Find if a user exists and create them if they do not
if 'andrew.bunday@pve' not in [x['userid'] for x in p.access.users()]:
    p.access.users.post(userid='andrew.bunday@pve', comment="test user", password="strawberries")