denvr

A Python SDK for interacting with Denvr Cloud.

This package is largely autogenerated from our Denvr Cloud API Spec

Installation

pip install -e git+https://github.com/denvrdata/denvrpy.git

Quickstart

Getting started with the denvr python sdk just involves loading and calling the client builder function, which returns a Client object for each denvr service (e.g., clusters, vpcs, servers/virtual).

>>> import json
>>> from denvr.client import client
>>> virtual = client('servers/virtual')

Lets start by fetching some of Denvr's VM configurations. We'll use json.dumps to make the output a bit easier to read.

>>> print(json.dumps(virtual.get_configurations(), indent=2))
{
  "items": [
    {
      "id": 5,
      "user_friendly_name": "A100_40GB_PCIe_1x",
      "name": "A100_40GB_PCIe_1x",
      "description": null,
      "os_version": "20.04",
      "os_type": "Ubuntu",
      "storage": 1700,
      "gpu_type": "nvidia.com/A100PCIE40GB",
      "gpu_family": "NVIDIA A100",
      "gpu_brand": "Nvidia",
      "gpu_name": "A100 40GB PCIe",
      "type": "nvidia.com/A100PCIE40GB",
      "brand_family": "NVIDIA A100",
      "brand": "Nvidia",
      "text_name": "A100 40GB PCIe",
      "gpus": 1,
      "vcpus": 14,
      "memory": 112,
      "price": 2.05,
      "compute_network": null,
      "is_gpu_platform": true,
      "clusters": [
        "Hou1",
        "Msc1"
      ]
    },
    {
      "id": 6,
      "user_friendly_name": "A100_40GB_PCIe_2x",
      "name": "A100_40GB_PCIe_2x",
      "description": null,
      "os_version": "20.04",
      "os_type": "Ubuntu",
      "storage": 3400,
      "gpu_type": "nvidia.com/A100PCIE40GB",
      "gpu_family": "NVIDIA A100",
      "gpu_brand": "Nvidia",
      "gpu_name": "A100 40GB PCIe",
      "type": "nvidia.com/A100PCIE40GB",
      "brand_family": "NVIDIA A100",
      "brand": "Nvidia",
      "text_name": "A100 40GB PCIe",
      "gpus": 2,
      "vcpus": 28,
      "memory": 224,
      "price": 4.1,
      "compute_network": null,
      "is_gpu_platform": true,
      "clusters": [
        "Hou1",
        "Msc1"
      ]
    },
    ...

Next we'll try Creating a new virtual machine:

>>> print(json.dumps(
    virtual.create_server(
        name="api-test",
        rpool="on-demand",
        vpc="denvr",
        configuration="H100_80GB_SXM_8x",
        cluster="Hou1",
        ssh_keys=["ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAA..."],
        operating_system_image="Ubuntu 22.04.4 LTS",
        root_disk_size=500,
    ),
    indent=2
))
{
  "username": "rory@denvrdata.com",
  "tenancy_name": "denvr",
  "rpool": "on-demand",
  "direct_attached_storage_persisted": false,
  "id": "api-test",
  "namespace": "denvr",
  "configuration": "H100_80GB_SXM_8x",
  "storage": 20000,
  "gpu_type": "nvidia.com/H100SXM480GB",
  "gpus": 8,
  "vcpus": 200,
  "memory": 940,
  "ip": "",
  "private_ip": "172.16.0.84",
  "image": "Ubuntu_22.04.4_LTS",
  "cluster": "Hou1",
  "status": "na",
  "storage_type": "na"
}

Similarly, we can also fetch info about a specific vm.

>>> print(json.dumps(virtual.get_server(id='rofinn-intel-dev', namespace='denvr', cluster='Hou1'), indent=2))
{
  "username": "rory@denvrdata.com",
  "tenancy_name": "denvr",
  "rpool": "reserved-denvr",
  "direct_attached_storage_persisted": false,
  "id": "rofinn-intel-dev",
  "namespace": "denvr",
  "configuration": "15",
  "storage": 27000,
  "gpu_type": "intel/GAUDI2",
  "gpus": 8,
  "vcpus": 152,
  "memory": 940,
  ...
  "image": "habana-1.16.2",
  "cluster": "Hou1",
  "status": "ONLINE",
  "storage_type": "na"
}