Cumulus Linux (NVUE) — Management

Day-2 inventory, facts and config management for Cumulus Linux (NVUE) data center fabrics (VXLAN/EVPN). Part of the multi-vendor Data Center automation guide.

Repository

Cumulus

NVUE

REST API

VXLAN / EVPN

Model the fabric once in a vendor-neutral inventory

Collect facts from Cumulus via API/SSH

Push config from intent, never hand-edit per box

Setup, key commands & run examples

inventory/fabric.yml

Inventory

Describe each Cumulus device once — role, mgmt IP and VTEP — so the same workflow targets it and every other vendor in the fabric.

inventory/fabric.yml

yaml

leaf21:
  role: leaf
  vendor: cumulus
  mgmt: 10.0.0.21
  vtep: 10.255.0.21
  napalm: null   # API/gNMI driven

connect.py

Connect

Open a session to a Cumulus Linux (NVUE) device. This platform is API/gNMI driven, so use its native client.

connect.py

python

import requests

# Cumulus Linux (NVUE) — NVUE REST API
base = "https://10.0.0.23:8765/nvue_v1"
auth = ("cumulus", "secret")
r = requests.get(f"{base}/interface", auth=auth, verify=False)
print(r.json())

facts.sh

Day-2 facts

The day-2 management read-outs you will automate on Cumulus: version, EVPN peers and interface state.

facts.sh

bash

# inventory & health snapshot — Cumulus Linux (NVUE)
nv show system
nv show interface
nv show vrf default router bgp neighbor

Best practices

One intent model, many renderers
Keep a single fabric description and render Cumulus config from it — never SSH in to edit CLI by hand.
Prefer structured APIs
Use the Cumulus structured API (NVUE REST) instead of screen-scraping CLI.
Version your inventory
Treat fabric.yml as the source of truth in git; every change is a reviewable PR.
Idempotent runs
Management scripts should be safe to re-run — collect, compare, only change what drifted.

Hands-on exercises

Guided practice tasks — copy the starter snippet, run it against a lab device, and extend it to meet the goal.

Exercise 1

Cumulus inventory report

Goal: Connect to the leaf, pull version + model and print one inventory row.

exercise_1

python

import requests

# Cumulus Linux (NVUE) — NVUE REST API
base = "https://10.0.0.23:8765/nvue_v1"
auth = ("cumulus", "secret")
r = requests.get(f"{base}/interface", auth=auth, verify=False)
# TODO: print hostname, model and OS version

Exercise 2

Interface drift check

Goal: Compare live interface admin/oper state against the intended list and report mismatches.

exercise_2

python

intended = {"Ethernet1", "Ethernet2"}
# TODO: read live interfaces and report any intended port that is down

Curated videos & guides

Checklist — pitfalls, gotchas & verification

Common pitfalls

•Screen-scraping Cumulus CLI when a structured API exists — output formats change between releases.
•Storing credentials in the script instead of a vault / env vars.
•Editing boxes by hand so the inventory no longer matches reality.

Gotchas to watch

•Cumulus needs its API enabled (NVUE REST) before automation can connect.
•Management VRF: make sure your reachability is over the mgmt VRF, not the data plane.

Verify it works

•Re-run the report and confirm every device returns facts.
•Confirm the inventory row matches the physical device (model, serial, OS).