Netbox Device Sync Selection

[asergeant01]

Currently, the selection of which devices to sync from netbox is driven from the ClusterImport custom resource.

This resource's most atomic unit is "clusters", whereby the list of devices to sync is extracted from queries and filters on the netbox /api/virtualization/clusters/ api.

In order to provide more flexibility for other use cases, I would purpose changing the atomic unit to "devices", whereby the list of devices to sync is extracted from queries and filters on the netbox /api/dcim/devices/ api.

A purposed CR is supplied for illustrative purposes.

kind: DeviceImport
metadata:
  labels:
    app.kubernetes.io/name: argora
    app.kubernetes.io/managed-by: kustomize
  name: deviceimport-sample
  namespace: argora-system
spec:
  devices:
  - name: "xxx"
    site: "xxx"
    role: "xxx"
    type: "xxx"
    status: "xxx"
    cluster:
       name: "cc-b0-qa-de-1"
       region: "qa-de-1"
       type: "cc-kvm-compute"
    tags:
      - "tag1"
      - "tag2"

[defo89]

Using clusters as the import unit is intentional — it keeps the data model consistent and is aligned with how infrastructure is actually managed. Any device that is to be used in metal automation is part of a specific cluster (+ cluster is of given type).

In the later stage we would not even need to filter for names but rather just search for cluster type in a region.

For testing purposes we could add DeviceImport with a name filter but I do not see a clear benefit of changing the atomic unit to Device.

Is there any specific use case where you are limited with working on a cluster basis?

cc @SchwarzM