nutanix / docker-machine

Rancher Node Driver for Nutanix AHV
https://www.nutanix.com/products/acropolis/virtualization
Mozilla Public License 2.0
13 stars 17 forks source link

Cannot create clusters using UI in Rancher 2.9.x #124

Open nnoethling opened 2 months ago

nnoethling commented 2 months ago

In the current Rancher version 2.9.1 the Nutanix node driver is not working out of the box via UI. When trying to create a new cluster with a network the UI show an validation Error and its not possible to create the cluster.

Nutanix Cluster Information

Current versions

Rancher Version

v2.9.1

Rancher Node Driver Version

v3.6.0

Node template

In Rancher version 2.8.x the vmNetwork field was an array field with multiple networks:

image

In 2.9.1 vmNetwork is not a field that can hold multiple networks

image

Debug Output

No logs visible in Rancher Pods regarding that issue. UI Error message says: NutanixConfig.rke-machine-config.cattle.io "nc-test-pool1-xwwj9" is invalid: vmNetwork: Invalid value: "string": vmNetwork in body must be of type array: "string" Even with empty Network field its not possible to create the Cluster.

Request response:

{
    "type": "error",
    "links": {},
    "code": "Invalid",
    "message": "NutanixConfig.rke-machine-config.cattle.io \"nc-test-pool1-vcggj\" is invalid: vmNetwork: Invalid value: \"string\": vmNetwork in body must be of type array: \"string\"",
    "status": 422,
    "type": "error"
}

Cluster Creation is working via helm chart or CRD. I think this issue is based on a behaviour change in Rancher UI because vmNetwork and vmCategories are array values and are not displayed correctly as array field in Rancher.

Expected Behavior

Cluster can be deployed via UI

Actual Behavior

Cannot create cluster because Error is blocking creation with Error message NutanixConfig.rke-machine-config.cattle.io "nc-test-pool1-xwwj9" is invalid: vmNetwork: Invalid value: "string": vmNetwork in body must be of type array: "string"

tuxtof commented 2 months ago

Hello @nnoethling

this is due to Rancher change between RKE1 and RKE2

with RKE2 you need to use the nutanix rancher extension to correctly specific network with RKE1 it was part of the almost built-in UI

In all cases, for the moment, there are some issues with Rancher 2.9 and the external plugin that cause trouble with the entire process to deploy RKE2 cluster on Nutanix with the UI

we are actually working with Rancher, to find a fix

stay tuned