Editing an VMware NSX-T Data Center Edge Node results in: Transport node refresh failed

Products

• VMware NSX

Issue/Introduction

Symptoms:

• In System - Fabric - Transport Nodes - Edge Nodes, when you attempt to edit the Edge node, you are presented with the following alert:

"Transport node refresh failed: [Fabric] Refresh edge <edge uuid> placement configuration failed. Check network connectivity of the edge node"

• And when tried to sync the edge node configuration, we see the error as:

" Failed to refresh the transport node configuration:[fabric] refresh edge<edge uuid> placement configuration failed. check network connectivity of the edge node.

• The vCenter (Compute Manager) which this Edge node was deployed to, no longer exists as a Compute Manager in NSX-T.
• In the NSX-T manager log /var/log/proton/nsxapi.log we see ERROR:

ERROR task-scheduler-1 EdgeNodeFabricUtils 23929 FABRIC [nsx@6876 comp="nsx-manager" errorCode="MP16047" level="ERROR" subcomp="manager"] An error occurred when refreshing edge VM placement config ########-23d7-423b-a5be-############. Exception null Error java.lang.NullPointerException
        at com.vmware.nsx.management.inventory.service.CmInventoryService.buildErrorInfoFromCmStatus(CmInventoryService.java:1276)
...

• In the NSX-T manager log /var/log/syslog.log we see the following ERROR:

nsx-mgr-01 NSX 24212 SYSTEM [nsx@6876 audit="true" comp="nsx-manager" level="INFO" subcomp="manager"] UserName:'admin' ModuleName:'common-services' Operation:'GET@/api/v1/fabric/compute-managers/########-6845-4514-ab42-############' Operation status: 'failure' Error: The requested object : ComputeManagerModel/########-4514-ab42-############ could not be found. Object identifiers are case sensitive.

• The vCenter (Compute Manager) in which this Edge node was deployed on, no longer exists as a Compute Manager in NSX-T or the edge was moved to another Compute Manager.
   
• Using the API: GET https://<nsx-manager-ip>/api/v1/transport-nodes/{edge-node-uuid} to check the deployment config, in the output file the field "vc-id" (vCenter-id) indicated will be different than the actual "vc-id" (vCenter-id) on which edge node vm currently resides

Environment

VMware NSX-T

Cause

• This issue occurs when a compute manager was removed or the edge VM was moved in vCenter to another compute manager, NSX-T still thinks the edge nodes deployed on the removed CM or CM where it was originally deployed on. 
  Since NSX-T Data Center 3.0.0, it is not possible to remove a Compute Manager which has NSX-T VMs deployed on it, such as: NSX-T Manager, SVM's or Edge Nodes.
  From the above log entry, the compute manager '########-6845-4514-ab42-############' was removed while the edge nodes where still deployed on it.
  For further details, please review VMware NSX-T Data Center Installation Guide.
  The following API can be used to confirm the Compute Manager no longer exists:

GET /api/v1/fabric/compute-managers

 

Resolution

Since VMware NSX-T Data Center 3.0, it is not possible to remove the Compute Manager without first moving the deployed components to another Compute manager or deleting them.
The action of attempting to delete the compute manager will fail and present an alert indicating the objects which need to be moved, before you can delete the Compute Manager.

Workaround:
Deploy a new edge node on the correct CM and replace it in the cluster, with the edge node that has the issue
If you believe you have encountered this issue and the workaround does not work, please open a support request with Broadcom Support and refer to this KB article.