Compute Manager cannot be edited in NSX UI due to stale Transport Node Collection entry in Corfu

Products

• VMware NSX

Issue/Introduction

• Compute Manager settings cannot be modified in the NSX UI
• Another symptom, compute manager reports status as down in NSX-UI, after replacing vCenter CA signed certificate .
  • Any attempt to update the new thumbprint fails with error 400
    
    
  • New Thumbprint at vCenter and NSX are reported correctly via openssl s_client output.
    
    Get thumbprint :
    From NSX-mgr-->  openssl s_client -connect <vcenter-ip/fqdn>:443 </dev/null 2>/dev/null | openssl x509 -fingerprint -sha256 -noout -in /dev/stdin
    From vCenter --> openssl s_client -connect localhost:443 </dev/null 2>/dev/null | openssl x509 -fingerprint -sha256 -noout -in /dev/stdin 
    
    
  • Ping and port connectivity between vCenter and NSX manager :
    
    From NSX-mgr cli:
    ping vcenter-ip/fqdn 
    nc -zv vcenter-ip/fqdn 80
    nc -zv vcenter-ip/fqdn 443 
    
    
  • Found thumbprint mismatch error in both nsxapi.log and cm-inventory.
• Logging in /var/log/cm-inventory/cm-inventory.log shows a NullPointerException at VlcmValidator.isVlcmEnabledOnComputeCollection:

ERROR http-nio-127.0.0.1-7443-exec-2 JsonRpcErrorResolver 4412 - [nsx@6876 comp="nsx-manager" errorCode="MP100" level="ERROR" subcomp="cm-inventory"] Error discovered by JSON RPC
java.lang.NullPointerException: null
        at com.vmware.nsx.management.lcm.invcallbacks.validator.VlcmValidator.isVlcmEnabledOnComputeCollection(VlcmValidator.java:113) ~[liblcm-inventory-callbacks.jar:?]
        at com.vmware.nsx.management.lcm.invcallbacks.validator.VlcmValidator.isVlcmEnabledOnComputeCollection(VlcmValidator.java:109) ~[liblcm-inventory-callbacks.jar:?]

Environment

VMware NSX-T Data Center

VMware NSX

Cause

The NullPointerException is caused due to a stale entry in VMware NSX-T Corfu database, for a vLCM Cluster which no longer exists.
While editing the Compute Manager details, NSX-T internally checks if any vLCM cluster is prepared.
An exception is thrown even when there are no vLCM Clusters prepared in NSX when stale Corfu entries are present.
Any attempt to Edit ComputeManager with a new thumbprint fails because of java.lang.NullPointerException" in method "VlcmValidator.isVlcmEnabledOnComputeCollection"

Resolution

This issue is resolved in VMware NSX-T Data Center 3.2.3 and VMware NSX 4.1.1, available at Broadcom downloads.

Workaround

• If you believe you have encountered this issue, please open a support request with Broadcom Support and refer to this KB article.
•  For more information, see Creating and managing Broadcom support cases.