NSX manager node(s) are down in the cluster after issues with storage/vSAN partitioning

Products

• VMware NSX

Issue/Introduction

• After encountering a storage issue, NSX manager VMs may not be working properly. The manager nodes may show error messages from the VM consoles related to services not starting.
• After rebooting the manager nodes, the error message may indicate a corrupted file system. 
• An example console message is "Failed to start default target: Transaction for nsx-custom.target/start is destructive (emergency.target has 'start' job queued, but 'stop' is included in transaction)."

Environment

• VMware NSX
• VMware NSX-T Datacenter

Cause

Environmental issue caused NSX Manager VM file system corruption.

Resolution

There are several ways to restore NSX manager cluster depending on if there is still at least one manager node remains working.

Scenario 1: At least one NSX manager node is still running, NSX UI is available when connecting to the node's IP/FQDN

• Access the NSX UI and go to System, Appliances. 
• Locate the problem NSX Manager node, then initiate the deletion.
  • If the Delete option is not available, access the cli of one of the existing NSX manager VMs, identify the UUID of the faulty node via the get cluster status command and then remove it from the cluster via the detach node <UUID> command (see Detach the Failed NSX Manager Node from the NSX Manager Cluster).
• Once the problem node is removed successfully, a new node can be deployed from the same page or via OVA deployment (see Redeploy the Failed NSX Manager Node).

Scenario 1a: At least one NSX manager node is still running, however, NSX UI is unavailable on any manager nodes

• If there is only one NSX Manager showing as unavailable
  • Follow this guide and detach nodes showing unavailable
• If there are two NSX Managers or the entire cluster showed as unavailable
  • Follow this guide to deactivate NSX manager cluster
• Check if the following command from the admin shell outputs the status of the surviving node as up or down (instead of unavailable).
  • • nsx-manager> get cluster status
• Shut down the problem nodes.
• Reboot the surviving NSX Manager node.
• Run the above "get cluster status" command on the surviving Manager node and wait for all the services to show as "UP". 
• Once the NSX UI is confirmed working, delete the broken manager node VMs from vCenter. 
• Access the NSX UI and re-deploy NSX managers 2 and 3.
• If NSX UI has errors when deploying the additional nodes, deploy the new manager nodes using OVA file.
  • Join the newly deployed NSX manager to the surviving node/cluster
• Verify the NSX manager cluster is now showing "stable" under System > Appliance.

Scenario 2: All NSX manager nodes are not booting to login shell and showing storage corruption

• Attempt the file system repair
• If file system repair does not make at least one manager node reboot successfully to login shell, deploy a brand new NSX manager using OVA file and restore backup from the SFTP server. 

Scenario 3: All NSX manager nodes are not booting to login shell and no valid backup was made and available for the NSX managers 

• Please note that Broadcom Support do not have a way to restore your NSX environment if no valid backup is available.
• Please raise a Support Request with Broadcom Support to assist attempts of recovering the manager nodes.

Additional Information

Restore an NSX Manager Node to an Existing NSX Manager Cluster