Impact and Considerations for Re-registering ESXi Hosts in an HCX Environment

Products

• VMware HCX
• VMware vCenter Server
• VMware vSphere ESXi

Issue/Introduction

In an environment where VMware HCX is deployed, re-registering an ESXi host (disconnecting, removing, and adding it back to vCenter) can have side effects if HCX appliances or HCX-related resources are associated with that host. This article describes the potential impact and considerations on HCX components.

Environment

VMware HCX
VMware vCenter Server
VMware vSphere ESXi

Cause

From an HCX perspective, no impact is expected provided that HCX-related appliances are not running on the target ESXi host during the operation. However, the following factors must be considered:

• Running Appliances: If HCX appliances are running on the host, they must be moved. Moving the HCX-NE via vMotion carries a risk of affecting L2 Extension communication.
• Local Storage: If the local storage of the target ESXi host is explicitly defined in the HCX Compute Profile, deleting and re-registering the host may cause HCX to lose recognition of that storage resource.

Resolution

To safely re-register the ESXi host without disrupting HCX operations, please follow the steps below:

1. Evacuate Virtual Machines:
   • If HCX-related appliances are running on the target ESXi host, place the host into Maintenance Mode. This will evacuate all running virtual machines to other available ESXi hosts.
   • Note: Be aware that L2 Extension communication may be momentarily affected after the vMotion of the HCX-NE : Communication failure after HCX Network Extension appliance vMotion.
     
     
2. Re-register the Host:
   • Once the host is empty, proceed to disconnect, remove, and re-register the ESXi host using its FQDN.
     
     
3. Verify HCX Compute Profile:
   • If a local datastore on the target ESXi host is selected as a Deployment Datastore in the HCX Compute Profile, removing the host will also remove that datastore object from vCenter. When the host is re-registered, the datastore is created as a new object and HCX may still reference the old, now-invalid datastore in the Compute Profile / Service Mesh, causing deployment or resync operations to fail : HCX Service mesh creation failed with error: no host is reachable to any selected datastores.
     
     
4. Operational Precautions:
   • Refrain from performing HCX migrations or other HCX configuration changes while HCX appliances are being migrated between hosts.