Edge Upgrade Failure After Configuring 1GB HugePages on Edge VM

Products

• VMware NSX

Issue/Introduction

 

• An NSX-T Edge upgrade to version 3.2.3.1 fails with the following error:

Edge 3.2.3.1.0.22104592/Edge/nub/VMware-NSX-edge-3.2.3.1.0.22104592.nub switch OS task failed on edge TransportNode <UUID>: clientType EDGE, target edge fabric node id <UUID>, return status switch_os execution failed with msg: An unexpected exception occurred: <......snip.....> ERROR: NSX Edge configuration has failed. 1G hugepage support required

• The Edge VM is running on an ESXi host within a cluster configured with Enhanced vMotion Compatibility (EVC) enabled in "Intel Westmere" mode.
• After attempting to configure the featMask.vm.cpuid.PDPE1GB parameter to enable 1GB hugepages (as referenced in Edge upgrade or new deployment is in a Failed state), powering on the Edge VM results in the following error:

Feature '1 GB pages (PDPE1GB)' was absent, but must be present

 

Environment

VMware NSX-T Data Center 3.2.x

Cause

In previous versions of NSX-T Data Center, both 2MB and 1GB hugepages were supported. However, starting with version 3.2.0, Edge VMs require 1GB hugepages.

On Intel Westmere, Sandy Bridge, and Ivy Bridge generation CPUs, ESXi does not expose 1GB hugepages by default.

Additionally, if the cluster has EVC (Enhanced vMotion Compatibility) enabled and is set to "Intel Westmere Generation," updating the hugepages setting (featmask.vm.cpuid.PDE1GB) will fail because the CPU feature level is too low. To successfully configure 1GB hugepages, the EVC mode must be set to a higher CPU generation, such as "Ice Lake."

 

Resolution

This is a known issue impacting upgrades to NSX-T Data Center 3.2.x.

FixedInVersion#: 3.2.1

Workaround:

1. Create a new vCenter Cluster and configure EVC with a CPU mode higher than Intel Westmere, Sandy Bridge, or Ivy Bridge (e.g., Ice Lake or newer) or disable EVC.

2. Migrate the hosts running the Edge VMs into the newly created Cluster.

3. If the Edge upgrade is already in a failed state:

   • Manually edit the featMask.vm.cpuid.PDPE1GB parameter on one Edge VM at a time and set it to "Val:1" (Edge upgrade or new deployment is in a Failed state).

4. Power on the Edge VMs.

5. Retry the upgrade.

After applying this workaround, the upgrade should proceed successfully without encountering the hugepages requirement error on the Edge nodes.