vMotion fails with CPU compatibility errors on identical physical processors (Sapphire Rapids)
search cancel

vMotion fails with CPU compatibility errors on identical physical processors (Sapphire Rapids)

book

Article ID: 448581

calendar_today

Updated On:

Products

VMware vCenter Server

Issue/Introduction

When attempting a live vMotion migration of a virtual machine between two ESXi hosts, the migration fails during the compatibility validation wizard. The target host reports that it does not support the virtual machine's current hardware requirements.

This issue occurs even though both the source and destination hosts utilize identical physical processor models (e.g., Intel Xeon Gold 6438Y+ / Sapphire Rapids) and have Enhanced vMotion Compatibility (EVC) disabled at the cluster level.

Error Messages: The validation check displays one or more of the following unsupported CPU feature warnings:

  • The target host does not support the virtual machine's current hardware requirements.

  • Move 64 bytes as Direct Store (MOVDIR64B) is unsupported.

  • Move Doubleword as Direct Store (MOVDIRI) is unsupported.

  • Cache Line Demote is unsupported.

  • Advanced Vector Extensions 512 support for BF16 (AVX512BF16) is unsupported.

Environment

  • VMware vCenter Server

  • VMware vSphere ESXi 7.0.x / 8.0.x

  • Hardware: Identical CPU hardware generations on both source and target hosts (e.g., Intel Sapphire Rapids Generation)

Cause

A host-level hardware configuration discrepancy or microcode patch misalignment is causing the destination hypervisor to mask native CPU instruction sets (such as MOVDIR64B, MOVDIRI, or AVX512BF16).

Even though the physical processors match, ESXi cannot expose disabled or unpatched CPU features to the guest OS. If the virtual machine was powered on on a source host where these instruction sets were unmasked, it locked those features into its CPUID mask at boot; attempting to migrate it live to a host masking these features will trigger a compatibility failure to prevent a guest OS crash.

Resolution

Option 1: Align Host BIOS and Firmware Configuration (Recommended for Live Migration)
  1. Place the destination ESXi host into Maintenance Mode and reboot the server into its BIOS / System Setup utility.

  2. Audit the CPU configuration settings. Ensure that high-performance instruction sets (such as AVX-512, Direct Store, and virtualization helper features) are configured identically to the source host.

  3. Compare the server BIOS and out-of-band management controller (e.g., Dell iDRAC, HPE iLO) firmware versions. Update the destination host to match the source host's firmware levels.

  4. Verify the CPU microcode version applied on both hosts. If security mitigations (e.g., GDS/Downfall patches) have restricted CPU instructions on the target host, apply matching patches to the source host or vice versa.

  5. Exit the BIOS, boot ESXi, exit Maintenance Mode, and retry the live vMotion.

Option 2: Cold Migrate via Guest OS Shutdown (Workaround)

If a host BIOS reboot cannot be scheduled immediately, a cold migration must be performed:

  1. Gracefully shut down the guest operating system of the affected virtual machine.

  2. Migrate the powered-off virtual machine to the destination ESXi host.

  3. Power the virtual machine back on.

  4. Upon boot, the guest operating system will renegotiate and initialize using the restricted CPUID instruction set currently exposed by the destination host. Future migrations between these hosts will succeed while the VM remains in this matched feature state.

Additional Information

For more information on configuring EVC, please refer to the VMware Cloud Foundation Tech Docs Portal.

ERROR: The target host does not support the virtual machine’s current hardware requirements.