All hosts are running at a reduced CPU clock speed of 28.6%

Products

• VMware vSphere ESXi

Issue/Introduction

Multiple hosts within the cluster are reporting a consistent, throttled CPU clock speed of exactly 28.6% of their rated frequency. This performance degradation is uniform across the fleet and persists regardless of OS-level power management settings.

• Uniform Throttling: Every host in the environment displays the same 28.6% cap.
• High Latency: Significant increase in application response times and processing duration.

Cause

The specific value of 28.6% is a signature of a Power Supply Unit (PSU) Redundancy Failure or a Power Capacity Mismatch at the chassis/blade level.

Most enterprise servers (e.g., Dell PowerEdge, HPE ProLiant) are designed to "fail safe." If the server detects that the available power from the remaining PSUs cannot support a 100% load during a spike, the Integrated Dell Remote Access Controller (iDRAC), Integrated Lights Out (iLO), or Chassis Management Controller (CMC) will trigger an Extreme Power Cap.

Potential Cause	Description
PSU Loss	One of the redundant power supplies has failed or lost its AC source.
Power Budget Over-subscription	New hardware (GPUs, high-draw NVMe drives) was added without upgrading the PSU capacity.
Firmware Bug	A glitch in the Power Management Controller incorrectly reporting available wattage.
DAPC/OS Conflict	The BIOS is set to "Minimum Power" or a specific "Power Cap" policy has been pushed via management software.

Resolution

1. Validate Power Hardware

1. Check the physical and logical status of the PSUs:
2. Log into the Out-of-Band management interface (iDRAC/iLO/IPMI).
3. Navigate to Power -> Supply and check for "Failed," "Input Lost," or "Mismatch" status.
4. Ensure the PSUs are plugged into separate power circuits (A/B feeds).

2. Review Power Cap Policies

1. Check if a Static Power Cap has been accidentally enabled at the Cluster or Chassis level.
2. In the management console, ensure the Power Profile is set to "Performance" or "OS Control" rather than "Extended Power Saving."
3. Clear "Throttling Flag" via Power Cycle
4. If a PSU was replaced but the speed remains at 28.6%, the firmware may be "stuck" in a protective state.

3. Gracefully shut down the host.

1. Remove all power cables for 30 seconds to drain the capacitors (Cold Boot).
2. Reconnect and power on.

4. Update Firmware

1. If hardware is healthy, the issue may be a known bug where the BMC (Baseboard Management Controller) incorrectly calculates the power budget. Update the following:
2. BIOS/UEFI
3. BMC/iDRAC/iLO Firmware
4. Power Management Controller Firmware

5. Update Firmware

     Change the power settings in power management of the iDrac/iLO to use A/B Grid Redundant. (Should always be set to Redundant)

 

 

Additional Information

https://www.dell.com/support/kbdoc/en-in/000202926/poweredge-power-settings