Unexpected ESXi host reboot with Processor Transition to Non-recoverable event in IPMI SEL and no PSOD

Products

• VMware vSphere ESXi

Issue/Introduction

An ESXi host reboots unexpectedly. The following symptoms are observed:

• No Purple Diagnostic Screen (PSOD) is observed on the console.

• No core dump (vmkernel-zdump) is generated.

• In the ESXi logs, such as /var/run/log/vmkernel.log, the logging stops abruptly without indicating any specific error or panic message.

• After the reboot, checking the IPMI System Event Log (SEL) via commands like localcli hardware ipmi sel list reveals a hardware event recorded around the time the ESXi logs stopped.

Example of the log entry:

Record:X:
 Record Id: X
 When: YYYY-MM-DDTHH:MM:SS
   Event Type: 7 (Fatal/NonRecoverable)
   SEL Type: 2 (System Event)
   Message: Assert + Processor Transition to Non-recoverable
   Sensor Number: 13

Additionally, Machine Check errors or Fatal errors related to the CPU may be visible in the BMC (Baseboard Management Controller) interface, such as iDRAC or iLO.

Environment

VMware vSphere ESXi

Cause

This issue occurs because a fatal hardware failure (such as a processor error) causes the system to halt or reset at the hardware level immediately. This happens before the ESXi software can handle the exception, generate a PSOD, or write any error details to the logs.

Resolution

This is a hardware-related issue. Please perform the following steps:

1. Check the IPMI SEL or the hardware management console logs (e.g., iDRAC, iLO, IMM) to identify fatal errors related to components such as the CPU, memory, or motherboard.

2. Collect details of the recorded hardware errors, including screenshots or hardware log exports.

3. Contact your server hardware support to request an investigation and resolution for the hardware failure.

Additional Information

Japanese version: IPMI SEL に Processor Transition to Non-recoverable イベントが記録され、PSOD を伴わずに ESXi ホストが予期せず再起動する(421431)

When ESXi logs stop abruptly and only boot logs are visible in vmkernel.log after a reboot, it often indicates a hardware-initiated power cycle or reset.