The hostd service might intermittently fail

Products

• VMware vSphere ESXi 8.0

Issue/Introduction

Frequent hostd crashes due to memory corruption in IpmiIfcSensorGetThresholds when an invalid SDR header record is detected
The hostd service might intermittently fail due to an invalid sensor data record

Environment

This issue affects ESXi 8.0 Update 2 and all earlier versions.

Cause

The cimsvc thread in hostd periodically polls IPMI sensor data from the BMC. One of the sensor data records returned by the hardware server's BMC was invalid. Instead of rejecting this invalid record, the ESXi code processed it, causing the hostd process to crash.

Resolution

In rare cases, the Baseboard Management Controller (BMC) might provide an invalid sensor data record to the hostd service and cause it to fail. As a result, ESXi hosts become unresponsive or disconnect from vCenter.

 

This issue is resolved in ESXi 8.0 Update 2b. The fix implements checks to validate the sensor data record before passing it to the hostd service.

 

Technical documentation for reference: release notes

https://techdocs.broadcom.com/us/en/vmware-cis/vsphere/vsphere/8-0/release-notes/esxi-update-and-patch-release-notes/vsphere-esxi-80u2b-release-notes.html