"vSphere HA virtual machine monitoring action" alarm is triggered due to Storage I/O Timeouts.

Products

• VMware vSphere ESXi
• VMware vCenter Server

Issue/Introduction

The following symptoms are observed in the vSphere environment:

• vSphere HA Alarm: "vSphere HA virtual machine monitoring action" is triggered, resulting in an automatic reset of the affected virtual machine.

• The result of investigation inside the Guest OS indicates significant delays or timeouts during file I/O operations prior to the reset.

• The vmkernel.log on the ESXi host shows HBA driver abort operations.

Log Example :

YYYY:MM:DDTHH:MM:SS cpu48:2097564)qlnativefc: vmhba2(12:0.0): qlnativefcEhAbort: abortCommand mbx success.
YYYY:MM:DDTHH:MM:SS cpu25:2105403)qlnativefc: vmhba2(12:0.0): C0:T4:L1 - FCP command status: 0x5-0x0 (0x8) portid=012400 oxid=0x715 cdb=a30a00 len=3064 rspInfo=0x0 resid=0x0 fwResid=0x0 host status = 0x8 device status = 0x0
YYYY:MM:DDTHH:MM:SS cpu48:2097564)qlnativefc: vmhba2(12:0.0): C0:T4:L1: Abort command succeeded -

Cause

When the HBA driver issues an "abort" command, it indicates that an I/O request did not complete within the expected timeframe at the storage layer.

This may lead to:

• Guest OS Hang: The Guest OS becomes unresponsive because it cannot complete critical disk writes or reads.
• vSphere HA VM Monitoring alarm: vSphere HA detects the lack of heartbeat (and potentially a lack of I/O activity) from the VMware Tools and triggers a VM reset to recover the service.

Resolution

The root cause resides at the storage connectivity.
Contact your storage hardware vendor to investigate the cause of the aborts.

Additional Information

• During application Maintenace some service restart inside guest OS is delayed