Enabling VAAI UNMAP feature in ESXi 5.x on an EMC VMAX Array environment results in VMFS metadata and guest data corruption
search cancel

Enabling VAAI UNMAP feature in ESXi 5.x on an EMC VMAX Array environment results in VMFS metadata and guest data corruption

book

Article ID: 340007

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

Symptoms:
When VAAI UNMAP feature is enabled in ESXi 5.x on an EMC VMAX Array environment, you experience these issues:
  • VMFS metadata is corrupted.
  • Guest data is corrupted.

In both cases the valid data appears from the ESXi host / guest as zeroed (overwritten). The range of zeroed data is an even multiple of 768Kb.


Environment

VMware vSphere ESXi 5.5
VMware vSphere ESXi 5.1
VMware vSphere ESXi 5.0

Cause

When an ESXi host first detects each device, it sends one NULL UNMAP to test that device's support for Block Delete operations. The VMW_VAAIP_SYMM plug-in in ESXi use SCSI UNMAP commands with a ParameterListLength of 0 (A 'NULL UNMAP' in this document) to detect whether a device on an EMC VMAX array supports the Block Delete VAAI Primitive.

When the EMC VMAX receives the NULL SCSI UNMAP command, it may in error UNMAP a series of tracks. These tracks may contain guest data or VMFS metadata.

Resolution

This is not a VMware issue.

EMC has released Enginuity fix 72255 to address this issue. Contact EMC support.
 
 
Notes:
  • You require an EMC account to view the preceding link.
  • The preceding link was correct as of September 3, 2014. If you find the link is be broken, provide a feedback and a VMware employee will update the link.


Additional Information

To be alerted when this document is updated, click the Subscribe to Article link in the Actions box

Attachments

Internal_2073511_EMCKB183439.docx get_app