Patch available for critical vSAN issue for all-flash clusters with deduplication enabled

Products

• VMware vSAN

Issue/Introduction

A critical patch for vSAN is available through VMware vSphere 6.0 EP11 and vSphere 6.5 EP4.

This patch release contains a fix for a rare vSAN bug that is exposed only in all-flash clusters with deduplication enabled, and with very specific I/O patterns. The following vSphere versions might be exposed to this issue: 6.0 p04, 6.0 U3, 6.0 P05, 6.5, and vSAN versions 6.6 and 6.6.1.

Symptoms
If you have a vSAN all-flash cluster with deduplication enabled, you might see the following two vSphere Observations (VOBs) in the Events tab:

vSAN detected an unrecoverable medium or checksum error for component uuid on disk group uuid.

vSAN detected and fixed a medium or checksum error for component uuid on disk group uuid.

In addition, you might notice I/O read errors. In some cases, you might encounter severe symptoms, such as inaccessible VMs, host failures, or stuck resyncs. This rare issue manifests itself only under highly specific operations and I/O patterns.

Environment

VMware vSAN 6.5.x
VMware vSAN 6.2.x
VMware vSAN 6.1.x
VMware vSAN 6.6.x
VMware vSAN 6.0.x

Resolution

If you have seen the messages listed above, please call VMware Global Support Services and opening a support case.

If you have a vSAN all-flash cluster with deduplication enabled with assistance of support perform the following steps:

1. Procure swing storage
2. Migrate all data off of the affected vSAN cluster 
3. Upgrade to the correct patch release as listed below
4. Destroy/recreate all disk groups
5. Migrate all data back to the newly created vSAN datastore

Use the following KBs to upgrade your cluster to the correct patch release.

• For ESXi hosts running vSphere 6.0.0 P04 or above
• For ESXi hosts running vSphere 6.5.0

Additional Information