Resync in a hung state after a RAID5 storage policy change on vSAN ESA clusters of 5 hosts or less.

Products

• VMware vSAN

Issue/Introduction

This occurs under the following specific conditions:

•  vSAN ESA Clusters of 5 nodes or less
•  vSAN ESA Stretch Clusters with 5 or less nodes on each side of the stretch cluster
•  Large size VMDKs (1TB or more) exist in the cluster
•  vSAN Administrator changes Storage Policy for one or more large size VMDKs
  Standard Clusters
  
  • non-RAID5 to RAID5
  • Changing any other policy parameter (not affecting the existing RAID-5 configuration)
  
  Stretched Clusters         
  • non-RAID5 to RAID5
  • Changing any other policy parameter (not affecting the existing RAID-5 configuration)
  • RAID5 Keep data on Preferred/Secondary Site to RAID5 Mirroring

 

Possible Symptoms:

• Resync latency affects all workloads in the cluster, and VMs become unresponsive (stunned).
• vSAN datastores can run out of logical/physical space
• Resync gets stuck in a loop and doesn't progress
• Hosts fail to enter maintenance mode with ensure accessibility/full evacuation

Environment

VMware vSAN 8.x ESA

Resolution

Engineering is aware of this issue, and a fix is due in an upcoming release
If the environment is currently in a hung resync, open a case with vSAN Support for further assistance

Workaround:
If possible, please wait for the upcoming release before changing the policy on the VMs with large VMDKs if they meet the above conditions. Otherwise, changing the policy on one VM with large VMDKs at a time may help if there is more than 50% free space.