Error: INVALID_FAULT_DOMAIN_NAMES while expanding a vsan stretch cluster

Products

• VMware Cloud Foundation

Issue/Introduction

Change the Fault domain names when we encounter the above error while cluster expansion.

Symptoms:

While performing the expansion operation on a vsan stretched cluster we get the following error :

"errorCode":
The minor error code
"INVALID_FAULT_DOMAIN_NAMES"

Cause

The error is encountered when the fault domain names that were set by the SDDC while stretch cluster creation has been changed manually outside the VCF environment.

Resolution

Rename the fault domain names in the following format on the vcenter itself:
<<cluster_name>>_primary-az-faultdomain and <<cluster_name>>_secondary-az-faultdomain

After the renaming is done, trigger a new cluster expansion workflow with these updated azName fields in the clusterExpansionSpec.

Ex: If the cluster name is "Test_Cluster" , the primary and secondary fault domains should be named as: 
Test_Cluster_primary-az-faultdomain and Test_Cluster_secondary-az-faultdomain

Note: It is highly recommended not to change anything outside vcf environment which is set up by SDDC.

 

Additional Information

Impact/Risks:

When we perform the vsan stretch cluster expansion, the sddc looks for the  below format: 
 

• For non VxRail environment: <<cluster_name>>_primary-az-faultdomain and <<cluster_name>>_secondary-az-faultdomain 

• For VCF on VxRail environment: <<cluster_name>>_az1-faultdomain and <<cluster_name>>_az2-faultdomain

for primary and secondary fault domain names and when this format is not found the cluster expansion task fails.



Note: The sddc does not store this entry anywhere in the SDDC database it just looks for the format when we perform cluster expansion for vsan stretched cluster. We might not get error while unstretching or replacing a failed host in vsan stretch cluster as the clusterUnstretchSpec and clusterCompactionSpec doesn't contain the entry to specify azName.