Lost connectivity to volume during storage vmotion running on ESXi 6.7

Products

• VMware vSphere ESXi

Issue/Introduction

Symptoms:

On ESXi 6.7 you may see higher number of unmap requests sent to backend storage array during Storage vMotion or after files have been deleted on VM.
This may cause performance degradation on backend storage array and/or ESXi host may get disconnected from vCenter.

Environment

VMware vSphere ESXi 6.5

VMware vSphere ESXi 6.7

Cause

This issue is caused due to incorrect calculation of unmap rate on ESXi under certain condition and if the backend storage array is unable to process the number of unmap commands sent during a given period.
Automatic space reclamation (unmap) was introduced in ESXi 6.5 and the unmap rate to backend array could be controlled with unmap priorities ( low, medium and high). Beginning with ESXi 6.7 we started supporting “configurable unmap rate” where unmap rate can be configured to reclaim space much quicker or limit the rate depending on performance of backend storage array.
Apart from the configurable rate in ESXi 6.7 , the unmap priority low from UI and medium, high priority from esxcli is supported.

Due to a incorrect calculation of unmap rate in ESXi 6.7 we can see a high unmap rate of 300MB/Sec and higher and up to 1000 MB/sec while unmap priority is set to “low”. This can cause large number of unmap commands being sent to backend storage array.
 

Resolution

This issue is resolved in ESXi 6.7 Patch ESXi670-201912001. To download go to the Customer Connect Patch Downloads page    
For more details refer to -  Configuring Space reclamation for  VMFS 6 Datastore  and  ESXCLI Command to Change Space Reclamation Parameters


Workaround:
To workaround this issue disable automatic unmap feature.
To reclaim the space run manual unmap - Refer to Manually Reclaim Accumulated Storage Space 
Please check with backend storage array vendor before disabling automatic unmap.