Virtual Machines on vSAN using Independent Non-Persistent Delta and Related Redo vmdks May Become Inaccessible After Shutdown

Products

• VMware vSAN

Issue/Introduction

The Virtual Machine is configured with an independent non-persistent delta disk and associated redo VMDKs (typically in VDI environments such as Citrix or Horizon).
When the VM is powered off, it appears as Inaccessible in the VC UI and shows as Invalid in Host UI.



From VC UI - vSAN Cluster - Monitor - Virtual objects for the affected VMs appear as " Inaccessible"



Example configuration 
Disks listed in vmx:

scsi0:1.deviceType = "scsi-hardDisk"
scsi0:1.fileName = "disk_IdentityDisk.vmdk"
sched.scsi0:1.shares = "normal"
sched.scsi0:1.throughputCap = "off"
scsi0:1.present = "TRUE"
scsi0:2.deviceType = "scsi-hardDisk"
scsi0:2.fileName = "disk_TemporaryStorage.vmdk"
sched.scsi0:2.shares = "normal"
sched.scsi0:2.throughputCap = "off"
scsi0:2.present = "TRUE"
scsi0:0.deviceType = "scsi-hardDisk"
scsi0:0.fileName = "disk-########-####-####-####-###########-xd-delta.vmdk"
scsi0:0.mode = "independent-nonpersistent"

From the VM’s vmware.log, the events observed during power-off are normal and expected.

2021-11-30T23:17:49.415Z In(05) vmx - DISK: Discarding redo log for nonpersistent disk scsi0:0: ./disk-########-####-####-####-########d7f3-xd-delta.vmdk.#####_#####
2021-11-30T23:17:49.418Z In(05) vmx - DISKLIB-VMFS : "vsan://528d96d3########-######625fae60fc/########-####-####-####-########8196" : open successful (1114113) size = 0, hd = 0. Type 20
2021-11-30T23:17:49.418Z In(05) vmx - DISKLIB-VMFS : "vsan://528d96d3########-######625fae60fc/########-####-####-####-########8196" : closed.
2021-11-30T23:17:49.419Z In(05) vmx - OBJLIB-VSANOBJ: VsanObjUnlink: Unlinking '/vmfs/volumes/vsan:528d96d3########-######625fae60fc/########-####-####-####-########6d96' -- 'vsan://528d96d3########-##############/########-####-####-####-########8196'.
2021-11-30T23:17:49.447Z In(05) vmx - DISKLIB-LIB : DiskLibUnlinkInt: Disk delete successfully completed { result:0, Msg: 'The operation completed successfully', fileName:'./disk-########-####-####-####-########d7f3-xd-delta.vmdk.#####_#######'}

From hostd.log you see the following:

2021-11-30T23:18:10.925Z error hostd[2103066] [Originator@6876 sub=Vmsvc.vm:/vmfs/volumes/vsan:528d96d3########-######625fae60fc/########-####-####-####-########9b50/vmname.vmx opID=HB-SpecSync-host-#####@1091845-########-##-348b user=vpxuser] Invalid transition requested (VM_STATE_INVALID_LOAD -> VM_STATE_RECONFIGURING): Invalid state

Environment

VMware vSAN 7.x
VMware vSAN 8.x

Cause

When a VM configured with an independent non-persistent disk is powered on, a shadow REDO disk is created to track changes. Upon powering off, the non-persistent disk and its associated REDO disk are deleted.
Since the REDO disk vSAN object remains registered with the vSAN APD (All Paths Down) monitor, the vSAN APD monitor reports disk as inaccessible, causing the VM to transition into an Inaccessible/Invalid state.

Resolution

The issue is fixed in ESXi 7.0U3q and 8.0U2.

Workaround:
Remove the Inaccessible/Invalid VM from the VC inventory and then register the VM.

Additional Information