Unable to access the VMware virtual machine file system datastore when the partition is missing or is not set to type fb
search cancel

Unable to access the VMware virtual machine file system datastore when the partition is missing or is not set to type fb

book

Article ID: 340059

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

  • Unable to access the virtual machine file system (VMFS) datastore. The partition is not set to type fb.
  • Unable to access the VMFS datastore. No VMFS partition table on the LUN.
  • Affected volumes are showing up with a partition type of 42 (SFS). Verify this by running fdisk -l from the service console.
  • Error message in the log /var/log/vmkernel :

    Aug 22 15:55:44 esx01 vmkernel: 145:20:59:19.562 cpu1:1037)WARNING: SCSI: 6693: Partition vmhba3:0:12:1 is active: partition table was not updated


Environment

VMware ESX Server 3.5.x
VMware ESX Server 3.0.x

Resolution

The Consolidated Backup proxy server is the only Windows server that can see the VMFS volumes.

When the automatic drive letter assignment function (automount) is enabled within diskpart on the VMware Consolidated Backup (VCB) proxy (default setting), the Windows diskpart initializes the volume on discovery and autoassigns a driveletter to the volume which results in the change of the partition type to 42 (SFS) and ESX losing access to the datastore volume.

To setup VCB properly, see Virtual Machine Backup Guide.

After the VCB proxy configuration is corrected, log in to the ESX console and double-check which volumes are affected.

This example shows how the partition looks after being connected to VCB proxy where diskpart autoassigned the drive letter and changed the partition type to 42 (SFS):
[root@localhost root]# fdisk -lu
To correct this issue, change the partition type back to fb on all the LUNs that are supposed to be VMFS datastores.
For more information, see Recovering a lost partition table on a VMFS Volume (1002281).
Warning: The SFS volumes may be legitimate Dynamic Disks in windows presented as RDMs. Make sure to verify that they are not RDMs presented to Windows OS as Dynamic Disks.

The following step must be performed only after creating backups as the result might lead to complete data loss.
After this change, perform a rescan.

[root@localhost root]# esxcfg-rescan vmhba1

If the volume does not appear as VMFS datastore on the ESX, you may need to align the partition start to the block 128 as this is the default when ESX does create partitions and this is another aspect that Windows does change when initializing the discovered volume.

To set the starting sector to 128 instead of 63 (the default) on the ESX host.

First to confirm that the current Start is 63
This is how the actual change from 63 to 128 is performed.
This is the result after successfully changed.
Rescan at this point and the volume appears on the ESX as VMFS datastore.

[root@localhost root]# esxcfg-rescan vmhba1