I have the following hardware and configuration:

  • Dell SCv2020, with loads of storage, with 2 volumes configured
  • 2 Dell R430's with VMWare ESXi 6.5 installed
  • Each R430 is connected to the SCv2020 with dual-path SAS cables
  • For testing purposes, per ESXi host, a Windows 10 VM which is connected to a dedicated vhd, and a shared vhd via a separate SCSI adapter, which has bus sharing set to Physical
  • Each Win10 VM has VMWare Tools, and so also all proper drivers installed, and disk cache disabled

Now I have the following issue:

The "shared"-disk is mounted as E: on both VMs, when I write something to the disk via VM1, it's not visible on VM2 on the same disk. After reboot the files are visible on both VMs

What do I have to do to configure this properly so files are visible on both VMs ?

  • You need a shared drive for what in those Win10 VM ? Please explain your need, as I think you did a bad approach to fix want you wanted to do – yagmoth555 Feb 27 '17 at 13:41
  • You can't do block device sharing w/out clustered file system (GPFS, VMFSv3/v5) or local file system + access arbiter (NTFS/ReFS + CSVFSv1/2/3). Good write up here -> forums.starwindsoftware.com/viewtopic.php?f=5&t=1392 – BaronSamedi1958 Feb 27 '17 at 16:02

You cannot just do what you're doing and expect it to work. NTFS is only meant to be mounted one time, by a single operating system.

You will either need to switch to a proper cluster-aware file system or more likely use CIFS/NFS/etc to share out the storage from one central location to all of your clients.

  • 1
    I actually expected that. I'm just looking for a way to share storage between VMs, without a single-point of failure (not counting the Storage Controller). I will look on further. Thanks. – Sander Feb 27 '17 at 13:48

Check shared VHDX - VHDS - which is available in Windows Server 2016. Source 1 and Source 2

Since you want to eliminate a single point of failure, you should consider a file or a block-level data replication.

As for file level check DFS or FreeNAS. As for block-level, Microsoft S2D, StarWind Virtual SAN, HPE VSA will do the job of highly-available storage.

  • 1
    Be careful with DFS as it can't do concurrent file access properly - it has no locks. You can use guys like PeerLock but they are costy. – BaronSamedi1958 Feb 27 '17 at 16:03

You can actually do this but only using Windows server and even then it's bad bad idea. What you've actually done is massively-corrupt your volume, consider it now entirely dead and unsavag

What you'd do is create a new VMDK, attach it to the first VM with the VMN shutdown - ensure it's set to 'multiwriter' mode then attach this same exact VMDK to the second VM, also shutdown and again ensure 'multiwriter' is set. When they come up you need to have MS Cluster Services/MSDTC configured and define the new disk as a shared/cluster-managed - then only one node has read/write on the disk but the second node can read the volume as MSDTC is doing the locking.

It's still a bad idea, do what everyone else says and use a NAS.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.