5

Environnent is full Windows Server 2012 R2.

I have a Hyper-V cluster on which I host a testing fail-over cluster file server with 2 guests. Guest cluster is running and file server role installed and running fine with many shares on it. For now a single drive is hosting the data.

Since it is impossible to resize a shared vhdx I wonder if I choose the right solution for my needs. File servers has to be VMs.

What would be the solution to extend the data drive when it will get full ? Any solution without multiplying the number of volumes ?

Thank you.

EDIT : I would like, as I'm used to be on open source environments, to end up with a full failover file server on which I could extend the data volume without down time and to avoid using twice the necessary storage as for a DFS cluster without shared volume.

Cyril
  • 51
  • 2
  • 1
    How about using two file servers and DFS instead? – joeqwerty May 20 '16 at 16:32
  • Hello joeqwerty, thank for your answer. Using 2 DFS servers would do the trick, but I will end up with double space consuming with one, not shared, drive per server. – Cyril May 23 '16 at 06:30

3 Answers3

3

In order to create a true “shared-nothing” failover file server you can use a free Starwind solution https://www.starwindsoftware.com/starwind-virtual-san-free.

Basically it can do the same that Windows Server 2016 will do later but unfortunately only in the Datacenter edition. Starwind can do it already even with a free Hyper-V 2012 server https://slog.starwindsoftware.com/smb3-0-fileserver-on-free-microsoft-hyper-v-server-2012r2-clustered/.

Net Runner
  • 5,626
  • 11
  • 29
2

Apparently you can take the shared VHDX offline (the shares will be unavailable), either by shutting down the clustered guests or by unmounting it, and then change the size of the VHDX and then bring it back online (or boot the guests).

A few web searches have turned up people writing and agreeing that that's the only way to do it. There's one such conversation and a powershell script here that purports to automate the process with minimum downtime.

Todd Wilcox
  • 2,831
  • 2
  • 19
  • 31
  • Hello Todd, thanks for your answer. Yes I could, but this would break all the ideas of a high availability clustered file server. – Cyril May 23 '16 at 06:10
  • @Cyril You didn't specify zero downtime in your question. Scheduled downtime is normal even with high availability, so if you are looking for a zero downtime way to do this, you might add that to your answer. – Todd Wilcox May 23 '16 at 12:06
1

Because this is only a testing environment have you considered creating several shared VHDX disks and using them to create a Clustered Storage Space? that way you can add more disks to the storage space to increase capacity as needed.

Michael Brown
  • 3,204
  • 2
  • 9
  • 10
  • Hello Michael, thanks for your answer. I tried this, but the server will then end up with as many disks as I add space, no ? I haven't find out how to "merge" the CSV to appear as one on the guest. – Cyril May 23 '16 at 06:23
  • 2
    Hi, you would see more disks but from them you can create a single volume that crucially can appear larger than the disk space that you have available and that can be extended in the future. – Michael Brown May 23 '16 at 06:59
  • Not sure to understand. I would extend the drive on my host, create a new disk, add it and share it among the guest. Then, from the guest OS I can add this new disk, but it will appear as a separate disk instead of only one. Using the file server role inside failover cluster manager I can add a new storage to the role, but again, this will add a disk (and a new default administrative share) instead of extending the existing one. – Cyril May 23 '16 at 08:03
  • 1
    @Cyril Windows has the capability of having one volume span multiple disks. If I'm reading correctly, that's what Michael is suggesting. Add disks and then extend the volume to include the added disk(s). – Todd Wilcox May 23 '16 at 12:07