I have 2 HyperV server (2012 R2) nodes with shared ISCSI storage (ubuntu 16.04) in a Failover Cluster.

I want to make that 3 servers (windows server 2016 and/or 2012r2) each with 2 partitions, one for the system and one as storage.

How would I make this work if I have to do maintenance on one of the servers ?

With what I have right now I trasnfer the vms to the working node and do maintenance on the second one, the storage being external it doesn't change anything to the vms. With how I think I want to do it, closing one server would make it's storage unavailable and everything on it would crash with it.

Right now the storage servers have all their disk in a hardware RAID 1+0, so if a disk fails it won't cause problems to the server, would it be possible to do the same thing for the cluster in case one of the servers has to be closed ?

I've read that using hardware and software raids are a no go but I don't see how I could make this work without both ?

If I do a storage pool in the failover cluster, will that make it possible to close one server and it'll update its storage when I bring it back up without making the vms unavailable ?

Would I need to transfer all the data out of the storage to one that will stay up and then do maintenance on the server and re-transfer it all back afterwards ? Or I stay with external storage and hope to never need to close it since it's pretty much the same problem ?

Some considerations : Old servers HP proliant Gen 7, Dell Poweredge Gen 12 and maybe IBM Server X5. Also, can't get other raid controllers, I get whatever comes in these servers so no pass-through, JBOD, etc

  • 1
    Are you considering shared storage (an array) or distributed? You can try a distributed, converged architecture and run your storage on your compute nodes with something like Storage Spaces Direct. – John Mahowald Dec 22 '17 at 3:12
  • 1
    S2D requires Datacenter edition all-around, and an upgrade from 2012R2 -> 2016. Bad news they are ripped off from the current production version of Windows Server because Microsoft S2D update was completely non-functional :( I'd pass by OP's place. – BaronSamedi1958 Dec 22 '17 at 11:00

As comments bellow say, Windows Server 2016 Datacenter Edition has built-in Storage Spaces Direct (S2D) that allows to implement/setup hyperconverged cluster of 2 or more nodes. However, it has been having too many issues so it was excluded in the latest Windows Server built. Source.

Despite that, there are still options to set up shared storage for Hyper-V Failover Cluster. StarWind Virtual SAN and HPE StorageVirtual VSA do the job for years before S2D was released. The solutions allow mirroring underlying storage between nodes for the purpose of clustering. So you can use the solutions to implement 2- or 3-node Hyper-V cluster. The difference between those is that StarWind is free and native for Windows and has a Linux-based version, while HPE VSA is free but only a Linux-based storage appliance.

With that being said, I would suggest you using Ubuntu iSCSI storage as a backup repository, and implement StarWind for Hyper-V.

| improve this answer | |

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.