A Hyper-V Guest that has direct connected iSCSI disks cannot be replicated?

[mfeightner34]

According to Microsoft, Hyper-V Replication can be done before any physical disks are connected to a Hyper-V guest Virtual Machine. After replication is completed, add the physical disks back.  However, this would have to be done anytime you wish to re-replicate the VM again, as any physical (iSCSI) disk will be detected upon the next replica cycle. 

Does anyone know if this has changed at all?  Is there any way to regularly replicate VMs that have iSCSI attached disks?

Thanks Nimble Community!

[jshuot16 · Accepted solution]

For now there is no other way that I know. I'm in the process of moving all my clients VM with iSCSI target to VHDX drive. It's not always desired configuration but this is the way to go if you want to be more dynamic and ready for the Cloud. If you find anything better let me know, I might be interested.

I don't know exactly what you're trying to do but that article might help you.

Shared VHDX Files - My Favorite New Feature in Windows Server 2012 R2 - Windows Storage Server - Site Home - TechNet Blo…

[mfeightner34]

Hi Jean-Sébastien,

Thanks for sharing your thoughts.  I'm wondering why, specifically, you are moving from guest-connected ISCSI to VHDX -- could you share some of the details that resulted in this decision? Does this also include more critical servers such as Exchange and SQL? 

Thanks,

Matt Feightner

[mfeightner34]

BTW, the article on shared VHDX files -- very interesting....  Thanks!

[jshuot16]

Hi Matt,

I was previously doing some site to site replication using block based replication. In the beginning it was for everything slowly it became only for the SQL and Exchange and now we're looking toward the cloud. If we want to migrate to something like Azure Site Recovery we need to have everything inside the VM. In the end we're looking to simplify the DRP actions, less complexity = less errors.

[foodandbikes]

Old thread, but there is a way.

You edit the VM Integration Services in Hyper-v and uncheck the "Backup" box.

I've got a few VMs replicating like this with no ill effects.

[mfeightner34]

Hi Dan.  Just to be clear, your solution would allow replication of a VM even though some or even all disks are not "local" to the VM.   i.e. iSCSI connected?

I will have to try it out.  Thanks for the post!

[foodandbikes]

It will allow Hyper-v to replicate the VHD/VHDX files. The iSCSI connected disks will NOT be replicated. To replicate those disks you would need to use Nimble-to-Nimble replication.

So, Hyper-v is replicating the C: drive while Nimble is replicating the D: drive.

[kevin_hsieh]

When I moved to Nimble, I converted in guest iSCSI attached volumes to VHDX. This includes Exchange, SQL, and file servers. In general, I don't see any advantages that a raw iSCSI device connected to a guest initiator is beneficial over using VHDX. A VHDX attached via virtual SCSI adapter can be grown online, detached and connected to another VM, and cloned (by Nimble). The only case where I can say for sure than in guest iSCSI is better is when doing failover clustering between two guests using shared storage, and you would like to be able to dynamically grow a volume. Hyper-V 2012 R2 supports shared VHDX for guest level clustering, but that VHDX cannot be grown online. Microsoft's solution it to add additional VHDX, which can be done online, or upgrade to Hyper-V 2016. Microsoft removed the limitation on shared VHDX in Hyper-V 2016, in which case it's really hard to make an argument that in guest iSCSI is better than VHDX in any situation.

[klitman51]

I know this post is a little old but we have been using Hyper-V Replication for some time now on Windows Server 2012 R2 with VM's that have direct iSCSI volumes. Our environment has 2 Failover Clusters, one in 2 different sites. Along with that, each site has a Nimble array with all volumes on each replicated to the other. In the event I need to bring up a Replica VM, I would first bring up the VM (obviously without the iSCSI disk) and then either handover or promote the Nimble snapshot and then connect that volume to the replica VM and reboot. If you're able to handover the volume collection, you obviously won't have data loss but in the event that's not an option, the volume collection can be promoted which will cause at least some data loss depending on the frequency of your snapshots.

[kareemtharwat]

Its work for me and solved the problem thanx