Can't use Move to migrate VMs to AHV? Manual migration available.

  • 24 March 2020
  • 0 replies
  • 2059 views

Userlevel 3
Badge +4

There are a number of reasons you might want to migrate a VM manually in or out of a Nutanix AHV cluster. You could be working with a situation not supported by Move, such as stand-alone ESXi without vCenter or no network path from the old hypervisor to the Nutanix cluster. Possibly you want to save a VM's disks to portable storage media and physically transport them rather than push them across the WAN.

If you're wanting to move VM data in or out of AHV, the methods have been provided in this article "Transferring Virtual Disks to an AHV Cluster".

If you're just looking to migrate current, working VMs from an ESXi or Hyper-V cluster Nutanix highly recommends using Nutanix Move to migrate VMs to AHV. The manual migration methods require attention to detail and generally involve a larger time investment and more downtime to complete the move.

I won't go into full detail on the steps, that's already done in the article I linked, but I will highlight that for any VM you’ll need VirtIO drivers before moving a VM to AHV. These are your network, disk, and memory drivers and while these are preinstalled in many Linux distros these aren't present by default in Windows up to current versions. Not having these means the VM won't boot with SCSI-attached disks. There's a process to attach the disks as IDE, install the drivers, then clone from the updated disks to make the new SCSI-attached disks but it's much easier to install VirtIO drivers first. Move can automate all of this for you so this is some of the time investment and attention to detail I'm talking about.

The methods are provided in order of increasing difficulty or effort.

The simplest method, described in Scenario one, uses the AHV Image Service to transfer the source virtual disk data and convert it to an AHV image in a single step. This will save you a lot of time. You just have to point to the disk file location by HTTP or NFS so the cluster can fetch the data, provide a name and any annotation you wish to add. You could also upload the file from your PC. The Nutanix cluster will automatically convert the disk file so you can use it in VM creation, creating the disk with the "clone from image service" option.

The second scenario covers situations where the source file can't be directly accessed by the cluster. This uses the filesystem whitelist setting from Prism to give an external machine access to the datastore used by AHV. Basically you mount the Nutanix container like any other NFS export and copy files to it. This is particularly convenient for pushing VMs from ESXi hosts but afterward you still need to convert the vm disk files. The primary method for this is to use the image service, but it is possible to use a CLI method. The command-line method uses the same utility as the image service but doesn't index the vm disk in the image service for later use. Instead, you'll have to utilize the filesystem path of the converted disk file when creating a VM from this disk, rather than selecting the disk image from a list.

The third and last scenario in the article describes pushing files to the Nutanix container using SFTP over port 2222. The follow-up steps here are the same as the second scenario, converting disks with either the image service or the "qemu-img convert" command.

Among these options you should be able to find a workable method for your situation.

If you’d like to read more, Alona offered some further insights in this post.


This topic has been closed for comments