ESX 6.5 and storing snapshots on an NFS mount

[PaulKessler]

I recently added an NFS mount to my datastores and I would like to start storing my snapshots there. I have followed the procedures outlined in this kb article: VMware Knowledge Base  but after I add the following to the .vmx file and unregistered and registered the VM it says the configuration is invalid. I used the command a command line to grab the path of the mount point. I must be missing something. Any assistance would be greatly appreciated.

workingDir = "/mnt/array1/Snapshots"

snapshot.redoNotWithParent = "true"

esxcli storage nfs list

Volume Name  Host            Share                  Accessible  Mounted  Read-Only   isPE  Hardware Acceleration

-----------  --------------  ---------------------  ----------  -------  ---------  -----  ---------------------

Snapshots    134.244.118.54  /mnt/array1/Snapshots        true     true      false  false  Not Supported

[a_p_]

Two thoughts about that:

• The workingDir has to point to a folder on a mounted datastore, i.e. something like "/vmfs/volumes/<datastorename-or-uuid>/<foldername>"
• Remember that snapshots are used in chains, i.e. all snapshots in an active snapshot chain are in use, and the weakest/slowest chain link dictates the VM's speed. It is recommended to store snapshots on the same tier as their base disks. See e.g. https://kb.vmware.com/s/article/1015180

André

[PaulKessler]

The NFS share is mounted as a datastore, I have even created a test sub-folder in the datastore browser and tried to use that folder and it still does not work. The reason for offloading snapshots is based solely on free space on the direct attached storage. I could simply use the NFS mount for full backups but I was trying to make things easier for the handful of developers on that host. Other suggestions are welcome!

[daphnissov]

I'll add on to what André has said and state that offloading snapshots to a secondary datastore is, in general, a bad idea and fraught with peril. Now you've doubled your blast radius–if anything should go wrong, at all, with either of those datastores (the one holding the base disk and the one with the snapshots), those VMs are hosed. It's basically the same concept as spanning a single VMFS datastore across multiple backend extents. I've seen this blow up multiple times for customers running in production and it's not a pretty picture (unless you have really good, regular, and tested backups). A better solution may be to simply move some larger and less-demanding VMs over to this NFS storage thereby freeing up more space on the other datastore while giving you some snapshot breathing room.

[PaulKessler]

I think this is the route that I will take, thank you!