Solved

Metro Availability and VM snapshot

  • 25 February 2016
  • 18 replies
  • 2410 views

Badge +7
Hi guys, finally i'm a new Nutanix customer and yesterday we have install two blocks in two sites with 3+3 nodes and 2x10Gbit connection that is in the same lan.
The goal is to create a metro availability for all VM's, but after the Nutanix and vsphere configuration the partner tell me that isn't possible to schedule a "protection domain" with VM snapshots in a metro availability configuration. It is possible only to backup and resotre an entire container, is that right?

Thanks in advance
icon

Best answer by mmcghee 1 March 2016, 15:15

Today the restores are driven from CLI. For example, from ncli you would do the following and restore the snapshot to a new location



pd restore-snapshot name=DRT snap-id=16788 path-prefix=/restore



You'd then utilize the files in the restored location or copy them as needed. You can then delete the folders for the other contents of the container which are not needed.
View original

This topic has been closed for comments

18 replies

Userlevel 7
Badge +30
We added interop for Metro and Snapshots in 4.1.4, so as long as you're running that or higher, you should be able to run both configurations at once

Let's take a step back, what are you wanting to do?

It sounds like you want metro, and thats great, but you also want to be able to restore a single VM from a snapshot?
Badge +7
Hi Jon and thank you for the reply!Yes i want metro and i want to make snapshot and restore single VM from it.
A stupid (not for me) example, last week cryptolocker come to us another time, because marketing manager click a link inside a personal webmail and in 30 minutes all Marketing department folders was crypted.No problem, with nutanix protection domain with 15 minutes snapshots (4 jobs per hour) I've restored the VM disk and extract the folder before the virus come in.Now if i put my file servers inside a metro availability and I want it to cover the B.C. plan, i can't make a restore VM/folder and i lose the fantastic protection that give me a frequent snapshot with no performance impact to the hypervisor.
In my opinion, but I think for more others people and environment this is very useful.
Userlevel 7
Badge +30
If memory serves me correctly, as of 4.1.4, you can do metro and snapshots. I dont have a test setup in my lab to try, but you could easily try this with a template VM, or heck, even an empty shell VM, and try it out.
Badge +7
Hi Jon, no i've just double check and if you have a metro availability all VM's inside that container isn't available for Async DR/snapshots.
Userlevel 4
Badge +18
Hello,

Snapshots for Metro are done at the container level, so the option to configure them does not show up within the async protection domains. You need to configure the snapshots by updating the Metro Availability protection domain.

-Mike
Badge +7
Hello Mike,

I've just try and yes the snapshot in the M.A. it's only at the container level, but i can't restore a VM from that kind of snapshot. Is it possible? If yes how?

Thanks

Userlevel 4
Badge +18
Today the restores are driven from CLI. For example, from ncli you would do the following and restore the snapshot to a new location

pd restore-snapshot name=DRT snap-id=16788 path-prefix=/restore

You'd then utilize the files in the restored location or copy them as needed. You can then delete the folders for the other contents of the container which are not needed.
Badge +7
Great if it is possible by cli, it's far better then nothing!
I've try from a CVM cli, i don't know if it's correct, i've put the container name after "name=", but i receive this error:

pd restore-snapshot name=v2-to-v1 snap-id=154241 path-prefix=/restore
-bash: pd: command not found

I'm on 4.5.2.1.

Many thanks
Badge +7
Mike, Google help me to find the mistake, i need to put ncli before, this is the correct syntax:

ncli pd restore-snapshot name=v2-to-v1 snap-id=154241 path-prefix=/restore

It's a great news, now i can scheduled a hourly snapshots of both container (site 1 and 2) and in case of restore i've a little procedure to do.

Any plan to put in the prism U.I.?

Thank you!!
Badge +7
Hi Nutanix eng., any news to snapshot at VM's level in metro?It's a very missing feature in my opinion and the bad thing is that simplivity can do it 😞
Userlevel 7
Badge +30
from the previous messages, it sounds like you were able to get this working via the CLI, is that correct?

Also, for the sake of tracking, can you file a support ticket with priority RFE - Request for Enhancement, and ask them to link your ticket with this internal engineering ticket: ENG-34942

That will track customer demand for fixing this up in the GUI/Prism.

Jon
Badge +7
Hi Jon, yes, it's possible to do that via CLI, but only the restore of an entire container.
Anyway Isn't possible to restore a single VM like a NON-metro cluster with Async DR.
Actually if you are in Metro and you need to restore a VM, the first step is to restore the entire container by CLI then delete all useless VM's in this container and add the only one you need in the vsphere. The result it's the same like Async DR, but with a lot of manual operations.(I hope to understand my basic English).
I'll open a ticket following your istruction with this integration.

Thank you for the support, have a nice day
Badge +7
Hi guys, i've open the request feature ticket in July, but sadly I can't see any news during the updates of 4.6.X and 4.7.X also. This morning, again, I've used the CLI procedure to restore a VM.
Any plans to implement it in the future releases?

Thanks
Userlevel 7
Badge +30
I pinged the ticket internally to see where it's at. As you can imagine, with any feature request, implementation is a matter of demand and priority. I can tell you there has not yet been that much demand, but we do recognize it as something that we should have.

The feature owner for this is MarkNijmeijer, so he can speak to the ins and outs.
Userlevel 1
Badge +1
Hello,

Any news about this feature request ?

Thanks
Marc
Userlevel 7
Badge +30
Still in the queue, as it hasn't gotten a lot of customer demand (or rather, other features have more than this one).

If you've got access to the Nutanix support portal, please file a ticket with priority "RFE" and reference internal ticket ID ENG-34942, that will link your request with the backend ticket to help us properly track demand
Userlevel 1
Badge +8
very needed
i have 2 costumers that wants it
Userlevel 7
Badge +30
Thanks EladRosen. Please follow my previous guidance above if you haven't already.