Backup and Restore VM to new TrueNAS scale server

boredandlazy

Dabbler
Joined
Jul 16, 2023
Messages
11
Hi,
I currently have a TrueNAS server running with a ZFS pool running on 4 SSDs that has my dockers and VMs etc.
For various reasons I need to redo my NAS build with a different motherboard etc. But I will be taking the CPU from my current build over to my new one. Therefore it isn't possible to have both online at once in order to use ZFS send/receive.
The new motherboard also only has 2 M2 slots available, so I also can't just relocate the 4 SSDs.
What's the best way to backup my VMs so that I can restore them on the new build?
Thanks.
 

probain

Patron
Joined
Feb 25, 2023
Messages
211
Got a spare drive you could use as temporary storage? You could then replicate the dataset containing the VM's zvols, onto that separate drive. Risk should be quite minimal then.
Obviously all of the warnings of only having data in one place etc etc, applies!
 

boredandlazy

Dabbler
Joined
Jul 16, 2023
Messages
11
I actually just had a thought. If I manually make a copy of the raw disk file from the current zvol, then make a zvol with the exact same size etc. on the new server I should be able to transplant the raw disk into the new zvol from the backup correct?
 

boredandlazy

Dabbler
Joined
Jul 16, 2023
Messages
11
Got a spare drive you could use as temporary storage? You could then replicate the dataset containing the VM's zvols, onto that separate drive. Risk should be quite minimal then.
Obviously all of the warnings of only having data in one place etc etc, applies!
I should be able to do this with usb attached storage? Then I can do the replication and just save the config of my Truenas, and when I load in the config on the new server it should pick it up from the usb drive? I can then just replicate it back to the SSD pool on the new server?
 

probain

Patron
Joined
Feb 25, 2023
Messages
211
I would personally hook up a high-capacity HDD via SATA internally. Not USB, since I read here so many times how USB is to be avoided.
Then replicate all of the relevant datasets I really need. Then export that 1 drive temporary pool.
Followed by a download of all relevant config files etc. Including screenshots of VM devices/configs etc etc. Better to have redundant datapoints with settings for restoring. I often still manage to miss one or two small things, when I do re-installs for things.

Extra thing to think about.
I would also expect if you're on Cobia, that any new pools you create. Would then be made with the newest ZFS flags. And as such would be incompatible with Bluefin.

My warnings still apply. And do a bit more reading/research. Would also be wonderful if anyone else could validate my train of thought.:smile:
 
Top