Upgrade from ElectricEel to Fangtooth, VM migration when?

Anyone have any idea when the VM Migration is going to be available? I’ve followed the manual migration steps several times on my test system and so far none of my VMs migrate over and I have to recreate them from scratch. Not a big deal with one or two but when you have lots (more then 10) it’s a pain. I get the need to update things but this seems like a big functional misstep to not have this on release. :frowning:

1 Like

I keep hoping that someone like stux will create a script that will do the heavy lifting to help us move our VMs. Right now that is keeping me on EEL…

1 Like

@davistw - ditto for me as well. would be nice to get some kind of offical statement on this. personally i’m getting tired of getting my system setup and working great and then after an upgrade having to reinvent the wheel cause some component changed yet again. :roll_eyes:

Well it was in the documentation that VMs are broken with this one which is why I am still on EEL

yup, agreed but yet a new version was released anyway… :thinking:

What makes you think it will be? I’m not aware of any statements from iX that they’re ever going to automate migrating VMs from 24.10 and prior to 25.04 or later.

Why? TrueNAS is just so poorly-suited as a hypervisor that I’m constantly amazed anyone is using it for non-trivial virtualization workloads.

@dan you are correct but if they aren’t then it should be stated explicitly so people can prepare to rebuild things manually after an upgrade. also the documentation states that VMs are “broken” which implies a fix is in the works and other threads have stated that a “fix” is coming. I’m simply asking when that “fix” is coming. iX has done this in the past with moving from FreeBSD to Linux. i’m using the VM setup under TrueNAS for my personal lab. I use it for HAPROXY and to test VMs for my lab. It allows me to run systems and test things out without having to have lots of systems running to support a virtual environment. TrueNAS is ideally suited to provide a one stop shop for hosting of this nature. I also disagree that it’s poorly-suited as it works well and has for me for years since the BHYVE days under FreeBSD.

again, we can debate the merits of VMs under TrueNAS till the cows come home. the point of the thread is to try and get some offical stance from iX on this topic. that is all i’m seeking so let’s please focus on that. :slight_smile:

(citation needed)
Here’s what the release notes say about migration:

I don’t see “broken” there. Nor do I see it here, nor here. The closest thing I see is “experimental,” further on which they say not to migrate production VMs until the feature stabilizes in “a future release.” I’d like to see a statement from iX that they’re planning a full-featured automated migration with 25.10, but I haven’t seen any statement yet that any such feature is going to happen.

Really, a true hypervisor OS (ESXi, xcp-ng, Proxmox) would be far better. Run TrueNAS as a VM under that if you want a single piece of hardware.

This is definitely a side issue, but I’ve run VMs under CORE, SCALE 24.04, and SCALE (sorry, “Community Edition”) 25.04. It’s painful. It’s so limiting. Initial experience with 25.04 suggests it’s less painful than under 24.04, but that’s with one VM for a few days. VMs under Proxmox are a vastly more pleasant experience than under any version of TrueNAS.

You do you, of course, but this whole fiasco is part of why I recommend against TrueNAS as a hypervisor for any but the most trivial workloads.

And yet it is used extensively. Mybe not for enterprise work but still it is used a lot.

1 Like

okay, broken might not be the official statement but broken they are as they don’t migrate and the manual steps don’t work for any VM i’ve tried. the wording in the version notes “implies” a fix and again as I asked in the thread i’m just curious as to when or are we going to get stuck on EE for the foreseeable future or should we bite the bullet and start recreating.

Instances are an experimental feature intended for community testing only. Users with production VMs on TrueNAS 24.10 should not upgrade to TrueNAS 25.04 until after this experimental feature stabilizes in a future TrueNAS release.

honestly, i use it in my company quite a bit cause it provides the ability to have AD integration, SMB shares, printing and VM services without having ESX hypervisors at small remote offices.

This is the only mention I’ve heard about any migration, but they don’t explicitly say migration from the previous VM implementation.

@troy you are right, doesn’t say migration but does state that the goal is the ability to seamlessly bring your machines in and out. thx!

In the last two seconds of the clip, he says

… and some migration capabilities

Hopefully that means we can read into the statement in the docs

Due to configuration incompatibilities between the previous KVM hypervisor (TrueNAS 24.10 and earlier) and Incus in TrueNAS 25.04, existing VM configurations do not transfer automatically during the upgrade.

Which specifically says Incus in TrueNAS 25.04 , so maybe Incus in a later version of TrueNAS can overcome the configuration incompatibilities

Removed - I misread the post…

I’m hoping that it can but IMO it’s such a widely used feature it doesn’t sit well that it was pushed out in this state.

I saw your post before you deleted it, but that brings up another point of confusion

This is were the wording is bad on the part of IX and I really wish the would use correct terms to explain. Technically they replaced libvirt with incus. The VMs are still using KVM and QEMU.

1 Like

Ah, now that explains things a bit more as the release docs as you quoted imply a full replace of KVM but they just changed out one of main IO libraries. This explains why the disks and such under my existing VMs are all inaccessible under Fangtooth.

I have two VMs that run in TN 24.04.02. One is a Linux VM with portainer that runs hundreds of docker images, and the other is a Windows machine that runs security camera monitoring and other tasks. These VMs have been perfect with no issues through reboots, SMB shares, etc. They always fire right up after a reboot with no babysitting. Simple replication each night. I chose to deploy via VMs rather than the Kubernetes/Truecharts garbage that TN used to have. This way is far easier to administer and maintain IMO than even their new native docker.

So… if we upgrade it will “break” these VMs? Or… does it just require that we migrate them by hand using their existing images?