Just wanted to say thanks for absolutely shafting the community with the 25.04 rollout where a simple popup was too much of an ask to alert that your whole system is going to be broken. The whole point of going with TrueNAS was for the long term stability, instead, apps stopped updating and there’s zero warning when performing the update you’re about to be left with a useless system. I set this server up when I had more free time, no idea how many hours invested getting all of the VMs and apps setup where everything was ticking along happily. I don’t have that free time anymore and now everything I used is gone (okay, not gone but might as well be). Thanks for this gigantic pain in the ass.
Mods, I get this is a new account and I’ve contributed nothing, I expect this to get nuked from orbit if it even sees the light of day, I wish no ill will within the community and I am grateful for FOSS projects such as this…I just need to vent a great level of frustration over something that could have been so easily avoided. I’d happily run a less-than-ideal setup for a bit longer that at least worked while I wait for time to do the upgrade properly but now I just feel stranded with no access to anything. If I’m starting from scratch all over again I don’t see much point in using TrueNAS as a base going forward when I can only expect this will happen again in the future.
If someone somehow made it this far I expect you’re just as frustrated as I am so please feel free to continue my rant.
In any case, I hope everyone has a great day, maybe iX Systems can use this as a learning experience or something and maybe provide some sort of a heads up next time.
And I have stepped through the doc about restoring VM’s but the one I tried is not going well and I don’t have the time to be troubleshooting that process at the moment.
I hear your frustration and I understand how mind-numbingly difficult it can be to get stuff set up and running as you wish, only to have something destroy it all. It’s one reason that I started to document for myself necessary steps to rebuild obscure stuff like deploying SSL certificates to Mikrotik routers, Pi-holes, etc. since usually, this is a one-and-done type event but then something happens and you get to relearn all that you thought you could leave behind.
At the same time, iXsystems has been extremely open about the trials and tribulations associated with getting Kubernetes, K8 or whatever it was in the previous generations to work and why they decided to switch to Docker in 25.04. They mentioned this multiple times in the release notes, including the lack of IP #s by App, which is a bummer but which will allegedly be addressed officially soon.
I made a point to move my SSL distribution system into a dedicated jail for this reason. That makes the system more portable and less subject to vagaries of iXsystems changes.
Bottom line: Read the the announcements and release notes before making upgrades? Anyhow, too bad you couldn’t just roll back to 24.10. Best of luck!
Yeah I’m not bothered by whatever system they prefer to use under the hood for containerization and virtualization, I’ve used many and can adjust accordingly, provided I am made aware. I spend very little time in forums and such these days and having gone through 3-4 TrueNAS updates in the past, even seamlessly backing down when 24.10 wasn’t playing nice with my GPU passthru (Intel ARC in Jellyfin) with really no trouble at all. Maybe TrueNAS is just a victim of its own greatness lol. They lulled me into the false sense that everything would be fine, and if not then it would be an easy rollback that didn’t take much longer than just doing a reboot.
To be fair, I got hit by both since I rolled back from 24.10 to 24.04 when 24.10 wasn’t handling my Intel ARC GPU passthru too well lol. So my apps are also gone.
Yeah that is a lot of shit to lose all at once. I get it. But you can just not upgrade and stay behind too. We get people showing up on Freenas 10 like a hermit emerging from a coma, moving to Fangtooth.
This shit ain’t Solaris, we don’t get a decade of binary compatibility, but the price is good. I was hardcore on Core 13 until earlier this year when Scale was evolving rapidly. I survived the TrueSharts massacre at Wounded Knee, so I know losing apps.
I probably would have stayed behind but a few apps were acting goofy, I thought it was strange when I hadn’t seen an update for them for a bit and given the easy rollback in the past I thought worst case scenario I’d just do that again.
Yeah, it (conceptually) follows the GNU naming convention. Making a free (as in beer) left-handed implementation of an expensive, professional, mature OS and toolset. The work that has been done to drag ZFS kicking and screaming onto Linux was probably incredible. Took some dudes at Sun a while to come up with it to begin with. But as time goes by, the BSD’s are losing nerd love, because Linux is sexier, or something. This is the best way to honor it’s memory, to copy it and keep working on it.
Yea, there are some cases where BSD had the idea first (like jails) or Solaris had it with “Zones” but Linux took those ideas, ran with it, and more importantly made them popular and commercially successful.
As a general response, I think we’ve been pretty transparent about where the biggest areas of “churn” have been. Coming from our BSD background into Linux, the Storage side of TrueNAS SCALE / Community Edition was pretty solid early on. The biggest challenges we’ve had has been marrying that up to modern containers and virtualization capabilities, which still are evolving as we experiment and push towards something we feel is more enterprise-grade, like the storage capabilities are. Docker is in pretty decent shape now and risk of that regressing is low. We will update with more knobs and features as time goes on. Containers (LXC) and VM’s are the next area which is under significant churn, but will eventually settle down as well.
But lets not look at the past with all rose-colored glasses as well. Jails / Plugins on CORE were pretty damn bad. Even with all the bumps in the road for proper Linux container support, what we’ve done has made our community usage explode. So many folks historically dismissed CORE outright due to how ancient and archaic running jails/plugins on FreeBSD are. That situation is only getting worse as time goes on IMHO. That is a technology stuck in the distant past and not going to evolve much further.
Yeah to be clear I’m not complaining that something broke, I get things change, y’all are trying to make improvements and many times that comes with new issues popping up. My complaint is just there was no warning when doing the update giving a heads up that this will break VMs and there will be no option to revert. It wasn’t an unforeseeable event, the impact a change like this would have was already known. I’ll concede that I should have went over the release notes, but iX Systems could also do better.
That part we agree on, we could have put that in more focus to warn folks better. Additionally we’ve agreed internally to not do these kind of sweeping / breaking changes again with introducing “experimental” features like this. So we’ll tackle it from two angles.
If the wheel was Not Invented Here, and is not chained by the GPL, the Linux spirit is to re-do the wheel from scratch to square things up.
Or you will drop it and try yet another VM manager, as happened with k3s → Docker.
It’s fairly clear that iX is trying to make “something” out of its new Debian base OS without a clear idea of what “something” may eventually look like. Scaling out? Yes, then no. Kubernetes? Ditto…
And thanks to the hint by CaptainMorgan that Enterprise does not provide VMs, it is now faily clear that iX is very free to do whatever it wants with the feature, including breaking changes, until is has matured enough to eventually make it to the Enterprise branch.
Let me spare @dan the trouble of pointing out that plugins were “pretty damn bad” because insufficient resources were put behind them. If you want to drown your dog, accuse it of rabies.
I still run Core 13.0, and the only way everything still runs is because I do NOT update. Otherwise some C libraries are replaced, everything is broken etc.
That’s why you have the base revert. Does Community edition not have Boot Environments? I used to just rollback or restore a snapshot and manually update components.
When I saw the post title I thought this thread might be about something else, and as the conversation has cooled a bit… I have been through every version from FreeNAS 8 to CORE to SCALE to CORE to SCALE, and even a brief time with Illumio. The one thing that I was always sure of is that my data and services would be fine as long I snapshotted all the datasets and downloaded the configuration before, and didn’t upgrade the zpools until I was sure everything was working. zfs with FreeNAS/TrueNAS’s configuration portability have not let me down in what must be around 15 years now. Thank you.
