Does this new configuration of the VMs overwrite the existing Electric Eel configuration of the same VMs? In other words, does it make it impossible to seamlessly revert to a previous boot snapshot?
Rolling back to previous BE will still work.
1 Like
I will never understand how iX ever decided to replace a perfectly working, rockstable solution by a half-baked alternative that doesn’t even reach half-way. At least they should have kept Core fully alive and updated until there was a solution worthy of replacing Core. Which, from all i am reading in this forum, is definitely not the case.
3 Likes
Fork the code and run your own version. They needed to do this to grow as a business, I am sure it kills them that your hobby system has to change to Linux if you want to keep up. Scale is working great for me and solves several legacy BSD issues.
4 Likes
So what do you need that you know SCALE currently doesn’t do?
Ah, nothing. Figured.
6 Likes
And this where you are wrong. My system is mission critical as i use it for my consulting business and all my business data is housed in Nextcloud to which others such as my accountant also have access to so i am not in a position to “just” switch over to an entirely different system and set up everything from scratch again. I am now forced to buy or build a new server, install Scale and Nextcloud again and test it throughoughly before i can migrate my data to the new one and switch over the domain after everything has been proven to work and being reliable and this while Core until now has been perfect for my use-case. From the forum it appears there are still a lot of diseases with Scale and especially with apps, whereas jails work flawless. Also i don’t like that i cannot assign a dedicated internal IP to an app. This is not a FreeBSD vs. Linux discussion…this is a Core vs. Scale setup and functionality discussion.
3 Likes
Jails are simply the superior container technology. With a bit of luck you can continue to run your Nextcloud jail well into FreeBSD 13.5. But the days of CORE are counted.
It’s not difficult to set up a jail on plain FreeBSD with iocage. Even less when you are familiar with CORE jails already.
Then dont switch I guess? Nobody forces you.
iX still supports their Enterprise products running Core.
Got a support contract with iX? Or is this a business use of open source software and a complaint that it’s not continuing to be free? Because from where I’m sitting this appears incredibly…shortsighted? Selfish? If this is a cost free option.
To alleviate some of your fear: I have seen zero diseases with apps. That’s from 23.10 to 24.04 then 24.10 and now 25.04. They work great. All my jails (all two of them ;)) were easy to move over as apps, as their data was already in a dedicated dataset.
That you definitely want, a dataset for the app data, independent of the iocage dataset.
The one thing I have seen in the forums are people struggling with 40+ apps on HDD and a hard-coded 120s timeout value. Fixed in 25.04, and maybe also those users may consider a) apps on SSD pool and b) why on earth 40+ apps. But if one must, then on SSD seems saner.
This looks like a good guide for nextcloud plus collabora with ngnix as the proxy: A full guide for installing Nextcloud on TrueNAS Scale with Collabora integration and outside access through Nginx Proxy Manager
I don’t think you need Incus for this, Nextcloud works just fine as an app. You can of course wait for 25.04 in April and make it an Incus container, more like a jail - but for this specific use case I’m not convinced it gains you anything significant.
You can set this up as completely separate hardware, but I don’t think you have to.
Make sure datasets holding nextcloud data outside iocage exist and have both snapshots and backups, make a backup of the config including key for encrypted parts of it.
Try the nextcloud setup fresh on a CE or SCALE VM, just to get a feel and take notes so you’re not learning during maintenance.
Remove the boot media, put in fresh boot media, install and restore config, run through your nextcloud setup as documented earlier, connecting your existing datasets for the data.
Do NOT update pools until a month or two of solid running.
All in all half a day to a day of switchover effort, not counting the time to dry run the setup - “just” announce there’ll be a little downtime.
Push comes to shove, put the old boot media back in, you’ll be right back where you were. Go back to drawing board to figure out what step might have been missed in migration. That’s worst case and I trust you’ll have sufficiently good notes from the dry run that you won’t need to.
If you’re really keen on not using apps and prefer maintaining nextcloud yourself inside something jail like (though see earlier - maybe more maintenance effort than desired without much or any gain), then you can reasonably try that idea on a CE VM with Incus container, then implement in April or May when CE is GA. No real rush after all, your CORE jail still works.
My experience has been that SCALE and now CE feels easier with apps than CORE ever did. Pre packaged apps actually work, I don’t need to manually create jails, I don’t need to hop into jails for updates and maintenance (or create cron jobs for that). After deployment, every so often I can click “update all” in the UI, and it does, and everything “just works”.
Apps survive across updates, even with the switch from k3s to Docker. I don’t need to manually handle OS updates for/in the jails.
I was wary of the switch from CORE to SCALE, which is why I waited for 23.10 before going for it. Turns out it was easy, and my fears were unfounded.
I did hold on to the iocage dataset for a really long time - finally deleted it just prior to moving to 25.04. After a year and a half, I’m now confident I won’t ever need to move back to CORE for any reason.
3 Likes
People come here asking for help solving issues with jails every month.
1 Like
Fixed that for you. “Jails work flawless[ly]”? Granted, you don’t see many posts on this (i.e., new) forum complaining about them–because relatively few people are hanging on to CORE any more. Look at the old forum and say with a straight face that jails work flawlessly.
And that’s not even addressing the apparently-unfixable security flaws in iocagge.
1 Like
Pretty good thread to read but a lot of it complaining about being forced to migrate to SCALE. I have a CORE system that will end up dying still as a CORE system. That is my choice and I am very comfortable with it. I like the jails and just how rock solid it is for me and I trust it.
With that said, I also have a SCALE machine that I built last year to replace the CORE machine, but the CORE machine has been running like clockwork. Almost 6.4 years on three of the main drives, only replaced one a few years ago. The drives will fail and I will retire that machine, probably sell it to someone who wants a reliable system, they just need to add hard drives or NVMe, whatever they like.
As for reliability, my SCALE system has been very reliable and I screw with it almost every day. It is on ESXi 8 (free) so that works great also for those interested.
As for Fangtooth, it works, not a lot of time for me to play with it. It just came out in Beta so I need to upgrade to that version and test the crap out of it, for what I do.
While I don’t do this often, I would like to praise the iXsystems teams as they are significantly better these days at developing a stable product. Who remembers Corral - FreeNAS 10? Need I say more? I do have my little complaints but as a whole, good job to those teams developing and testing. We haven’t seen another Corral happen since then.
Here is another positive spin, it will be nice to have a single TrueNAS software type for these forums, How many times do we ask what system are they running? The GUI’s, command line, etc are different so this will ultimately be nice to see from my perspective when helping someone out.
5 Likes
I think there’s a significant sense of betrayal by iX. When SCALE was announced, they unambiguously stated that development on CORE wouldn’t be affected, and that it’d continue to be as actively-developed as it had been. That, well, didn’t happen. And whether iX knew it at the time or not (and I think they did), there’s a perception–that IMO is quite justified–that we’ve been lied to. Some of us made (relative) peace[1] with that earlier, some later, and others have yet to. And that, as much as anything, drives the complaints about needing to migrate (IMO).
Then, of course, there’s the issue that change is always painful–and since all the UI development that’s happened over the past 3-4 years has gone into SCALE (they couldn’t even be bothered to take the working web shell from SCALE and put it in CORE), not into CORE, the UI has diverged pretty significantly.
Few do these days, I think, and those who do still shudder at the name–it’s The Release That Must Not Be Named for a reason. And you’re right that they haven’t had a screw-up of that magnitude since, but you’re setting the bar awfully low.
Sure, ten years from now when everyone’s made the switch. There was a thread the other day from a FreeNAS 9.2 user, so “ten years” probably isn’t an exaggeration.
In my case, “relative” peace means that I accepted that CORE was dying and SCALE was the way ahead, but iX’ poor (to say the least) messaging on this question is one of many things in the past few years that have greatly reduced my trust in them. ↩︎
3 Likes
We may have seen different messaging, or interpreted it differently.
What I saw was: “Here’s corporate speak that boils down to: We have Enterprise customers on CORE. We’re going to keep those happy while we move our product line to SCALE.”
3 Likes
Here’s the initial announcement:
It’s important to note that the SCALE project doesn’t change the plans for the continued support and development of TrueNAS CORE and TrueNAS Enterprise on FreeBSD. The full suite of software will be very complementary to each other.
Of course, a lot has changed since then–the “scale” piece has gone right out the window with the de-scoping of clustering. But what they were saying five years ago was clearly that CORE would continue to be a first-class citizen. See also:
This isn’t so much about reducing support for anything
Of course, there were some in that thread who (correctly, as it turned out) understood that iX was putting CORE out to pasture. I (foolishly, apparently) took them at their word.
…and
If you currently are happy with FreeNAS as it sits today, you can expect to keep updating it on BSD going forward.
1 Like
Even if we consider their blog post from last year, the one stating that CORE was moving to a “sustaining engineering phase”, that’s still 4 years since the statements you link.
4 years is a long time in the context of software development, don’t you think?
2 Likes
True. But the (apparent) truth of the matter is that CORE was in a “sustaining engineering phase” from roughly the time SCALE was announced. iX would likely point out in response to this that it’s seen two major releases (12.0 and 13.0), one semi-major release (13.3), and quite a number of point releases and bug fixes, which is true enough. And in the case of 12.0, there was at least the rebranding work happening.
But 13.0 and 13.3 don’t appear to be anything more than “rebase to upstream,” and in the case of 13.3 remove a few features that don’t work anyway.
2 Likes
I have to completely agree with that statement. Although I could see the writing on the wall (pessimist I guess).
And as I said, I will stick with CORE on my main system until it dies a horrible hard drive death, then migrate to the SCALE system, which is of course in use as my backup server and testing server.
The only thing I really do not like about the advancements on SCALE is the hardening of the system. I’m certain that corporate users may enjoy it, maybe not, but I know it makes some stuff difficult to do. This too shall pass.
1 Like
Presumably iX wanted to get enough homelab enthusiasts onto Scale in order to properly test moving from FreeBSD to Linux, while at the same time not creating nervousness or uncertainty with paying corporate customers on BSD, or for that matter pissing off BSD-loyal homelab enthusiasts, before they were absolutely certain about their architectural choices. Tough to get this right - they could have been much more vague instead, which likely would have created (even more) speculation. Or conversely been clearer all along regarding Scale and Linux being the way forward - but then risking painting themselves into a corner if they didn’t manage to pull Scale off.
I agree as well that the corporate speak can be a bit provocative - and seemingly much more vague recently. Particularly this recent ”unification” announcement which actually seemed pretty clear in its intent but still peppered with caveats and maybes, not sure why that’s needed any more.