I’ve been trying to bury my head in the sand for the last 1-2 years but as the EOL for FreeBSD 13.5 barrels toward us, it’s time to take it out and at least think about what this really means.
I’ve been running TrueNAS since FreeNAS 9.2.8 and finally (well, since about 11.2) have it working just as I want. I have jails running emby, Nextcloud, and WordPress and have a few bhyve ubuntu 24.04 VMs running CrashPlan, Mattermost and various other services using Docker containers.
Both my HDD and SSD pools are fuller than I’d like ~75% but still performing fine.
I can’t be alone (can I?) wondering where I go next?
As a scale user from minute 1 i always wonder what core users fear would happen if they switch to scale.
I get it that in the first 1-2 years there were issues and i went through all itterations of scale and the various issues with the kubernetes based apps system, but since 24.10 switched to native docker and all the changes regarding arc usage i don’t see an issue anymore…
If you already use vms for docker you should be able to migrate them to the native docker implementation, or keep using your vms, they should auto migrate from bhyve to libvirt.
So what are your actual concerns regarding scale or migrating from core to scale?
For me, I think it’s because my use case is simple and I use TrueNAS as just simply… a NAS and nothing else. For that, I value stability first and foremost and SCALE’s very turbulent beginnings that you mentioned is a part of it.
The other factor is all my other servers are also FreeBSD-based and I just prefer FreeBSD way of doing things as a server (ie. primarily VNET jails + pf) and also the fact that they don’t constantly reinvent the wheel the way Linux does. Now, I do use Linux for my workstations, but for my servers, I simply prefer stability and boring and not have to constantly adopt to changes.
As such, when CORE finally sunsets, I would probably just retire TrueNAS entirely and migrate it to my already existing FreeBSD machines, which shouldn’t be too much problem.
That’s for me, obviously most people likely find value in the Docker apps and that is ok, it’s just not my cup of tea.
You are not alone. I did leave CORE but only because I removed my primary computer. I was running one CORE and one SCALE, now just SCALE.
Here is my philosophy, and I’m being serious: “If it ain’t broke, don’t fix it.”
So, if what you have is working for you, why change it?
Okay, well one thing to consider is, is your NAS exposed to the WAN? If it is, then you could have a security risk. I’m not saying you do, but over time it could happen. If your NAS is not exposed to the WAN, then security is much less of a threat. But as you can imagine, TrueNAS (the company) will not be updating CORE unless it must, and even then, that too will end.
I’ve been here since FreeNAS 0.7 (Yes, I’m older than Dirt) and have been here for every step of FreeNAS/TrueNAS/Core and Scale. It took me over a year of running Scale before I felt comfortable shutting down my Core server.
Again I have to say it, If it ain’t broke, don’t fix it.
But if you do decide to “fix it”, due diligence is required to find out what will not migrate over. And "migrate is loosely used here, many of the migrations that were expected to work, didn’t work. Do not upgrade your ZFS features, I’d say most if not all home user would never need to do this and it would give you a possible path to return to CORE.
Good luck if you take the path of migrating from CORE to SCALE.
Because you cannot deploy new jails or update existing ones - which might or might not be your primary use of a TN CORE system. It is for me, so although I have also been running a SCALE/CE system in parallel and still do, I switched to Sylve from CORE. Jails are the most important FreeBSD feature for me anywhere - at home and at work.
My experience with Core has been quite intense, but with ups and downs.
Long story short: while chasing updates to keep using jails, I started hitting bugs in features that used to be rock solid on stable versions… And to run some specific service i slowly migrate to a VM with docker…
So with the Electric Eel release i bit the bullet and done the sidegrade, and i stick to general releases.
Since I’ve only ever used Docker (even for VMs thanks to dockur) and relied on third-party scripts for other features like multi-report, I haven’t had much to worry about until now, even though things have been a bit bumpy with certain releases
Storage only here, so I’m happy to fossilise on EOL 13.0 until I walk into the Sylve.
With both jails and Ubuntu VMs, you have a choice to make between migrating to work-in-progress Sylve and migrating to CE, re-doing your jails as “instances” or VMs. ANd I’m afraid you’re the only one who could answer.
But you have a bigger issue looming:
In this market, you really should start looking into opportunities to get more HDDs and/or bigger HDDs. Any purchase is going to be painful… but to do it in an emergency would be even worse. So plan ahead.
My answer to switching was to stand up a FreeBSD server and move all my jails over to it. I too like the FreeBSD way of doing things. It took a bit of work but It’s been running trouble free for a couple years now. I migrated to Scale on my main server and only run a couple apps on it that are only available via Docker.
Oh my, I didn’t know what Sylve is until you mentioned it and it looks uber promising, particularly this bit.
Why Sylve? Why not Proxmox, TrueNAS, or others?
Light on resources: Sylve runs in roughly ~384 MB RAM even with all major features enabled in a multi-node cluster.
And it’s also written in go, one of my preferred languages, I will have to give this a try. Looking at the UI, it’s clear that UI is inspired from Proxmox, the resemblance is quite apparent.
That is funny. But if the NAS is doing what it was designed for and the owner is happy with it, why mess with it. SCALE is not bug free, but in CORE, whatever bugs exist are not going to change, so running what works is perfectly fine. If the OP wants to cross over into SCALE, that person can, when they feel like it, not being pressured because someone says “the sky is falling”.
Always validate if the risk is worth the reward. Some people say yes, some people say no, or at least not yet.
FreeBSD manages memory better, especially in regards to ZFS and ARC, without the need to play with parameters. There’s a common theme with SCALE and CE users where they need to either limit the ARC or remove its cap, depending on workflow, Apps, and VMs. Without swap, there’s the risk of OOM, which you can browse in these forums for such examples.
It’s just not worth it to me. Core was a solid NAS. Jails are powerful yet light. If Core had been maintained like a first-class product, we would be running on a base of FreeBSD 15 with excellent memory management and a rock-solid NAS doing what NASes do best. We’d be using up-to-date versions of software, including Samba, NFS, and ZFS. We’d have an updated UI.
Any comparisons of Core to SCALE/CE are disingenuous. Not only is FreeBSD 13 two releases behind the current version, but the version of ZFS is also older.
The TrueNAS focus shifted to “Apps”. Sometimes it feels like TrueNAS is now an app center that also has NAS functionality as a secondary feature.
This is absolutely the reason why they shifted to Linux. Basically for access to (previously) k3s, and then Docker.
Actually, I find the switch from k3s to Docker interesting because I’d thought that TrueNAS is supposed to be first and foremost an enterprise product, but they actually listened to the home users and switched to Docker.
A VM does not have local access to the files on your pool but needs to use a sharing protocol like NFS or SMB. And if the jail runs e.g. Nextcloud and we are talking about multiple TB, that’s a bad idea and the main reason why jails running natively on the platform are preferred.
There is a community effort to keep “Core” alive - they forked the open sources and removed proprietary bits. Haven’t tested it, went with TrueNAS SCALE/Community instead.
That project seems stagnant or dead. There hasn’t been any update to a newer FreeBSD 13 version and now version 13 is EOL by FreeBSD. I don’t know if there has been any progress on moving to FreeBSD 14 or 15 as the base.
There are probably a few other threads mentioning the project, too.
Try and help improve Sylve instead. Funding secured for at least another year. Rough but working. Modern architecture. Improving at an incredible pace.
Oh, right, in my head that announcement seemed a lot more recent that a year ago. Also I thought they were already working on making it Freebsd14-based but the web page only lists it as a milestone on the road map…
