Will TrueNAS Core 13.1 have ZFS 2.2.2?

[Arwen]

The subject says it all:

Will TrueNAS Core 13.1, (expected in 1st quarter 2024), have ZFS 2.2.2?

Now I understand that 13.1 might be ZFS >2.2.2 due to other changes. Just want to see if that update is on the radar.

 

[Davvo]

Iirc it was confirmed it would get [at least] 2.0

 

[Arwen]

Iirc it was confirmed it would get [at least] 2.0

Don't we already have OpenZFS 2.1.14?

TrueNAS 13.0-U6.1 is Now Available

We are pleased to release TrueNAS 13.0-U6.1! This is a minor update to 13.0-U6 to update OpenZFS to version 2.1.14 and fix a data integrity issue discovered in that project. While this bug has been present in OpenZFS for many years, this issue has not been found to impact any TrueNAS systems to...

www.truenas.com

Thus, my question...

 

[Volts]

I'm pretty sure it will. Nightly from a few days ago already has 2.2.1-something (plus iX patches) and there have been more updates in the last couple days referencing 2.2.2 and all of the other hot ZFS activity.

iX doesn't use ZFS from the FreeBSD base, they maintain a port based on OpenZFS (plus iX patches). Look at the recently-active project branches here:

Branches · truenas/ports

FreeBSD ports tree. Contribute to truenas/ports development by creating an account on GitHub.

github.com

Branches · truenas/zfs

OpenZFS on Linux and FreeBSD. Contribute to truenas/zfs development by creating an account on GitHub.

github.com

 

[Arwen]

Excellent!

Yes, that is conclusive in my opinion. With the recent bug being fixed in 2.2.2, they will use at the bare minimum 2.2.2.

I just wanted to confirm it was not going to be 2.1.14.

 

[Davvo]

I just wanted to confirm it was not going to be 2.1.14.

How so?

 

[Arwen]

If they are using 2.2.1 in the nightlies, then that is higher than 2.1.14.

Plus, iX would have to include the bug fix in 2.2.2. Thus, my opinion that they will be using at least 2.2.2.

 

[Davvo]

Yeah my bad, I mean why do you require 2.2.2? Is there any special feature you need?

 

[Volts]

Yeah my bad, I mean why do you require 2.2.2? Is there any special feature you need?

More fasterer. Good prefetch and ARC improvements. ZStd early abort makes it more attractive. Various little nice things.

I want to do some testing with Blake3 and nopwrite.

 

[Arwen]

Yeah my bad, I mean why do you require 2.2.2? Is there any special feature you need?

No, I don't "need" OpenZFS 2.2.2.

I was seeing if Core would get some feature parity with SCALE. SCALE update 23.10.0.1 uses OpenZFS 2.2.0-rc4, so the update due next week would have to include 2.2.2 for the bug fix.

Perhaps too much circular logic for non-elves .

 

[victort]

Didn't want to start a new thread so, when is TrueNAS going to get FreeBSD 14?

 

[winnielinnie]

I think TrueNAS Core 13.1 is going to be based on FreeBSD 14.

 

[the_jest]

Didn't want to start a new thread so, when is TrueNAS going to get FreeBSD 14?

According to @Kris Moore in this Reddit thread, it's not.

 

[winnielinnie]

We have no plans for a FreeBSD 14-based TrueNAS at this time, and the 13.1 release will be a longer-lived maintenance train for those who want to continue running on the BSD product before migrating to SCALE later at some later date.

On the SCALE side, it is where the future of TrueNAS is going, all new features and development activities take place there now. It is where we are seeing the largest growth in TrueNAS adoption, breaking all kinds of records for us these past couple years. This goes beyond just "Converged Apps and VMs", but includes 'core' NAS functionality as well, where the basic NAS functionality has been at feature parity and beyond compared to CORE for some time now.

When I read something like that, it makes me nervous about Core's future, and hence my future as a TrueNAS user.


"[SCALE] is where we are seeing the largest growth in TrueNAS adoption, breaking all kinds of records for us these past couple years."

Part of the reason for SCALE's adoption can be attributed to the fact that Core is stagnant when it comes to its GUI, polish, new features, and even bugfixes in some cases. Imagine if Core (or a non-Enterprise derivative) received the same level of development parity as SCALE? Imagine if it had a robust Plugins/Apps system? Imagine if it got the same upgrade to its GUI? Imagine if it fixed longstanding bugs, such as the nearly useless Shell?

If the time comes, I'm unlikely to migrate to SCALE. Sorry, but for what can be done with jails/iocage, I don't want the overhead of K3s and "Apps" being stuck on "deploying". Jails are greatly underrated. @Patrick M. Hausen can attest to that. Not to mention ZFS/ARC as a first-class citizen on FreeBSD.

 

[Arwen]

I agree with @winnielinnie for many points. I can sum up mine with:

• ZFS ARC on Linux needs serious work, not just a bandaid.
• The various App issues, (un-related to actual specific Apps), need to be cleaner. Here are what I see from the outside:
  • More network flexibility
  • Easier access to ZFS datasets
  • Need method to total disable Apps, including App chart updates. For those that want NAS only, or just VMs
• Linux kernel is too unstable, (and the Linux kernel people WANT it this way!!!) So TrueNAS SCALE almost needs to both use a long term support kernel. AND then track kernel issues to see if they apply to SCALE, with potential back porting of fixes.

For the last, this would be similar to the RedHat Enterprise Linux model. RHEL uses a specific kernel and back ports curated fixes. This helps avoid the mess that a new kernel release can introduce. Even using a LTS kernel, (Long Term Support), does not guarantee stability.

---

SCALE almost needs a pre-install tool to verify hardware will be suitable. Need output wording to not imply that the hardware will work perfectly. Perhaps with these results:

• Specific device is known to work
• Specific device is not ideal, but can work, (RealTek NICs; Non-ECC RAM; Really old CPUs; etc...)
• Specific device is known to be problematic and not recommended, (SMR disks; USB attached data disks; USB attached Network: etc...)
• Specific device is known not to work. Or if it does appear to work, can cause data loss, (like HW RAID controllers)

The screwy part of the above, is that iX has little incentive to write such a tool. In the Enterprise space, they sell the hardware, so they already know about compatibility and suitability of the hardware. For the free users, their is such a wide range of hardware, and new hardware coming out all the time, it would be difficult to keep such a "compatibility database" up to date.

Further, to complicate the issue, are categories of USERS:

• Those that won't bother with the pre-install check. Or it ran but they did not bother to read the output.
• Then the 2nd type: Those who absolutely think they know better and TrueNAS MUST support their hardware because that is what they have. Or desperately want to use.

Sun Microsystems used to have such a tool & list for Solaris x86 when Solaris was free to download. I know VMWare has such a compatibility list and I am sure some other software vendors have such too.

Every time we get a SCALE user who thinks their hardware is special and must work, I get a little sad. Some of them attempt emotional blackmail by saying "If TrueNAS can't support my hardware, it sucks!". Except today, I see it as confirmation that not all hardware is suitable for all purposes.

---

Sorry for the long rant. It's been years in coming.

Linux is good, even great at times. But, it is absolutely NOT a magic bullet. Both AIX & Solaris have features that the Linux people say they don't need. Except in an actual data center, as a Unix SysAdmin, I WANT THOSE FEATURES, on Linux. I mean it is possible that Linux SysAdmins don't know such features exist, thus don't expect them on Linux.

 

[Volts]

When I read something like that, it makes me nervous about Core's future, and hence my future as a TrueNAS user.

Yup.

The blog post a few days ago was about how 13.0-U6.1 was the highest-quality TrueNAS release ever.

But now CORE is admitted to hospice care. Oh well. Glad it went out on a high note!

I could not in good conscience recommend CORE to any new users at this time, knowing that it is a dead end. There isn't even a stated product support lifecycle date we can point to.

I also can't yet recommend SCALE to any serious or enterprise users. It's very promising, but I will never again recommend products based on future promises or expectations. (You can't even buy a Mini X+ with SCALE today!)

I know it's not possible for iX to have feet in both the CORE and SCALE canoes at the same time. The change in messaging and lack of clarity in documentation is my biggest disappointment.

 

[Arwen]

Weirdly enough, this implies that TrueNAS Enterprise is now based on Linux:

Compare TrueNAS Editions - Powerful Storage Platform

TrueNAS is the ultimate Open Source storage platform. Choose from CORE, Enterprise, or SCALE to get the best performance and reliability.

www.truenas.com

Their are plenty of references that imply Linux:

• Apps - Yes (K8s or Docker)
• Virtualization - Citrix, Veeam, vCenter Plugin, Built-in KVM / Kubernetes
• Multi-Channel SMB - Yes (23.10)
• Built-In RAID Protection - Mirrors, RAIDZ1/Z2/Z3, dRAID (23.10)

But, to be fair, they do list:

Base OS - FreeBSD (13.0) or Linux (23.10)

 

[Volts]

Both 13.0-U5.3 and 23.10.0.1 are listed in the TrueNAS Enterprise column here https://www.truenas.com/software-status/

I think of "TrueNAS Enterprise" as the umbrella label for the configurations they're willing to sell support for.

 

[Davvo]

I guess it's time to start organizing for a community-mantained CORE. Sad thing is, I'm not a programmer. I can only contribuite with my good intentions and time.
I'm not really against SCALE, but it needs some serious work on levels that I don't know if iX can or is willing to do. I'm aggravated now

 

[Volts]

ZFS 2.2.2 is now in nightly.

Code:

root@truenas[~]# uname -a
FreeBSD truenas.local 13.2-RELEASE-p6 FreeBSD 13.2-RELEASE-p6 truenas/13.1-stable-75ff390ee31 TRUENAS amd64
root@truenas[~]# zfs -V
zfs-2.2.2-1
zfs-kmod-v2023121400-zfs_b88422e3f