For those of us who might remain on TrueNAS CORE?

[winnielinnie]

If one decides to stick to the TrueNAS CORE releases (FreeBSD), yet pass on SCALE, are the following assumptions correct? I've gleaned different ideas from perusing the Jira tracker:

• TrueNAS CORE will always remain based on FreeBSD
• Each major version of CORE will follow the major version releases of FreeBSD (e.g, 12, 13, 14, 15 ... )
• All the features currently on CORE (12.0-U8) will remain indefinitely for every upgrade of CORE releases [1]
• No new features will ever be added to any future CORE releases [2]
• CORE will continue to follow OpenZFS releases and updates (i.e, CORE users will inherit improvements in ZFS over time)
• CORE will only be maintained strictly to follow FreeBSD's cycles, and to fix bugs and cosmetic issues

Is the above all true? Did I miss anything important to consider?


[1] I'm not sure about this one, since this ticket implies that the developers are contemplating the removal of less frequently used features. Even if few people use certain features, they might have established a workflow, which can break many setups with future upgrades to CORE if they are removed. The feature / code is stable, so why consider removing it? It's only a matter of maintenance, right? (Seriously. Read the description of the ticket linked in this footnote.)

[2] Except perhaps the in-house "CLI" program (which uses middleware calls) will be officially supported in CORE, rather than be an exclusive feature fore SCALE?

 

[Patrick M. Hausen]

Why "might" remain? Why would I run anything else in production? SCALE does not have jails. Case closed.

Second, CORE is the code base for the entire Enterprise product line by iXsystems. I cannot picture them categorically ruling out new features and improvements.

The single question that I can answer with some confidence: of course CORE will follow OpenZFS releases. Simply because FreeBSD does. There won't be a way to run an older OpenZFS version on e.g. FreeBSD 14. And I am not even asking "why?"

Kind regards,
Patrick

 

[winnielinnie]

Why "might" remain? Why would I run anything else in production?

Shots fired!

Second, CORE is the code base for the entire Enterprise product line by iXsystems. I cannot picture them categorically ruling out new features and improvements.

I'll have to rummage through the website, these forums, and the Jira Tracker again, but I do recall it was mentioned that from CORE 12.x and onwards, it would be a "feature-freeze", and any new features would only be introduced in SCALE. In other words, CORE will be "as is" feature-wise, but continue to benefit from maintenance and bug fixes.

Also on your point, I provided an example in the reverse. They're contemplating if features should be removed if the usage stats show that not enough people are using the feature. (This is more concerning to me than an indefinite feature-freeze. For example, I've setup rsync tasks using the Rsync Service (Modules).

Jira Ticket NAS-111949

We need to track rsync mod in usage stats.

We will need it to decide if we will remove that feature or not.

 

[Patrick M. Hausen]

from CORE 12.x and onwards, it would be a "feature-freeze"

Well, CORE does exploit the FreeBSD ecosystem to a great degree, doesn't it? Unless there are substantial new technologies introduced to FreeBSD in terms of virtualisation and containers, I do see CORE as being largely feature complete. If they really refactor that bridged networking, that is

I wasn't aware of this discussion regarding rsync. Thanks.

 

[Kris Moore]

I'll try to provide some clarity here.

New features focus is going to be primarily on SCALE. We may backport things to CORE if they make sense, but we'd like to keep that release stable above all else. The underlying FreeBSD OS doesn't innovate particularly quickly, so apart from keeping up with some of the latest kernel / packages, there won't be much change we have to do there to keep it humming along.

On SCALE however, there is a much wider variety of things we can do long-term and big picture, so we'll expect most effort to spent there. Apps would be a good example. Jails on FreeBSD haven't changed much in decades, however Linux containers have vastly evolved and pushed the technology envelope forward. Unless you are a jails die-hard, running Apps on SCALE I expect will always provide a better experience and support for more types of 3rd party software. I could say the same thing about hardware support and other critical technologies that will help drive the TrueNAS SCALE experience forward, no real surprises on that front.

 

[winnielinnie]

The underlying FreeBSD OS doesn't innovate particularly quickly, so apart from keeping up with some of the latest kernel / packages, there won't be much change we have to do there to keep it humming along.

When I refer to "features", I mean more along the line of the middleware / appliance (not the operating system or ZFS code).


Here is an example of what I mean. You'll notice I filed this feature request in May 2020:

https://jira.ixsystems.com/browse/NAS-106300

It was later accepted in September 2021:

https://jira.ixsystems.com/browse/NAS-112452

But from what I can tell, it's destined for SCALE, while CORE users will likely be passed over. This is what my bullet-point #2 was asking.

• No new features will ever be added to any future CORE releases

This type of feature has nothing to do with the base OS, nor the ZFS code, nor even ZFS features. It's specifically to expose an already-existing ZFS feature in the TrueNAS GUI.

So my question was sort of implying that for those of us who choose to stick with CORE, it's possible we won't inherit the same beneficial features as SCALE users? (Which also implies there will be a "feature-parity mismatch" between CORE and SCALE that goes beyond the obvious jails vs docker containers.)

 

[Kris Moore]

At the moment we plan on being much more restrictive on CORE features, including MW & UI Changes. Focus will be on stability, not introducing new code which can/will lead to regressions. We may selectively bring features back into CORE, but *only* after they have been through some SCALE release cycles and we believe the benefit is worth the effort to potential regressions.

Still too early to say which features those will be. With TrueNAS 13.0 we have some pretty pressing reasons to get that out the door quickly (including Samba CVE's that can't be backported to 12), so we've made the decision to make the 13.0 cycle pretty short and minimal with regard to consideration of any new features.

 

[winnielinnie]

Thank you for the clarifications, @Kris Moore.

I'm looking forwards to TrueNAS CORE 13 as a "brought into the present" version of TrueNAS, especially considering NFS will be the dearly-needed version 4.2.

At the moment we plan on being much more restrictive on CORE features, including MW & UI Changes. Focus will be on stability, not introducing new code which can/will lead to regressions.

While we're on this subject, can you speak to the concern that CORE might not be spared from feature removal?

Because if it is "on the table" to remove the Rsync Service / Modules, then it goes against the strategy of being restrictive in regards to middleware and GUI changes. (Not to mention breaking current setups and workflows.)