Me too. But again would be good to have a known EOL date unless I missed that memo?
That would give clarity. And clarity seem to be precisely what they want to avoid.
You do not have to upgrade to the latest release, you can choose a previous release. Or you could choose to only upgrade to .2 releases, which will have many fewer updates to them. You have choices.
There are systems that are not exposed to the Internet, do not need the new features, and are working just fine for the workload they have that just do not upgrade and stay with old releases. This is fine until something goes wrong and you need help and no one can tell if you are hitting a corner case bug in the old release you are running.
The issue with any change to underlying code is not what or how much changed, but that for any change to underlying code the entire system needs to be fully tested, end to end, to ensure that no new corner case bugs were introduced.
2 Likes
As always, of course. But here I will agree with previous posters - future roadmaps, as well as lifecycle schedules of past and current releases, is totally non transparent and in addition seems fluid over time. You very much have to play it by ear and follow these forums closely to understand what’s going on. To illustrate my point, Internet is full of product pages with good examples of what clarity could look like - e.g. RedHat: Red Hat Enterprise Linux Life Cycle | Red Hat Customer Portal
Indeed - in other words, a bad idea… I don’t see that any well-run enterprise would run a non-supported version for any extended period of time. As for home users, each to their own… But if you really care about that category of users, you would try to make life easier and more predictable for them, not the other way around.
4 Likes
Yes, that is the cadence we use. It allows us to meet customer needs in a timely fashion while retaiin a well organized engineering process.
Many users/customers choose to stay on a particlar release or skip-through releases to the version they want.
Its a fair criticism, but we had tens of thousands requesting per-App IP addressing. What would you do?
No, they could not easily update the Linux kernel more regularly.
Their are several basic reasons:
- Whence updated to a newer Linux kernel, iX more or less can’t go back. That means even more kernel updates if a new reliability or security problem is found.
- Newly introduced drivers more or less become “standard features” that users may depend. And if un-reliable, then yet more kernel updates are needed to fix.
- Besides based on Debian, the kernel is custom built for TrueNAS.
- Last, and critically the most important, OpenZFS is built & tested against a specific kernel. While OpenZFS versions support a range of kernels, (unlike Linux kernel drivers), iX has to choose which OpenZFS version to use. And OpenZFS always needs customization for the latest Linux kernel releases. Might only be a few days, but could be something really annoying like removing floating point access for non-GPL kernel drivers.
In my opinion, the “Linux way of unstable kernel interface” is no longer really viable for the Enterprise. This “Linux need” to have the option of changing the kernel interface every release is unsustainable.
Yes, changing the kernel interface between major releases is reasonable. But, every release? This does not just impact OpenZFS, but other software like nVIDIA’s drivers.
I mean, theoretically this might mean more “major releases” that truly need to change the kernel interface. But, their is a LOT to be said for stable kernel interfaces for Enterprise software.
1 Like
It was a popular request to be sure, but tens of thousands? And while it’s somewhat of an orthogonal question (because you should try to give the users what they want, even if they don’t really need it), I rather wonder what proportion of those requests had any actual need for that feature–I suspect it was quite small.
Design them better in the last release so the feature was already there? Surely you already knew people wanted it, even if your k3s-based apps never allowed it (though TC’s did). And if the issue with that is that you didn’t have time to get that working reliably (enough) prior to 24.10’s release, well, that’s kind of the point of this thread.
1 Like
@dan I can’t argue with you, you make many valid points. I’m sure my perspective is skewed as I really only use TrueNAS as a NAS, not as a hypervisor, nor do I use most of the features many people do use.
It did take me a while to jump from CORE to SCALE, I knew SCALE was new and had some problems so I waited. I still have one CORE and one SCALE system.
My only gripe is that NVMe drives still cannot be scheduled for SMART testing, even though the GUI makes it appear you can, the test never actually runs. It was blamed on smartmontools v7.4 and earlier, and I’m not doubting that was the issue. v7.5 was released a few months ago. I manually installed version 7.5 and the NVMe SMART test still do not work. I suspect the middleware needs some manipulation to launch the test.
I guess that from my use, I’m a happy camper as they say.
2 Likes
I appreciate the 6 month update cadence with the knowledge that I don’t have to update to the latest and shiniest immediately.
The only part where that changes is security updates. If a severe flaw is found I would like to see that being backported at least one, maybe two versions.
3 Likes
There are LTS kernel releases which get minor updates but not breaking changes. Most enterprise distros also just backport patches.
2 Likes
We are generally backporting at least 1 version… especially if that version is used for “mission critical”. That allows the annual update cadence and still have good security.
Reminds me of the Joke:
Why could God create the world in 7 days?
He didn’t have an installed base.
The reality is we have to make many decisions based on keeping Community and customers happy in an impatient world. We don’t give the engineering team free rein to slow releases or ignore pressing issues. The need to migrate existing customers to new versions does put constraints on what is viable. There are compromises made… we don’t deny that.
2 Likes
Many good points have been made.
In my view it comes down to what the “target audience” needs. An enterprise customer has very different requirements compared to a small business compared to a home user.
As to release cadence, I used to work for a reasonably large company that did enterprise software (like for Fortune 20 companies). At one point product management had somehow convinced themselves that customers wanted features sooner. So they switched to a 6-month release cycle.
Guess what: Nobody made use of that. Customers stuck with upgrades every 2-5 years. Simply because the upgrade created an effort of hundreds to thousands of person days. And it is important to remember that such an update in and off itself provides zero business value.
In my view the core value proposition of enterprise software is investment protection. That is why IBM etc. can charge what they charge. These guys close contracts of several hundred million dollars. Because it is still cheaper and less risky for customers than to switch to a competitor (incl. open source).
Coming back to Linux, that is also why Red Hat is so successful. By back-porting fixes and delivering a stable platform. Not by providing new features.
At the end of the day iXsystems needs to decide what market they want to address. Personally I don’t see them anywhere near the enterprise space. And that is completely a leadership issue.
For a long time they claimed that TrueNAS Core was the enterprise platform, only to suddenly drop that line. Perhaps there was non-public communication with critical customers. But from what I have seen iXsystems disqualifies itself for the enterprise market.
8 Likes
I guess we can wonder what the enterprise customers think of iX and their decisions. But I think we wont learn that here on community forums.
I mean, if nobody here knows any CEO or something? 
If the paying customers are ok with how things are done by iX thats what matters.
For me at least TrueNAS is nice alternative to brands like Synology or QNAP. I wonder if these brands are categorized as enterprise market or consumer market, or something in the middle?
At least with TrueNAS you can use your own HDD
1 Like
I think you are reading that a bit wrong… it is clear that some people here are very much speaking from first hand experience from large scale enterprise.
1 Like
That’s nice, so who’s here from any company that pays for Truenas Enterprise systems? Will be nice to hear your experience 
There’s a bunch if case studies on the public web site. On a quick browse none of them look like the scale that @ChrisRJ is talking about and I’d be very surprised if any of them run 25.04 also 
but yeah, isn’t that the gist if this thread… I.e. what is the actual target segment for the product and therefore the most important feature sets and the most efficient delivery method.
When you want to be on the bleeding edge, nothing better than Arch
1 Like