As I am thinking about some changes to my Rockstor Server, I am also curiouse about the plans of Rockstor in moving beyond openSUSE Leap 15.6 (which is only maintained till April 2026).
Is there already a discussion (or decission), what the next “recommended” base system will be?
Simply Leap 16.0
Is the new openSUSE MicroOS distro on the table for discussion?
Or maybe remove the whole OS version dependency by using slowroll?
As I could not find a discussion neither in the forums nor on github, and as the end of Leap 15.6 is approaching, I am just curiouse about the plans of Rockstor moving forward.
These are questions we constantly ask ourselves but there’s unfortunately no quick answer. I’ll thus take the time to reply properly a bit later today as I’m currently a bit short on time at this very moment, I’m afraid.
Depending on how slow the roll is in the rolling release, may affect stability of the Rockstor software as a NAS appliance. If the rolling release is too fast it will definitely affect stability of the Rockstor NAS software. Using rolling releases can also introduce breakage even more so than stable, which would require more testing before each release build.
First off, my sincere apologies for such a delay in getting back to you, @simon-77.
This is an important topic (thank you so much for your interest in it and for taking the time to bring it up) and as I mentioned in my previous message, we are constantly asking ourselves these questions. The short answer to your questions could be “yes” but that would be overly simplistic and hide a lot of the reality. It actually does follow a simple logic, but the road is a bit more complex, of course. I’ll thus try to share my current train of thoughts on that logic in an attempt to illustrate that point thoughts on each of the different options you’ve listed.
Before I do that, though, and as you prefaced it yourself, this is all in the context of the upcoming end-of-life of Leap 15.6. While this was originally scheduled to be by the end of this year if I remember correctly, it has been pushed by a few months and Leap 15.6 will be “maintained until at least April 2026” (Lifetime - openSUSE Wiki).
Now that we have a timeframe set, let’s briefly discuss the options you listed as these are indeed exactly those that need to be considered:
This is indeed the most natural next step for us given Leap is our current main target. Leap 16, however, does come with some requirements that explain why we have not yet released an rpm for it. @phillxnet recently detailed this well in a Github issue (Missing build dependencies #101 by phillxnet · Pull Request #104 · rockstor/rockstor-rpmbuild · GitHub) but the gist of it is as follows: the only python version available on Leap 16 is 3.13, which means Rockstor needs to be running on Python 3.13, which requires a Django upgrade, which itself requires us to upgrade our dependencies. We started that work but that’s a substantial upgrade throughout the project (rockstor-core, but also our rpm build, ISO build, our infrastructure); this and its related issues are part of milestones: GitHub · Where software is built and GitHub · Where software is built.
Of note, there is no overlapping Python version between Leap 15.6 and Leap 16, which further complicates things as we now need to maintain two pipelines, one for 15.6 and a different one for 16.
This is all good, though, as all these upgrades were planned and we’ve already accomplished many of them; it just explains why we don’t yet have a Leap 16 release.
There is also an important point about Leap 16, one that is unrelated to Rockstor: the openSUSE release summary explains it well:
The hardware requirements have changed. Leap 16 now requires x86-64-v2 as a minimum CPU architecture level, which generally means CPUs bought in 2008 or later. Users with older hardware can migrate to Slowroll or Tumbleweed.
It’s unclear how many Rockstor users will be impacted by it; maybe not many but it is something to keep in mind.
That brings us to the next target you mentioned:
We’ve had the exact same thought and it is something I’ve been personally excited about. On paper, it seems to match exactly what we would need and that’s why we’ve included it as one of our targets (although for advanced users for now: Downloads). It is, unfortunately, not a universal solution. We would lose aarch64 compatibility if we were to focus exclusively on Slowroll as no arm builds/repositories are available at the moment and it is unclear when/if this would happen. There are also a few other points of concerns and uncertainty that do tone down my excitement about it for the moment.
Last but not least:
Yes, this is something that has been in the back of our head (MicroOS or Leap micro). It has been brought up a little while ago and @phillxnet did detail the points we needed to consider at the time: Base on openSUSE MicroOS · Issue #2217 · rockstor/rockstor-core · GitHub. It definitely is an interesting option but while we’re closer to being able to actually consider it, we are not quite there yet, I think.
We are a small team and as you can see from the above, there already are a lot of targets and architectures that we do need to support and provide releases so we unfortunately need to remain realistic in our choices. We’re making substantial improvements in our infrastructure and much progress to ease the workload required for each rpm release and target but each still requires a lot of work so we’re trying to be careful and not stretch ourselves too thin.
The openSUSE project is moving fast and making a lot of exciting releases. It does come with its own “pain points” at times as there’s a lot to do, but it’s great to have all these options at our disposal too!
I probably rambled a bit too much (and sorry again for taking such a long time to get back to you), but I hope I was able to provide some elements of answer.
Thanky you @Flox for taking your time writing this insightfull post and taking us behind the scenes.
After reading about all the hurdels regarding the different python versions and supported CPU architectures, I can definitely imagine that the choise of the right openSUSE OS is not as straight forward as it might seem in the first place.
Of course I do understand the relevance of stable release distributions for servers, though I think that openSUSE did release two ingeniouse distros with MicroOS and Slowroll, that sound really promising for servers too.
About 2 years ago I set up another server with openSUSE MicroOS and I was amazed how easy the installation and setup procedure was and how stable the daily automatic updates did go, including the possibility to roll back to a previouse version. I would actually be very keen on trying that distro on my rockstor server tpp, once it does get compatible.
Regarding Slowroll, I assume that it might be a more balance middle ground as server OS, though I would think of Slowroll more as a Leap with faster release cycles than a slower Tumbleweed . I would actually love to see the future of Rockstor beeing built on Slowroll, though I was not aware of the limitations regarding available CPU architectures, which is very unfortunate.
. I would actually love to see the future of Rockstor beeing built on Slowroll, though I was not aware of the limitations regarding available CPU architectures, which is very unfortunate.