Workstation, server, IoT? No worries. Fedora 36 is out – all 13 editions of it Version 36 of Fedora, the free community Linux distro sponsored by Red Hat, is here. And let it be known that there's a lot more to Fedora 36 than just the well-known distribution, with its different desktop spins. There are multiple parallel products under the Fedora banner, and they've all got new versions out, too. The …
-> There are multiple parallel products under the Fedora banner, and they've all got new versions out, too.
Multiple ways to confuse people. Fedora has its own branch of YALD.
-> Silverblue is a desktop distro
s'/is a/is yet another/'
-> The idea is that if an update goes wrong and something stops working, then you can just reboot and revert back to an older snapshot in which everything worked.
As seen in Solaris and beadm.
-> Fedora has no snapshot support at all by default. You can enable it, but it's not trivial
Yeah. Take something that is actually *useful* to have and make it not trivial to use.
-> Without it, there isn't really any significant benefit to using Btrfs
Well it makes it different. It means that some people can say "We use Btrfs".
-> Btrfs can be fragile
Your'e telling me. Let's just call it unreliable and advise people should not use it if they value their data. That is the only trustworthy view to take.
-> Red Hat has its own next-gen filesystem effort, Stratis
What the Linux world need is file systems, file systems sweet file systems. How about just trying to fix btrfs? No, let's have yet another file system.
"Multiple ways to confuse people. Fedora has its own branch of YALD."
We (I work for RH on Fedora, as the Fedora QA team lead) do try hard to avoid this, which is why if you go to the download page - https://getfedora.org/ - you are directed prominently to just the three core Editions, for very distinct purposes: Workstation, Server, and IoT. Other flavors of Fedora are intentionally de-emphasized a bit and shown lower down.
On btrfs, we do actually get some benefit from it; since Fedora 34, zstd-based transparent compression has been enabled by default, which usually saves quite a lot of space. The folks pushing the incorporation of btrfs in Fedora also have grander plans down the line, we're just moving carefully on it to ensure folks don't get bitten by the vaunted "btrfs stability problems".
Stratis is not actually a next-gen filesystem effort, that's slightly inaccurate (though to be fair RH docs do refer to "Stratis filesystems", which makes this a bit unclear). Stratis is actually a management layer on top of a bunch of pre-existing technology - LVM, device-mapper, and the xfs filesystem. It aims to provide next-gen filesystem-like functionality by marshaling all those technologies together. See https://stratis-storage.github.io/ for more on this.
I think you must've misunderstood something - I don't think we ever had May 24th down as a release date for 36. It was actually delayed a couple of times, the original "well...we *might* hit it..." early target date was 2022-04-19 , and the more serious "we really *want* to hit it" date was 2022-04-26. We missed that by two weeks and released on 2022-05-10 instead. See https://fedorapeople.org/groups/schedule/f-36/f-36-key-tasks.html .
Glad you're enjoying it, though!
"Stratis is not actually a next-gen filesystem effort, that's slightly inaccurate (though to be fair RH docs do refer to "Stratis filesystems", which makes this a bit unclear). Stratis is actually a management layer on top of a bunch of pre-existing technology - LVM, device-mapper, and the xfs filesystem. It aims to provide next-gen filesystem-like functionality by marshaling all those technologies together. See https://stratis-storage.github.io/ for more on this."
Strange - btfrs was meant to do all that and more? I mean does it have encryption yet. Is this an admission that butterfs has failed, or has IBM, RH's corporate daddy decided that it doesn't want to play with something it cannot control/own?
Yet again, it has nothing to do with IBM. I dunno why people are so keen to believe IBM, having bought an extremely large and successful business, would immediately start interfering with every aspect of how it works, but it really isn't doing that. IBM hasn't said squat about what filesystem it wants RH to use. I doubt IBM cares very much, so long as the money keeps coming in.
I obviously can't say too much about internal matters, but, AIUI it boils down to this: RH has a storage team that makes these kinds of decisions, and after quite a lot of research and discussion, they made the call to pursue the Stratis strategy rather than going with btrfs. I'm not part of that team or much of an expert on filesystems, so I can't really judge whether that was the right call or not for myself, but that's the call they made. AFAIK it was purely a call made on their interpretation of the technical merits.
It'd be odd to frame it as an "admission" that btrfs "has failed", since RH doesn't control btrfs' development or anything. To RH it's a technology we could choose to use or not use, like hundreds of other F/OSS technologies. And of course decisions aren't set in stone; of course in future RH might decide to change course, like we have in the past on other similar technologies.
-> Strange - btfrs was meant to do all that and more? I mean does it have encryption yet. Is this an admission that butterfs has failed,
I think it's fair to say that btrfs is not yet finished. If you do a search for 'btrfs reliability' you will find very mixed responses. You will see "highly experimental", "I don't know", "expected to offer better scalability and reliability", "reliable and stable", "corruption issue", and the like.
Meanwhile in comparison for zfs you will that find that it is reliable though it uses a lot of memory (an oft-seen rule of thumb is 1GB of RAM per 1TB of disk, though more is still better).
I would be happy to store any amount of important data on zfs, but not on btrfs. I don't want to go through the pain of finding out that the problem IS btrfs.
'we're just moving carefully on it to ensure folks don't get bitten by the vaunted "btrfs stability problems".'
I don't think that an "enterprise" Linux distribution will be able to solely base itself on btrfs.
Oracle specifically prohibits database installations on this filesystem in 2290489.1: "Oracle DB has specifically said that they do not support using BTRFS filesystems... BTRFS is optimized for *non-database* workloads."
XFS also dominates the TPC benchmarks (at tpc.org) for Linux. There is no way that btrfs is going to supplant that status within the next few years. SGI definitely contributed an extremely valuable asset.
I'm also a bit surprised that Fedora's boot is ext4, as it's XFS under CentOS (7).
-> The idea is that if an update goes wrong and something stops working, then you can just reboot and revert back to an older snapshot in which everything worked.
-->As seen in Solaris and beadm.
beadm and ZFS = fantastic.
Or for time being, duplicate Solaris 10 Live Upgrade ability (non-ZFS filesystems). Gosh I loved that ability. Wish the Linux camps would either do a ZFS shoe-in compatible (done legally somehow) or provide a Live Upgrade ability to flip-flop between boot environments.
Well, we definitely do not advise anyone to run it in production, and stuff can still go wrong. But in truth we (and I personally - as noted above, I work for RH on Fedora as the QA team lead) have been working quite hard for a few years to make Rawhide much less scary than it used to be, and most of the time it is. I run it on my personal system sometimes (my policy is to run "the next version", so after a stable release I upgrade to Rawhide, then when the next release branches off from Rawhide I follow that branch until release, then go back to Rawhide) and it's fine most of the time. We have quite comprehensive automated testing of each Rawhide compose, which you can refer to to see if anything blew up in the latest compose, and avoid upgrading if it did - see https://openqa.fedoraproject.org/tests/overview?distri=fedora&version=Rawhide&build=Fedora-Rawhide-20220512.n.0&groupid=1 for today's test results for instance.
My latest effort is to enable automated testing not just for the daily *composes* of Rawhide, but for individual Rawhide updates, as we already do for stable and branched release updates. Ultimately I'd like to have those updates gated on the testing, again as they are for stable and branched releases, so updates that break core functionality would never make it to a compose. For now the Rawhide update tests are running in the test instance of the QA system for the last couple of weeks, and have already helped us identify the package responsible for one critical bug - https://bugzilla.redhat.com/show_bug.cgi?id=2082359 - and find and fix a subtle bug in a recent systemd release candidate - https://bugzilla.redhat.com/show_bug.cgi?id=2083374 . Once we get to the stage of gating Rawhide updates on these tests, those issues would never make it to a Rawhide user.
So, Rawhide is still officially scary and you really shouldn't use it on a critical system, but in practice it's already less explode-y than it was a few years ago, and we're actively working to make it ever less explode-y over time!
Sadly, VoT seems fixated on the idea that 'choice' is intrinsically a problem, totally failing to recognise that the price of 'freedom' is 'eternal vigilance', i.e. needing to take responsibility for one's own choices, and learning to manage the consequences of other people's choices.
I actually have to agree. A lot of software often makes an explosive bolt out of the gate and then slowly rots to death as interest wanes.
I'm all for people playing with a distro and having a stab at getting one going but you only have to speak with teams who run successful distros to realise how much hard work they have to put in to actually making one and more importantly promoting it to get people spend more than 5 mins trying it out. VM/HyperVisors make it so easy to whack in an ISO and just as quick to spot one small problem and just drop the VM and ISO with it.
More distros are a Good Thing
I've tried dipping my toe in the waters of Linux a few times but I genuinely get confused by the proliferation of distros. Having distinctions between workstation, server, etc. makes sense but I've never understood why having more distros is such a good thing.
The Linux community wants widespread/mainstream adoption of Linux as a desktop OS - in use by general Tom, Dick & Henrietta users for whom a computer is just a means of doing their daily job, hobbying, etc - they don't care about the OS. That goal has not come about yet. Has anyone considered that if the community put all of that effort into making a distro which is as good as it possibly can be, rather than effectively duplicating effort across multiple distros, the less tech-savvy people might embrace LInux?
The non-techy person looking for Linux puts me in mind of the Not The Nine o'Clock news sketch about the guy in a hifi shop - look for it on YouTube - can't post a link here as I'm at work and the Fun Police forbid such things.
It like..."I'd like Linux please"
"Certainly, which one? There's Ubuntu, Fedora, Debian, Centos, Mandrake..."
"...errr...."
"...or there's AlmaLinux, MiraLinix, Oracle Linux, Red Flag Linux, Ricky Linux..."
"...errr..."
"...or there's Berry Linux, BLAG Linux, Fuduntu, Korara..."
"You know what? I think I'll just stick with Windows"
Linux (as a desktop or server) is for people who like to have a choice. Not some BigCorp telling them what the next OS version is like. Don't like the direction Fedora is going? Use some other distro. Actually even this is not usually necessary, because you can customise Fedora to you liking. I mostly use Fedora, but hate the Gnome desktop, so I use XFCE (conveniently, there is a ready Fedora "spin" with this, but even if there were not, I could install XFCE myself).
Contrast this with the moaning one heard from Windows users whenever MS makes some change existing users do not like. Sometimes the moaning causes MS to change its mind, but usually not, and you just have to swallow the new version. (And there is no alternate Windows "spin").
As to the huge number of distros, most of them do not matter for a beginner. They tend to be derivatives of the three or four big ones. (Usually born of someones dislike of how the original distro evolved! Or a pet customization of it). I would tell a beginner to pick well-known one with a good user community, learn to use it, and then, after maybe a year, perhaps sample others, if interested in alternatives.
"The Linux community wants widespread/mainstream adoption of Linux as a desktop OS - in use by general Tom, Dick & Henrietta users for whom a computer is just a means of doing their daily job, hobbying, etc - they don't care about the OS."
I'd say your premise there is a bit shaky. Especially these days. The desktop wars were kind of a 90s/00s thing. It feels a bit weird these days when desktop/laptop computers are less of a big deal overall anyway.
I suspect quite a large chunk of "the Linux community" doesn't actually give much of a toss what anybody else runs on their computers, they just like Linux. I don't really care what Tom, Dick or Henrietta is doing any more.
Things are similar for the big Linux companies. None of them make money from desktop users. All of them tried to before, and found it doesn't work. The other commercial Linux companies of the past also tried it, and went broke. The money turns out to be in the enterprise, and not the enterprise desktop. RH, Canonical and SUSE don't really care what Joe Public runs on their desktop or laptop computers, if they even have one any more.
"Has anyone considered that if the community put all of that effort into making a distro which is as good as it possibly can be, rather than effectively duplicating effort across multiple distros, the less tech-savvy people might embrace LInux?"
Only about ten zillion internet commenters. But, again, there are some problems here. "Too many cooks spoil the broth" is a real thing.
In theory, you certainly could make one bigger better distro by combining all the work that goes into several different ones. But there are some devils in the details. The work that would really help out the bigger distros is probably not the work that people who maintain smaller distros want to do. It's often kind of thankless behind-the-scenes stuff like rote testing of updates or maintaining lesser-used packages or updating documentation or doing user support. You're not necessarily going to get anywhere by saying to people working on Mint "hey, how about you stop doing the fun stuff on Mint and start updating documentation for Ubuntu instead?", and so on.
Basically, people aren't just pawns to be moved around a chess board at will, especially when a lot of them aren't getting paid.
At the level of people getting paid, the major players paying people to work on Linux - Red Hat, Canonical, SUSE - are essentially competitors and have zero motivation to combine into a super-distro, or make a serious move on the consumer desktop space. It's just not going to happen.
[Author here]
There shouldn't be just one distro, because there isn't just one user.
Choice is good. Diversity is good. I want to see more diversity, not less. (Indeed I have a forthcoming piece on this exact theme.)
What do you want from it?
To just get online with minimal fuss? ChromeOS Flex.
A simple, vaguely Windows-like, general-purpose OS that as much as possible Just Works? Linux Mint.
Something simple, fairly clean, mainstream, solid, that you can run for the typical lifetime of your PC and won't need to be reinstalled in 4-5 years? Ubuntu.
The latest version of everything? OK, got some skills or want to learn: Arch. No, just want it to work: openSUSE Tumbleweed.
I don't mind learning a bit of my own essential maintenance, but I want to be solid and reliable over the long haul? Debian.
I work in business, my employers pay for their boxes and I maintain them and want to know what's coming? Desktop: Fedora. Server: Arch or Rocky.
I want it light, clean and independent of the big names? Void.
I work with containers and want to run the same OS I test and develop on? Alpine.
I paid top dollar for this box, I overclock it, and I want the ultimate performance and to know my OS as well as I know the hardware I hand-built? Gentoo.
I could go on.
Almost every distro out there has a use case.
And occasionally experimental distros come up with amazing new features that everyone copies, and that's great. Occasionally companies commit to bleeding-edge tech and deliver something amazing that everyone should adopt.
If you have a use case, and you know what you want, then there is something for you. If you don't, the question becomes:
Do you want to learn new stuff and maybe do a tiny bit of maintenance? Ubuntu or Debian.
No, just dead easy and general-purpose please? Mint.
Do you just want a laptop that works like a tablet? ChromeOS Flex, or better still, buy a ChromeBook.
It is not as hard as you are making out.
And yes, this summary does reflect my personal biases a bit. Everyone has those.
While I accept some of what you say, what about this:
-> Do you want to learn new stuff and maybe do a tiny bit of maintenance? Ubuntu or Debian.
Why should I have to learn the plethora of different package managers if I run different distros? What is the possible benefit to me? It is this sort of thing in the Linux world which is both stupid and unnecessary. Diversity here is not good. It is nothing but a waste of time and effort, both for the end users and for the developers who create it in the first place.
Big surprise, buzzah: It's not about you.
Not at all.
You want to merge Linux distributions project for no better reason than your own convenience?
Why should a package maintainer of the Debian community have to deal with the corporate kerfuffle at Red Hat or Suse? Why should an Arch supporter have to deal with the basic democratic shenanigans of the Debian community? And why on earth should someone of the Nix community be coerced to try to make their rather radical approach palatable to any members of any of the other projects?
Different distribution projects exist for different reasons. You're making it explicitly clear that you can't bring yourself to care about these reasons, raising the question why anyone should bring oneself to care about your opinion.
I keep hearing and seeing comments about Linux and Fedora which are COMPLETELY at variance with my own experience. So....I thought I'd add my ten cents.
(1) Linux is hard to use
Well....not in my experience. I like RH/Fedora.....but Linux Mint or elementary OS seem EVEN EASIER!
(2) Fedora is "bleeding edge"
Well.....not in my experience. Pretty easy actually. Now if you REALLY want "bleeding edge".....there's always Slackware!!
(3) Gnome is "polished" and/or "elegant"
Gnome is cr*p. Version 3 was cr*p. Version 4 is cr*p" They should have stuck with version 2!!
(4) Users say they can't get along with Gnome...."What can I do?"
Well....see item (3) above. The Fedora XFCE Spin might be an answer. Or Linux Mint. Or elementary OS.....lots of choice out there!
(5) Quote from article: "Fedora defaults to formatting the hard disk with Btrfs ... but it's worth noting that this doesn't confer much benefit yet."
Well......"No sh*t, Sherlock". Don't even go there.....there's NOTHING wrong with ext4. PLEASE use ext4!!
That's it for now.....hope someone out there has found this to supply serious corrections to common misinformation!!
[Author here]
If you want ext4, you can have ext4, and you won't lose anything important. You are free to do a custom install if you like.
For what it's worth, I tried MATE on F36 and XFCE before that. If you want a simple clean Windows-like desktop with no mess and no fuss, you can 100% have it.
Me personally, I like a vertical taskbar, and Xfce does that better than anyone. It is, incidentally, totally broken in LXQt now.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
