I use Fedora Kinoite daily and find it to be the only OS to make sense really.
I find Fedora CoreOS totally confusing (with that ignition file, no anaconda, no user password by default, like how would I set this up anywhere I dont have filesystem access to?)
But there are alternatives. I would like to build my own hardened Fedora server image that can be deployed anywhere (i.e. any PC to turn into a secure and easy out-of-the-box server).
As modern server often uses containers anyways, I think an atomic server only makes sense, as damn Debian is just a pain to use.
Experiences, recommendations?
What makes Debian a pain to use on servers?
Automatic updates are overcomplex and not even preinstalled. Install a package, change some configs, so some more.
I dealt with it and its annoying.
And there is a lot more that is completely manual with no good default presets
cron
run as superuser script.sh
apt get updates
apt get upgrade -y
??? profit?Why is there
apt-get
andapt
? Also on regular updates there are sometimes package conflicts that need manual configuration. Maybe-y
deals with some.What monstrosity are you running and calling Debian that there are package conflicts on regular updates?
…or, are you talking early-2000’s Linux, where SuSe was the only consistent distro and package management hadn’t really been fully sorted out?
Probably I got none, just this “do you want to use the maintainers version” which is always a bit confusing. VirtualBox also gave issues but just dont use that crap.
unattended-upgrades is annoying? How so?
Its overcomplex. For sure I could get used to it and maybe this is the way to go.
But you could wrap this tedious process in a function.
Fedora has a distro upgrade command (that totally sucks but okay) since many years, while on Debian I needed to follow some random Guide to get on the hyped Debian 12.
Debian releases a migration guide with every new version release. And sorry but if you have trouble updating your system then replacing the source.list file and then updating your system again, you should reconsider running a server yourself, imho.
I was looking for such a guide but could not find it back then.
Which may be overcomplex but it is complete and lots of things where not intuitive at all.
As I said, you could easily automate this step, instead of making it that manual. Or course I can do that, but why need to, if a
sudo apt distro-upgrade
would do it?https://wiki.debian.org/DebianUpgrade
Because those steps need manual review. Things change, packages get removed, packages get upgraded, config files need to get manual reviewed and merged etc.
On a simple System without much configuration that stuff does not matter, but when you use different package repositories and backports you need to be careful. I am not sure how introducing a new command does solve those complex issues. Imo only the system admin can decide what the best steps are.
Thanks! Will look into that
If you’re on Debian, it’s the tried and true method. The config is dead simple for most upgrades, just un-comment the line in the config file next to the type of upgrades you want, stable or testing. It can take some debugging if you have a package with it’s own APT repo. It’ll just ignore those updates by default.
NixOS works really well as an image based server. Use nixos-generate to create a pre-configured image and put it on a flash drive/PXE share, and you’re good to go. Automatic updates are a bit confusing and not really documented, but doable. I have code examples.
I would be interested in automatic updates on NixOS!
This wiki page has an example on how to do automatic updates on a normal install of nixos:
https://nixos.wiki/wiki/Automatic_system_upgrades
But this won’t work for nixos-generate because nixos-generate doesn’t have a configuration.nix file in the booted system.
Here is the code I use for my nixos-generate flake that I use to generate all of the nixos images in my homelab:
The way this works is that it includes the flake source code as a folder in the nix store on the booted system, and the nixos-upgrade timer will then use the flake to build an updated version of itself. Note that nixos-generate uses the
packages
output of the flake, while nixos-upgrade uses thenixosConfigurations
output of the flake. I have written the flake so that they build identical systems, but it means there’s some code that I had to write twice in flake.nix.Feel free to try it out yourself, though note that you will probably have to rip out the agenix stuff to get it to build.
Nixos isnt really that user friendly yet, but insanely powerful once you understand how it works. Feel free to ask questions if anything seems confusing.
You can use Fedora IoT which is essentially rpm-ostree based Fedora Server. It would be less confusing if it was just named Fedora Atomic Server.
This new distribution of Fedora is FAT!
Omg yes thats true. Thanks!
But CoreOS is also using rpm-ostree, how are they different?
I didn’t try CoreOS as I didn’t even get how to set it up. As I understand it, it uses a completely different workflow for administering the system compared to regular distros.
Yep, and thats all cloud-first I suppose. It sounds cool but you need to create an ignition file (which sounds very possible) but then you need to get that to a server that doesnt yet have a user account.
I dont understand anything of that. I dont think mounting a drive with that file is possible everywhere, and how do you setup LUKS?
Just no. I see if IOT is actually atomic but normal.
Like, just use a cli installer that can load a file to automate it. Or have a backup user password. There is an issue that addressed this, its old and closed, yeah.
the only OS to make sense really
How does it make more sense than Fedora KDE?
I want a server haha.
And yes, atomic ftw.
Because containers (Distrobox, Flatpak, etc.) are bae.
You can read my post I made a while ago for more information: https://feddit.de/post/8234416Once you “get” image based distros, you probably never want to go back. Traditional distros just feel… off now for me.
Containerisation is the biggest strength in Linux, we use it all the time on servers, so why not on the desktop?
Atomic OSs just make more sense for me, not only because of security/ bug/ whatever reasons, no, also because they feel simpler and are pretty convenient and robust.Interesting. I didn’t realize it was structured this way. I thought it was similar to NixOS for some reason.
@Pantherina i use fedora kinoite and yeah it’s really awesome! new packages and a safe system.
I wanted to use Fedora CoreOS on my server but no providers offer it so I ended up installing AlmaLinux instead. But yeah the ignition file setup is really painful. tried in a vm but never managed to get everything i wanted. i’m gonna stick with enterprise linux until they make it easier, i think
I tried IOT too and it the bootloader didnt install.
Then I just installed Atomic Sway (because not that much bloat), and before logging in rebased to secureblue server-main-userns-hardened. It worked but I have no DNS? Damn…
@Pantherina have you checked if systemd-resolved is working properly and that systemd-networkd or networkmanager is used? only one of them shall be used. i had a similar issue when upgrading from 38 to 39 because then both were active. i’m using NetworkManager on my desktop and disabled systemd-networkd and then it worked…
Strange, Fedora39 to Fedora39, I use that atomic base always (like 15 different installs, GNOME, Plasma6, Secureblue, Cosmic, Sway,…)
@Pantherina
I see. At least systemd-resolved needs to be running and /etc/resolv.conf needs to be 127.0.0.1I rebooted and now it works. /etc/resolv.conf is not the file you edit, but that localhost DNS is interesting. Saw that a long time ago (Obi wan face)
@Pantherina
Awesome! Great to hear that it works :)
@selfhosted
I made a similar post a few weeks ago.
I will try uBlue core and give you all a small update about it.I feel similar about Debian. It’s a good distro for sure and I don’t have any issues with it for server use, but somehow, I still don’t like it somehow. RPM-/ OSTree based distros are more my taste, and I don’t even know why.
I am completely confused about ublue currently, (okay all they did is remove the image list, its the same on Github)
Debian is old and crusty with all its tooling. Apt sucks, automatic updates are strange, there are no snapshots afaik, it uses ext4, its like Fedora was 10 years ago
People who use Debian servers typically just install Docker on a basic system and then use containers. Which is exactly the same concept that you describe.
What’s the filesystem of the server got to do with anything? You can take snapshots in half a dozen different ways, everybody uses the method they’re comfortable with.
A bunch of edgelord babies skimmed the selfhosted subs and noticed that “atomic distros” is a common buzzword ATM; they then flood said subs with opinionated posts that atomic subs are the best and everything else sucks 🙄