As long as there is software, you cannot shortcut the need for maintenance. Nothing is bug free, and telling people they will never need to upgrade/patch/maintain a system is a well-paved path to compromised systems.
This OS doesn't says it's maintenance-free! But it skips a whole load of maintenance you'd need to think about with a traditional base system, because 1) there's almost nothing there, and 2) the upgrade to that base is easy, you just reboot and restart your containers.
Obviously the software you run needs upgrades, but (again, but a layer down) it's based on Docker and probably someone else is maintaining it. So you pull that new container, restart and the OS is just making sure your data lands in the same place with the new container.
If you're happy with all your software running from Docker this seems like a step up from a Debian or Redhat, and it has a lot less bureaucracy than something like CoreOS.
Whether it's _usable_ I'm not sure (especially around storage management) but it's a really clear pitch.
> Nothing is bug free, and telling people they will never need to upgrade/patch/maintain a system is a well-paved path to compromised systems.
Of course nothing is. But there's a reason projects like "Talos" do exist: no terminal, no SSH, no package manager (how do we like package managers like NPM lately btw?), read-only filesystem, definitely no systemd, etc.
And then a minimal number of executables.
This does, definitely, reduce the attack surface.
I'm not speaking about this Show HN's project but there are such things as systems both more secure and requiring less maintenance than others.
Throwing in the towel and saying: "nothing can ever be 100% secure so we'll always need to patch so we may as well YOLO by accepting npm packages modified 3 minutes ago" is not the way to go forward either.
IncusOS is another - read-only root FS, interactions with the system exclusively through the Incus API, no package manager, blue-green OS updates (à la Steam Deck / Home Assistant OS).
Talos on IncusOS is likely a very interesting stack that I intend to play with hopefully in the near future.
This is relevant to what I have been learning about recently!
I'm getting ready to launch an online game and I'm dealing with "how do I just run my game server on dozens of boxes without dealing with linux stuff".
I don't really have an answer yet (leaning into "just get one really powerful box" lol), but my investigation into the problem so far has been pretty interesting.
You can conceptualize the "my program + the OS" as a single program. It's not a pretty picture. Lots of global mutable state. (Also it randomly modifies itself??)
The whole point of Docker appears to be "I just want to run my program", in the least painful way possible. Immutable Linux extends the "lean in the direction of sanity" idea. (The programming and OS worlds seem to be learning the same lessons, from different angles.)
And then there's "it turns out the OS solves problems I don't have, while creating many new problems", which leads to Unikernels. Fun stuff ;)
In a perfect world, I wouldn't need the OS at all. Docker gives me two Linuxes to worry about! The number of operating systems I want to worry about is zero!
Which brings us to Unikernels! Just ditch the OS! Technically the right answer, except... now I'm a kernel developer? Maybe that's the least bad option, long term.
If I had a firm requirement to have only one physical piece of hardware home server on bare metal to run further containerized things on, it would be running proxmox, because that grants the ability to run further QEMU, KVM virtualized things, and then to install docker containers inside of any KVM VMs. Even to use QEMU to fully emulate other CPU architectures if necessary.
Or if not proxmox, without a http GUI, just a boring debian stable x86-64 system to manually install QEMU and virt-tools, virsh toolset on to
run QEMU/KVM things on with purely CLI management.
This is an interesting general concept but being limited to only running docker containers is a huge constraint.
I'm a novice in this space I think. I've self-hosted for over a decade and around 2019 I moved over to Unraid, which is generally pretty visual (web portal or configuring and doing maintenance). I find the web portal very easy. How does one interact with your home server OS? I assume it's all via terminal because there are no pictures on the website?
I like the idea of something like this for swarm mode clusters; not sure if you’re focused on the home server aspect exclusively, but I’ll be following along.
Thanks!
I'm only announcing it for home servers because that's where most people are willing to try it out. But Lightwhale is already running in production, and it makes an excellent Swarm cluster.
So I’ve just set up my home server with Ubuntu server, installed docker with one line and I’m off to the races. What’s different/ exactly the value prop of this? You mention maintenance, of what exactly? Is your server a slimmed down version to run on less powerful hardware? Genuinely curious as I’m new to setting up a home server so seeing how this would benefit me.
I do the same thing. Being immutable is supposed to be great for updates. New image version and if there's a problem you can boot back to the last version no problem.
But functionally, like you I find Ubuntu server fine. I run apt update and upgrade a couple times a year and its local only with tailscale access.
I find these immutable OS's really nice on laptop or desktop. The home directory is the only thing that can be written to so the OS is supposed to be more stable and can't break easily
can't imagine a world in which I'd download a little known distro to put on my home network and use as a server. also, doesn't fedora already have something like this already?
I believe for anything home server (or even production), proxmox got you covered, it’s mature, stable, has strong community, and at the end of the day it’s still debian so you can mod it however you like. You can have containers, vms, firewalls, hdd zfs pools, backups, and more. And you can even use something like community scripts for easier installation, although always read the script before you install anything. I have also been playing with BastilleBSD too but I don’t think it’s there yet.
The source repository isn't very enlightening?
> The actual repository here hosts the source code for Lightwhale, and is not of any interest for most people.
> https://bitbucket.org/asklandd/lightwhale/src/master/
Obviously the software you run needs upgrades, but (again, but a layer down) it's based on Docker and probably someone else is maintaining it. So you pull that new container, restart and the OS is just making sure your data lands in the same place with the new container.
If you're happy with all your software running from Docker this seems like a step up from a Debian or Redhat, and it has a lot less bureaucracy than something like CoreOS.
Whether it's _usable_ I'm not sure (especially around storage management) but it's a really clear pitch.
I've long since thrown everything with a user count > 1 out.
Of course nothing is. But there's a reason projects like "Talos" do exist: no terminal, no SSH, no package manager (how do we like package managers like NPM lately btw?), read-only filesystem, definitely no systemd, etc.
And then a minimal number of executables.
This does, definitely, reduce the attack surface.
I'm not speaking about this Show HN's project but there are such things as systems both more secure and requiring less maintenance than others.
Throwing in the towel and saying: "nothing can ever be 100% secure so we'll always need to patch so we may as well YOLO by accepting npm packages modified 3 minutes ago" is not the way to go forward either.
Talos on IncusOS is likely a very interesting stack that I intend to play with hopefully in the near future.
https://linuxcontainers.org/incus-os/docs/main/
I'm getting ready to launch an online game and I'm dealing with "how do I just run my game server on dozens of boxes without dealing with linux stuff".
I don't really have an answer yet (leaning into "just get one really powerful box" lol), but my investigation into the problem so far has been pretty interesting.
You can conceptualize the "my program + the OS" as a single program. It's not a pretty picture. Lots of global mutable state. (Also it randomly modifies itself??)
The whole point of Docker appears to be "I just want to run my program", in the least painful way possible. Immutable Linux extends the "lean in the direction of sanity" idea. (The programming and OS worlds seem to be learning the same lessons, from different angles.)
And then there's "it turns out the OS solves problems I don't have, while creating many new problems", which leads to Unikernels. Fun stuff ;)
In a perfect world, I wouldn't need the OS at all. Docker gives me two Linuxes to worry about! The number of operating systems I want to worry about is zero!
Which brings us to Unikernels! Just ditch the OS! Technically the right answer, except... now I'm a kernel developer? Maybe that's the least bad option, long term.
Or if not proxmox, without a http GUI, just a boring debian stable x86-64 system to manually install QEMU and virt-tools, virsh toolset on to run QEMU/KVM things on with purely CLI management.
This is an interesting general concept but being limited to only running docker containers is a huge constraint.
Kudos to the great project!
But functionally, like you I find Ubuntu server fine. I run apt update and upgrade a couple times a year and its local only with tailscale access.
I find these immutable OS's really nice on laptop or desktop. The home directory is the only thing that can be written to so the OS is supposed to be more stable and can't break easily
first read looks good, excited to try.
> Can you please add wget, nano, $my_fav_app_omg_i_love_it to the root filesystem?
> No, not likely.
I am guessing the way to use software not already in the image is to use `docker run`.