Why use something other than SystemD?
I'll admit I have used Artix, Peppermint OS (Devuan), and Void Linux because I thought the idea of a SystemDless distro sounded cool. It wasn't because I had a problem with SystemD. It's more so that I love that numerous options exist for different system components. For me it's sad to see that certain features or applications require SystemD be present. I like SystemD, just to be clear. I also like the freedom to choose. Maybe, i'm missing something. What do y'all think? What do you use and why? If you use OpenRC or something, why? I'm guessing most people use SystemD and don't give it any thought which is fair.
29 replies
We were here first.
I don't like systemd's ways, through and through.
Things like KISS, "unix philosophy", freedom of choice of components, and not being an apparent embrace-extend-extinguish coercively marketed monopolosing tool of the corporation, matter to me.
Free software's only the first step to liberating cyberspace. I don't think we should start shuffling backwards. I think we should take more steps forward towards liberating cyberspace. I have a notion of "pedagogyware" as one candidate next step, parts of which (e.g.) are somewhat disavailed by systemd, what it is, how it integrates, and the context in what brought it to us.
Also, it's just smelly. How it's presented like the defacto choice. How it's promoted, like that. Even as far as like it's a necessary only way. Like it's "modern", implying all others are old and obsolete. ... An argument dismissing the many other init systems, not just older, but newer, some even more featureful. Not just promoting itself on its own merits. That kind of disingenuous promotion, is just smelly.
Cue downvotes from systemd enthusiasts.
I prefer init-freedom.
Top examples, Gentoo, and BedrockLinux [there will be others out there... LFS?]. They do not do either groupthink of "thou shalt use" for either systemd or "init-freedom" (as a euphemism for any init system except systemd). So, as an avid user of both, where does that fit with your OP question(s)? Though I suppose, as much as I appreciate the non-imposition and retention of user's choice, I do choose to use them without systemd. So then I'm back to answering why use something other than systemd, and this reply recurses from here to the start again.
Looking forward to you similar posts for each of the other init systems. If only to show this has not been (unwittingly or otherwise) another piece of the smelly. ;D
The main practical reason to not use it is if you don't need it and have an environment or task that would benefit you going for something else.
A classic example is containers, if they do have an init system at all, it's usually not systemd but something much simpler, because the environment/task doesn't need it and there's a memory and storage benefit to not shipping the extra bulk
Void uses runit because they want to optimise for quick and lightweight without caring about wide compatibility, so that's a choice aligning with their philosophy.
its an absolutely terrible example. they literally had to reinvent the wheel (see sidecars) instead of just using what was already available and works fine. (see podman and its integration with systemd)
I more meant a classic example of a situation where someone wouldn't usually use systemd, even if there's nothing stopping it, that most people would be familiar with
I'm not talking about the merits of containers
Systemd often just straight up doesn't work in containers due to containers having security restrictions that prevent processes in them from accessing certain parts of the Linux kernel. Systemd tries very hard to be a container manager (nspawn, portable services, cgroups controls, and so on), whereas other init systems don't do this, making them work inside containers
Because Microsoft has back channel influence on systemd development^1^
It's fine now, but the future trajectory is locked into the path Android has followed - extending dominance over all widely used distributions, then gradually using that dominance to choke and extinguish any alternatives.
Additionally, I saw a site around 3 months ago that VIGOROUSLY documented the financial ties from amutable to Microsoft and Google. I have been looking for a good 20 minutes but can't find it. If this rings a bell for anyone that can remember more details please comment!
https://itsfoss.com/news/amutable-linux-security/
I find that to be a pretty weak argument given that the Linix foundation is funded almost exclusively by big tech
It is almost as if they have a incentive to support the thing that makes them billions of dollars
No, you're misunderstanding me. They aren't trying to kill Linux.
Rather, they are trying to add a walled garden to it that they can use to drive out hobbyists and privacy enthusiasts.
I think they might just be supporting something that serves their use case well. Both those companies manage massive amounts of servers that benefit from systemd being what it is.
If it wasn't useful for them it wouldn't make sense to invest money in it. I am not trying to make the argument that this is some kind of single minded hatred playing out.
With that said, how is Linux improved by having a single " init system " effectively steered by the worst of big tech? Systemd is becoming harder to get away from and reducing interest in alternatives.
Yeah no that is definitely the case. As OP said, alternatives are good. I just think that having a single init system everywhere is not what they're trying to accomplish, they're just investing in what they want to use themselves.
Though I guess in the end what they're trying to do is not really what matters, what matters is what actually happens. And to that end I'm really glad that void and company still maintain alternatives to systemd.
I don't think it's a good idea to ascribe any benevolence to Systemd, though.
they are intentionally suppressing alternatives: https://lists.freedesktop.org/archives/systemd-devel/2010-September/000391.html
They have used their size to force others to work around their idiosyncrasies, causing outages and reducing security of a majority of Linux systems around the world.
I have used systemd extensively for work in huge multi tenant environments, and I am in no way saying it just sucks and has no benefits.
I am saying that accepting it as-is will lead to Linux becoming a locked down, big tech owned slopfest.
I don't see any evidence of that
Do you think Microsoft ever tried to use a tactic of embracing a competing technology, extending their influence over it, and then using their leverage to extinguish the threat?
If yes, do you think they have stopped doing that currently?
Do you think Google and Microsoft are currently aligned with the technocratic destruction of EU and US legal status quo?
Boot time.
Fuck, now I wanna try Artix again.
Having different components is like having immunity to certain diseases.
https://blog.nsrun.io/2026/01/15/systemd-vsock-openssh-server/
This alone is reason enough to question systemd.
Systemd is fantastic
I personally don't have many reasons not to use it outside of some edge cases
Your hitting on the main complaint with systemD. Its not just controlling the run levels but integrates more and more things which goes against unix philosophy which ironically (because its stallmans baby) so does emacs.
Talos Linux is an interesting Kuberntes focused Linux distro that has an extremely minimal init system, with only the components needed to launch Kubernetes.
https://docs.siderolabs.com/talos/v1.13/learn-more/components#machined
The idea is that, since Kubernetes can manage services and do basically everything Systemd often does on modern systems, you just have Kubernetes do it instead.
For me it's many things, boot time not even a priority. This is not an invitation to challenge my personal grievances.
If i wanted a monolithic system half of which depends on a single service/library, I would use windows or macos. Basically, I'm trying to avoid vendor lock. I want to be able to replace almost any part of the system on a whim.
I don't care if some user process is not stopping gracefully, I just want to shut down my machine as fast as possible.
When I change a user's groups, I expect that a simple logout/login will be sufficient to apply them.
I want sensible defaults, and systemd has the opposite in my opinion. I don't want to spend hours trying to find one maybe existing setting in some unnecessary system service (which I did not want in my system in the first place but which is shipped in almost every distro) that will probably be deprecated in the next release and I will have to do it all over again.
From what I've seen on social media, systemd's developers and its fandom are insufferable assholes with a superiority complex.
And many many other small and not so small issues that pile up over time (and systemd developers/adopters help with this a lot) and lead to a burnout.
Right now I'm using void btw. And ironically I'm running elogind, which is systemd code. But it does one specific thing in this case and doesn't try to replace the whole os.
I don't have an opinion on the rest of what you said, but you're factually incorrect about systemd being "monolithic". It's a pluggable framework, very much the opposite of monolithic software.
I worked in the industry before Systemd.
Systemd fills a niche it manufactured. It fills it poorly, supplanting established and reliable incumbents with something that is neither. In a world where Unix tools do one thing and well, Systemd does all things marginally.
Nfsroot. Portable skills. Clear source of truth. Pick one.
What innit do you recommend?
OpenRC
Why use something other than Windows? /s
I suppose this should be solved now, but when I tried it, it used to hang at random with the "waiting for xyz service to finish" messages at boot time or when powering off. So I got fed and said yuck, went back to ol' good OpenRC and never ever had even a slight temptation to even look at systemd ever again.
Because as soon as there were hints about requiring OS age verification in some states they rushed to comply quicker than even Google or Microsoft