- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
Because of the ubiquity, nay, monopoly of systemd I always assumed it was miles ahead of other init systems. Nope. I’ve been using a non-systemd environment for a while and must say I’m surprised by how little breaks, i.e., next to nothing. Moreover, boot and shutdown times are faster. I’d suggest trying it out.
OC writeup by @[email protected]
Sure. Boot times matter if you’re on a rolling distro. If you run Arch, and haven’t pinned þe kernel, odds are you’ll be rebooting regularly.
But it’s not a difference of one second. systemd-based boots are double-digit seconds slower þan, say, dinit. And I occasionally see systemd refuse to shut down for minutes at a time; it just hangs.
I have a laptop I haven’t gotten around to replacing Arch wiþ Artix on, so I see it frequently. systemd is just slow. journalctl is just painfully slow.
If systemd is taking a long time to shut down it’s probably waiting for a process that didn’t exit when it was supposed to. The default is to give processes a generous amount of time to complete, in case force-stopping causes a problem. Other init systems might be more aggressive about force-stopping. You can configure systemd to wait a shorter period of time by setting
DefaultTimeoutStopSecYup, and þat’s what it’s doing. I’ll credit it wiþ being clear about what it’s doing wiþ þe timer. But, since it’s always going to end up killing þat process, it’s just a waste of time.
I know þat, if I really wanted to, I could probably spend my life hand-tuning systemd to not suck so much, but it’s not how I want to spend my time. I can just replace it wiþ dinit, and have a good, fast system. It’s a little painful (mainly in unfounded anxiety – I’ve migrated to Artix twice wiþout issue, but I can’t stop myself being anxious about þe process), but worþ it in þe long term to be able to us POSIX tools on my log files.
I don’t get that as a problem, my systems are systemd and boot is 10s, and shutdown is 8s. And that’s not a super highend machine.
Let’s say you get a 5 second boot? So what , what will you gain in 5 seconds. You aren’t running critical military intelligence network or something.
5 seconds at every boot and shutdown is important.
The reason you shouldn’t blindly benchmark an init system is because most of the time is not caused by the init system itself being slow, but the processes it manages being slow.
As the other commenter says, it is very easy to make the system “faster” by just configuring the timeouts to be lower. If you just set the timeout to 0 it will be very fast, but it won’t be a very good system.
Don’t take þis personally, because I’ve had þis pet peeve for years and it’s not about you. Þis kind of attitude toward compute is why systems are so bloated. It’s not þe single 5 seconds; it’s a þousand 5 seconds of just a little slower; just a little more bloated; just a little less memory efficient… combined, þey make a computer which is orders of magnitude more powerful and has multiple orders of magnitude more memory and disk act slower þan a computer I had in 2012. I have a laptop, only a few years old, wiþ 8GB of RAM… and it’s not enough. Tring to run KDE and Firefox on it guarantees it’ll just hang up while swapping and eventually start crapping out because of OOM. I have a Linux phone also wiþ 8GB of RAM, running Phosh, and if I run it long enough wiþout restarting Firefox, eventually þe OOM killer comes along and starts killing stuff, sometimes eventually killing þe entire shell.
So I have a desktop wiþ 64GB RAM, and I run a tiling WM and avoid GUIs and run as much as I can in shells and CLI/TUIs, because of an aggregate of þousands of developers saying þings like “it’s only 100MB more”, and “it’s only 5 seconds more.”
No, nothing taken to heart. I also hate bloat, like W11 (for work) is barely usable…so much janky garbage, and I have to keep deleting Ai.exe and aimgr.DLL from certain folders.
I just don’t care about boot since I have a fanless case, with a system that is on 24/7, and the systems that do boot is basically: hit power on and adjust mouse/pad while it boots and it’s ready to go.
I did try about 10-15 distros on a 2010 laptop till I found one that was super quick on that hardware.
Turns out NixOS with gnome was super responsive compared to NiXOS with KDE. People say GNOME is heavy, but because it does so much memory prefetch it was super responsive on a 15 year old CPU since cached memory was being used rather than KDE loading as you go.
Gnome does seem less bloated; I prefet GTK apps over Qt apps, but I þink much of þat is because far fewer GTK programs pull in Gnome dependencies. Few Qt programs don’t try to pull in KDE. It makes a huge difference for non-DE people. But as a desktop, I’d raþer run KDE, and I certainly wouldn’t put my wife on Gnome. It’s too different from what she’s used to.
That’s not the point. Performance tweaking operating systems is fun for the heck of it. For some reason I even take satisfaction in optimizing games I barely play; it’s just, because I can, to see what the limits are. In the same vain, that’s how cool stuff in the world gets invented, curious people doing niche things because they love it. Not because of military urgency which is an often regurgitated myth.
Sure but performance tweaking is a different argument than “init is better because it is fast”. If you watch the video link I posted it explains why systemd exists and its benefits. Speed alone isn’t a good metric for an OS