The Tragedy of Systemd

it is likely faster to code oneself a tiny application under Unix (BSD), rather than googling for several freeware or "closeware" to suit the need under Win.

Actually, it is just 10 min coding, instead.

Additional benefit, if you make use of your term or ncurses, you can compile it on any other nix operating system.
If BSD is no longer, then plan B, plan 9 from outer space;) Plan9 : it's rock.
 
it is likely faster to code oneself a tiny application under Unix (BSD), rather than googling for several freeware or "closeware" to suit the need under Win.

Actually, it is just 10 min coding, instead.
I want to see you code up Visio in 10 minutes. Or Excel. You are off by many orders of magnitude.
 
There is too much mud slinging and personal assaults on these forums :-(
Having read certain recent people's posts, I'm starting to think that a few people discovered this forum, noticed it was a pretty peaceful place and decided to stir it up intentionally.
One of the reasons I've long abstained from creating a username on this forum (and only did that to post about something very specific), is that I wanted to make
sure the community was actually a positive one. It is, and a few people won't change this at all. Just don't get triggered and let them be. They'll get bored fast.

I'll try to steer this thread back on topic.
Even though Systemd has been received with mixed feeling across the communities, that shouldn't mean there is no value in it at all. Since I'm not a Linux user at all
and am really content using FreeBSD as my day-to-day OS, I lack experience with it. I would however be very much interested if someone with experience could list
up the benefits that he/she has experienced. Even people who don't like it must have something positive to say about it ;)
 
There are many reasons to use, or not use, systemd. One doesn't need to tally up various theoretical technical points. Personal preference is a perfectly good reason.
 
There are many reasons to use, or not use, systemd. One doesn't need to tally up various theoretical technical points. Personal preference is a perfectly good reason.

High resource usage is sufficient to not use it.
Furthermore a single Big program to control everything is not unix.
 
Moderator, please close this thread.

If there was a single post that addressed the factual issues with systemd (and there are plenty), I missed it due to all the noise.

BTW: I stopped listening to the OP's linked video at the point where the presenter stated in regard to the systemd portability limitations that Unix is dead and "today you have Linux and some rounding errors".
 
It's interesting that systemd can be so contentious that even FreeBSD users are fighting over it...

What is clear to me is that FreeBSD, NetBSD, OpenBSD, TrueOS, DragonFly, etc, etc, etc, will forever be a "perpetual downstream" (so far as ports/desktop stuff goes), working with and around, the crap coming down the pipe from the Linux/systemd/freedesktop/gnu world.
 
will forever be a "perpetual downstream" (so far as ports/desktop stuff goes), working with and around, the crap coming down the pipe from the Linux/systemd/freedesktop/gnu world.

Thats actually fine. I would rather have to handle a fossilised dinosaur turd than a fresh one!

I almost like to think of Linux in general being the CURRENT or UNSTABLE branch of UNIX. Then when things are finally correct, they are available for less hobbiest users. I also note that so much Linux stuff actually disappears (after making such a massive noise) before it ever hits production, so I am quite happy to wait rather than waste my time learning it only to see it get replaced next week.

The biggest example is Wifi and networking. There are so many different ways over the last few years to set this up:
  • /etc/network/interfaces
  • /etc/systemd/network/25-wireless.network
  • /etc/sysconfig/network-scripts/ifcfg-wlan0
I am quite happy to wait until they finally decide on one... Or stick with how FreeBSD has done it for ages. I think the most radical change we ever had was that you needed to create the wlandev before using it (FreeBSD 9?).

That said, I don't dislike Linux. I do however dislike "modern" Linux and what it is becoming. A massive time waste in many aspects.
 
Bla bla bla bla. If you don't like systemd then don't use it. Mod close this mess!

Sorry, but that is an extremely naive statement. systemd slowly engulfed the most popular Linux OSes one by one: Fedora (2011), Arch (2012), CentOS (2014), SuSE (2014), Ubuntu (2015), leaving fewer and fewer options to "just don't use it." That's how free software and services works: if not enough people forcefully reject what they see as bad practice, then bad practice becomes the norm. There is a pattern being set by systemd, which Linus himself worried about: it is a rejection of giving people choice -- Google did it with Android and Chrome, so Mozilla followed suite with Firefox; Apple does it with all of their products. They force people to accept what benefits their companies at the expense of personal freedom of the consumer (choice). What good is the freedom to choose when the options are all the same garbage in different flavor?
 
A former client of mine offered me a boatload of money if I would write some Linux drivers for their product line. To do some preliminary investigation, I started installing a few Linux distros to see which one I would use to do the work.

Linux is a zoo. Not the St. Louis Zoo or the San Diego Zoo. It's not Marlin Perkins' Wild Kingdom even. It's like wandering the bad parts of town and seeing rats creeping out of every corner. And I'm not done yet. And I AM being nice.

I might do a brief write up about this in a few days.
 
Bla bla bla bla. If you don't like systemd then don't use it. Mod close this mess!

Thank you. This anti-systemd stuff is getting out of hand. I didn't like it that much either when it took over in Arch Linux but I never had any problems with it beyond "a stop job is waiting for..." and a couple error messages at shutdown.
 
I almost like to think of Linux in general being the CURRENT or UNSTABLE branch of UNIX. Then when things are finally correct, they are available for less hobbiest users.
You seem to be saying that Linux is used by hobbyists, and by implication that Linux development is driven by the needs and wants of hobbyists. Nothing could be further from the truth. The vast majority of all Linux installations are used as servers, by "professionals", such as the FAANGs (Facebook, Amazon, Apple, Netflix, Google, and by implication other large cloud-based companies such as Alibaba or IBM), by industry in general (GE, GM, and other companies whose names starts with "General"), by government (whether is is weather forecasting or national security), and by supercomputers. And the number of machines in these uses is gigantic. For example, look at the Top500 list of supercomputers: Every single one runs Linux, and each of them are somewhere in the neighborhood of 10^3 to 10^6 CPUs. The big data centers of the FAANGs have millions of Intel chips too ... mostly running Linux. The Linux ecosystem is dominated by professional server use, if you count the numbers.

Or look at public-facing servers. There is a reason it's called a "LAMP" stack: They all have Linux. There is some debate about Linux' market share among public-facing servers (depending on how you measure it, the number is between 50% and 90%), but the number of these machines is still huge.

It gets even more stark if you look at the money flow. Linux today is a big industry, with the two largest players being RedHat and SUSE. Those are multi-billion dollar companies. They drive the development of Linux packaging and the Linux kernel, together with IBM and Intel (both of whom contribute large amounts of manpower and money to Linux development, to some extent through the foundation). Both RedHat's and SUSE's revenue comes from customers who pay for supported versions of the OS (famously RHEL, where the "E" stands for enterprise). In RedHat's case, they de-facto control Fedora (through their sponsorship) and CentOS (which is just a free clone of RHELs packaging). There are many million Linux machines that get their OS distribution and support through these support contracts; very few hobbyists pay for support. So if you measure influence by the amount of money changing hands, hobbyists are even more irrelevant to the Linux ecosystem than by raw machine count.

Face it: Systemd exists because several large Linux entities (foremost RedHat and SUSE) think it is a good idea. And these people are neither stupid nor criminals. Systemd does indeed solve a real-world problem, namely that the whole mess of how to start processes and services had become unmanageable complex. In my personal opinion, systemd is low-quality software, which is focused on the wrong use case (Lennart cares about his desktop experience, and grudgingly supports server features), and written by a famously sociopathic person. But it is (for better or worse) still preferable over the alternative, which is why the Linux big companies continue to not just support but embrace it.

Personally, I use and manage quite a few Linux machines with systemd, and have even have to write services that work under it (including configuring systemd). It's painful at first, and after a while you get it to work. Would I prefer to do it in a RC-based or Init-based setup? Absolutely, but I don't get to make that choice.
 
Would I prefer to do it in a RC-based or Init-based setup? Absolutely, but I don't get to make that choice.
I'm involved in a commercial project based on Beaglebone Black, and I have made that choice: I switched Debian to openrc – it's still possible in "official way".
So far, so good: hundreds of such controllers work in production for more than a year. From time to time I perform system updates using official Debian repositories, haven't seen any impact of missing systemd.
 
I'm unclear how systemd doesn't meet the "off topic" rule:

Since systemd isn't a part of any BSD operating system, not sure how this technical debate about Linux is allowed to continue.

I guess because the talk in the OP was given by a FreeBSD developer, although the recording was taken at a Linux event it originally debuted at a BSD event.

I must admit that I though of posting it here in the hope it would generate interesting discussion about a BSD "System Layer". I then thought that a certain maturity was needed for such discussions so decided against it. It's been fun watching how the thread developed ;)
 
You seem to be saying that Linux is used by hobbyists, and by implication that Linux development is driven by the needs and wants of hobbyists. Nothing could be further from the truth.
Obviously what you say is true. Linux is very much a staple of enterprise systems because it is convenient for servers.

However, when it comes to desktop Linux, I very much think the opposite is true. It is aimed towards hobbyists pretty much exclusively. The many Wifi systems, beta (albeit default) Wayland on many distros, experimental systemd features are all things that are keeping the Linux desktop back, pretty niche and far from fit for business purposes.

I think the main reason why we think otherwise is because many developers choose Linux; but I don't really think that is for the merit of the OS but purely because it is the closest match to their production setup.

But many distros are quite honest with this. Fedora is an early access to RHEL features (basically a beta) and thus the users of it can't really be anything other than hobbiests and enthusiasts. When you then realise that most other distros follow suite with the same versions of things like systemd because all distros kinda update their software at the same time... surely that makes their users also beta testing enthusiasts?
 
Honestly, I don't know what fraction of desktop Linux users are hobbyists and religious believers. A lot of software developers use Linux on the desktop (usually laptop), but in addition to the correct reason you give (matches their server production setup), there is also the reason that many software developers are diehard open source fans, and will refuse to use Windows or Mac or a Chromebook, because it is not "pure" enough; that's religion or dogma. But to each their own ... if they want to run it on their desktop, fine with me.

I know that there are non-hobbyist non-expert computer users who download and use Linux. Recently I've found a few examples among elderly (and often not very wealthy) people, for whom Linux is an inexpensive and easy-to use choice. And I know that there are corporations where Linux is the standard OS given to employees; usually those companies use supported versions (such as RedHat and SUSE). But I don't know whether this adds up to a significant number. It can't be a lot, given that the market share of Windows and MacOS together in the desktop/laptop market is about 97% or 98%, depending on whom you ask.
 
I am of the firm belief that 80% of all computer users wouldn't even own a computer if it weren't for games. All the other things they now use it for are coincidental. Note the uprise in people eschewing the desktop in favor of laptops and, now, phones alone that can play games.
 
Maybe it's because I'm not a "power user" but I would never have known I was even using systemd if I hadn't witnessed other people moaning about it.
I never play games. It seems a frivolous thing to do given my impending death.
I like coding . And I use my smartphone to keep in touch with those I love.
(It's a Huawei phone with a Google userland so it is probably the most Orwellian phone I could have chosen - lol).
Sorry for drifting off topic as usual...
 
I tried to research the systemd issue about five years ago. I collected references, blogs, newsfeeds, and
articles written pro and contrary. In fact, the very reason I started FreeBSD was that it was absent.
But, that does not mean I would throw away an entire learning-curve or experience all at once. This gave me some space to experiment both "with" and "without" systemd environments.

The history of software development has many waves of innovation that can improve performance and
reliability. There have been many, many litigation cases that steer the software developments which
exist in the world today. Basically, don't go against the stream of development. Complain and argue
the merits, and when there are breaches of ethical standards - then yes, please speak up.
 
systemd is an abomination. The only ones not convinced of this are those that don't administer systemd systems. That's the disconnect between the glowing praise from the speaker and reality. Point-by-point I can direct you to documentation from systemd's own developers of how to fix what systemd breaks by preventing systemd from doing what it was designed to do. The theory behind systemd is great: make all of these different Linux systems adhere to a general standard, so that knowing how to administer one translates to being able to administer all the others. But systemd is garbage in practice, unable to meet up to the lofty goals it set for itself.

The biggest reason I left Linux is because of the low quality standards adopted by its community -- everyone is so desperate to get their new toy published and get praised for it that they don't bother putting in the effort to make a high-quality product. FreeBSD suffers much less from that. It is a well-throughout operating system, where development decisions just make logical sense (for the most part).

Yes, the BSDs could benefit from a well-designed and well-implemented type of "systemd" (the opposite of existing systemd), which is what the presenter was trying to convey, but he's clueless about systemd in the real world.

You've met plenty of people who administer systemd systems and haven't had those problems, but you haven't noticed because people don't write long stories on the internet about how their software does exactly what it's designed to do. Unix stopped being a "lean and mean" "worse is better" system in the '80s and the sky hasn't fallen. Most of the major Linux distros (including distros targeted to "serious" admins like Arch) have switched to systemd and the sky hasn't fallen. The entire Android ecosystem, which is fragmented across hundreds of device manufacturers and at least two ISAs, runs on Java, and the sky hasn't fallen. Windows keeps getting bigger every year but the sky hasn't fallen. Lean software obsessives might as well be a rapture cult at this point. _ software ecosystem is certainly going to collapse under its weight this time! Just you watch!

But it. Never. Happens.

Maybe it's because I'm not a "power user" but I would never have known I was even using systemd if I hadn't witnessed other people moaning about it.
I never play games. It seems a frivolous thing to do given my impending death.
I like coding . And I use my smartphone to keep in touch with those I love.
(It's a Huawei phone with a Google userland so it is probably the most Orwellian phone I could have chosen - lol).
Sorry for drifting off topic as usual...
The Huawei part doesn't even matter--unless you're Chinese, the Chinese government couldn't care less about whether you're a dirty subversive. Worry about your own government.

Also IME the entire point of systemd is fading invisibly into the background, automating key aspects of system administration so you can direct your attention to more interesting things. Interfering with systemd doing its job is likely to do more harm than good.
 
The systemD concept is actually very good and was not new when they brought that, and so there were not just SMF for them to get ideas, but also there are some good ideas from Amoeba etc.

The real problem is the problem of almost all software inside the Linux ecosystem: low quality. Those who rely heavy on Linux, like Google or big irons (aka IBM), usually do a lot of in-house engineering to make that reliable (and of course never release the code), inclusive their own release engineering - so in practice it is not the same Linux we can freely download.

One thing I really don't understand is how they are able to have so many poor designed and implemented stuff. They are quite competent into that.

The Linux Foundation is supposed to support Linux, and looking on their Members it is really easy to guess they should get at very least x15 more money than the FreeBSD Foundation and still have a crap sound stack, with an ever worse sound server, crap graphics stack etc.

It is not like the Linux Foundation need to rely on some individual willing to write a (e.g.) great audio/sound stack and server on his/her free time for them, like they do. They could simply contract specialized professionals to design and implement one (like the FreeBSD Foundation does when there is money available for that).

[EDIT]

If I had to use Linux for something like development or desktop I would probably settle with Arch Linux but using Ravenports instead of their build/package system.
 
Back
Top