Linux - GeneralThis Linux forum is for general Linux questions and discussion.
If it is Linux Related and doesn't seem to fit in any other forum then this is the place.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
I've always felt that "systemd" was a well-conceived and well-executed project which is very helpful to corporate users who might need to maintain hundreds of remote systems. (Microsoft Windows® has always been very good at this.) The new design incorporates many new ideas that "init" never had, and it especially focuses on pragmatic management issues that we all face. ("A twisty little maze of individual files, all different™" is extremely tedious to navigate. Especially from a distance.)
"init" was just fine for the isolated computer systems of the 1970's. But, this being Linux after all, there always should have been – as Perl programmers like to say – "another way to do it.™" If you choose.
Last edited by sundialsvcs; 12-07-2023 at 09:51 AM.
I first came across systemd during the brief period when I was using Arch. Later I used it under Debian. My initial reaction was fairly positive. It booted faster than sysvinit (on my machine anyway) and shut down fast too, and I liked the systemd configuration files. They were easier to understand and write than the very complex init scripts that went with sysvinit. The division of labour between foobard and foobarctl also made sense to me.
But I went off systemd when it started to expand, growing more and more tentacles that hooked into more and more applications. I believe that an important part of free software is modularity, the freedom to combine software packages in the way that you want, so as to create a personalised system. After all, it's your computer! You should be free to have the init system you want and the applications you want, without them interfering with each other. Being told that you have to use a particular init system to make some application work is a limitation on freedom. I mean, the purpose of an init system is simply to get your system up and running. What does that have to do with running a printer?
Systemd seems more and more like an OS in its own right and I don't like that, which is why I don't use those distros any more.
I've always felt that "systemd" was a well-conceived and well-executed project which is very helpful to corporate users who might need to maintain hundreds of remote systems.
This is a good point. It's funny the bulk of people that go on and on about Systemd are home users. If it was really that bad then corporations wouldn't have let it happen or moved to something else. It clearly works for them. While Linux can be used in a home / non business setting it is definitely not the primary market. Enterprise stuff first. Home users if we have time seems to be the prevailing way of things.
Just because Ubuntu started making it a relatively easy to use Desktop world for the average user back in the day (pretty sure they were first?) doesn't mean that the ecosystem itself will cater to the miniscule linux home desktop user market. The fact that it is as usable as it is outside of a corporate setting is far beyond what I would expect the way the world works where money is the goal.
The levels of eloquence and understanding demonstrated in the description below match that of the video - at least for the short segment I watched - and thus anyone unsure whether to waste 16 minutes of their life can probably spend a few seconds looking at the below before deciding it's not worth the time.
Code:
$ yt-dlp --compat-options=youtube-dl --no-download --write-description Fv3tQbOkz-E
...
$ view "What is systemd, and why its getting so much hate online-Fv3tQbOkz-E.description"
Quote:
Originally Posted by What is systemd... .description
[removed a bunch of sponsors/links/etc]
00:00 Intro
00:42 Sponsor: 10% off your first website
01:36 Init systems and SystemD
03:21 SystemD is bloated?
05:48 Everything depends on it now?
07:01 It's a Red Hat project?
08:44 It restricts choice and modularity?
09:51 It makes Linux less secure?
10:59 Why use systemD?
12:37 Parting thoughts
13:52 Sponsor: Get a PC made to run Linux
14:52 Support the channel
All Linux based systems use an Init system, short for initialization: it's the first process that starts after you boot your OS, and it runs in the background while you're using your computer, to manage system services, and various processes. For many, many Linux distros, SystemD is this init system.
SYstem D is a relatively recent project, at the scale of Linux anyway, it started in 2010, and was spearheaded by Red Hat. Its goal was to replace the existing solutions, like SysV or Upstart, to make things faster and more resilient.
It quickly became the default on Fedora, obviously, then on Arch Linux, Debian, Ubuntu, SUSE, and many, many others.
The famous Bloat argument is one advanced most often. System D, as time went on, encompassed more and more features that were generally handled by individual services, not the init system itself, like device management, login, or network management and creating logs.
This can be perceived as going against the Unix philosophy, where a piece of software is supposed to do just one thing, and to communicate well with other small systems.
What's certain is that most distros that implement it are general purpose distros, that need to provide as many systems as possible, and so they tend to use most of systemD's features and modules.
SystemD also "hides away" certain configurations with its own tools, like systemctl, instead of exposing everything as a config file. Whether these things are important or not, though, depend on the person.
Another criticism levelled at System D is the fact that it has become so pervasive that a lot of other components are created with a hard dependency on it: without SystemD, they can't work at all, or will have a limited featureset. This results in some extra work for distros that don't want to use systemD, as they have to use an alternative implementation of these features.
Another regular criticism of SystemD comes from the fact it's mainly a Red Hat project, or at least was started by Red Hat. The fact remains that while systemD was started at Red Hat, it IS an open source project, and it is receiving contributions from a lot of people that aren't at Red hat.
Another criticism of SystemD is that it's making Linux based systems uniform and that it restricts choice. I'd argue this isn't really true, since there ARE other alternatives, like OpenRC, Dinit, SysVInit and more.
One final problem people identify with SystemD is system security. First, there's the fact that having one single system that powers the init and service management of most distros is a security risk: an attacker can target many, many systems by targeting systemD.
Second, some people would say that since SystemD is huge and does a lot of things, it has a very large attack surface.
But why would you WANT to use it, exactly?
SystemD is a unified project, which means you don't have to learn 20 different programs if you need to interact with something: you learn how systemD works, and you can manage everything.
Compared to other init systems, it's also simpler, as it opens various sockets that services can plug into, and services can start in mostly any order. And finally, systemD is written in C, and isn't the usual compilation of bash scripts, so it tends to be faster and more efficient than many other init systems.
The biggest problem with SYSTEMD is not from the developers of SYSTEMD at all, it is the project developers and distribution maintainers (starting at RH). When they made it a REQUIREMENT instead of RECOMMENDED and removed or forced the admin's (or user's) choice that they went off the rails. Linux, GNU, and FOSS is about freedom: and forcing SYSTEMD and removing the option to use any of the other good init systems reduced our freedom to choose.
BAD idea. Especially with this freedom focused community, it assured that there would be pushback.
The FOSS communities have always understood that, and allowed for choices. The proprietary drivers like Microsoft, IBM, Canonical, etc. have never really understood the FOSS community and the focus on freedoms. The attitude that "market share" and adoption are more important than choices is core to proprietary owners, but foreign to the FOSS community, and it seems to have driven the SYSTEMD adoption.
It is that, more than anything technical (although there are great technical reasons as well) that fuels opposition to SYSTEMD.
It is also that that fuels the projects that have gained support by providing SYSTEMD free alternatives. Because THAT is core to the FOSS community, when we see fewer choices we spin off something that provides more options. This behavior derives directly from the definitions and values that define the community.
The original concept of the UNIX philosophy as Ken Thompson wrote
Quote:
Make each program do one thing well. To do a new job, build afresh rather than complicate old programs by adding new "features".
SystemD tries to do far too much, does some of it badly, and keeps growing into more complicated and complex systems of failure. In the long run the advantage is in keeping it simple, easy to troubleshoot, easy to replace, and quick to fix. They have lost that, but WE have not.
"Although Systemd 255 mainly consolidates and builds on changes that were announced earlier, it does have some new features.
A visible one may prove to be the new BSOD service, which is genuinely more welcome than it sounds. Yes, it does stand for Blue Screen Of Death just like in Windows, but the idea is that LOG_EMERG level messages – meaning an inoperable system – will be displayed on the machine's console full-screen."
Indeed. Oh well, at least the Windows users will feel right at home.
As engaging and well presented as OP's video was, there was nothing in it that changed my mind. Interestingly, the presenter skipped mentioning the fact that the lead developer of the systemd project is now a direct employee of Microsoft.
The biggest problem with SYSTEMD is not from the developers of SYSTEMD at all, it is the project developers and distribution maintainers (starting at RH). When they made it a REQUIREMENT instead of RECOMMENDED and removed or forced the admin's (or user's) choice that they went off the rails. Linux, GNU, and FOSS is about freedom: and forcing SYSTEMD and removing the option to use any of the other good init systems reduced our freedom to choose.
BAD idea. Especially with this freedom focused community, it assured that there would be pushback.
The FOSS communities have always understood that, and allowed for choices. The proprietary drivers like Microsoft, IBM, Canonical, etc. have never really understood the FOSS community and the focus on freedoms. The attitude that "market share" and adoption are more important than choices is core to proprietary owners, but foreign to the FOSS community, and it seems to have driven the SYSTEMD adoption.
It is that, more than anything technical (although there are great technical reasons as well) that fuels opposition to SYSTEMD.
It is also that that fuels the projects that have gained support by providing SYSTEMD free alternatives. Because THAT is core to the FOSS community, when we see fewer choices we spin off something that provides more options. This behavior derives directly from the definitions and values that define the community.
The original concept of the UNIX philosophy as Ken Thompson wrote
SystemD tries to do far too much, does some of it badly, and keeps growing into more complicated and complex systems of failure. In the long run the advantage is in keeping it simple, easy to troubleshoot, easy to replace, and quick to fix. They have lost that, but WE have not.
war started.
I think it is just wrong. But anyway, check the video, it is explained. All of your arguments. Systemd is just a framework for doing something, and like any other tool, it can be used properly or completely abused.
And don't forget a host you buy today is much more complex than a device from 1980. Our software today is much more complex than their predecessors were 40 years ago. Therefore, it can be expected that the newer the software, the more complex it is and the easier it is to be misunderstood and misuse it.
In the last quarter of a century I've only had to modify initialisation twice. Once was with System V, the other time with systemd. Both operations were easy to do and both went off smoothly. A lot of people will tell you of their theoretical objections to systemd, but how many can actually cite real problems that it's caused? I have one computer running System V, one with systemd. Do I notice the difference? No. Most of this debate is simply ideological.
war started.
I think it is just wrong. But anyway, check the video, it is explained. All of your arguments. Systemd is just a framework for doing something, and like any other tool, it can be used properly or completely abused.
And don't forget a host you buy today is much more complex than a device from 1980. Our software today is much more complex than their predecessors were 40 years ago. Therefore, it can be expected that the newer the software, the more complex it is and the easier it is to be misunderstood and misuse it.
Perhaps I failed to explain this earlier, but I do not watch videos. I get my information from images and the printed word. I really value excellent documentation, but video has nothing for me. Not to mention the security issue with clicking on media links that you cannot verify are harmless.
In the last quarter of a century I've only had to modify initialisation twice. Once was with System V, the other time with systemd. Both operations were easy to do and both went off smoothly. A lot of people will tell you of their theoretical objections to systemd, but how many can actually cite real problems that it's caused? I have one computer running System V, one with systemd. Do I notice the difference? No. Most of this debate is simply ideological.
There are no differences that matter if you are a casual user and can simply reinstall when things go south. There are VERY important differences when you are a sysadmin who manages servers for thousands of users and dedicated enterprise applications. Troubleshooting SystemD connected behaviors can be a nightmare! If you are deeply invested in the simplicity and reliability of Linux soutions over more complex and less reliable platforms SystemD IS ABSOLUTLY a nightmare! The ideological issues are also real, and obtain directly from the technical issues, but both matter.
That said, they may not matter TO YOU! And that also is a valid viewpoint. Just not MY viewpoint, so I do not care to try to explain it.
Do I notice the difference? No. Most of this debate is simply ideological.
You're right. A non-technical user such as yourself won't even notice. Anyone who administers networks or regularly starts and stops services on their machine/s certainly will.
I personally like being able to use commands like "chmod +x /etc/rc.d/rc.some_service" to achieve this.
I also value the ability to edit text files to change system configurations, and know what the result will be instead of blindly running a tool that makes the changes for me and hoping for the best.
systemd was designed for desktops, and when RedHat made it the default there were lots of complaints from system administrators. You will easily find technical problems with systemd if you care to, but you probably don't.
Dismissing this debate out of hand on the premise that it's "simply ideological" is akin to dismissing the entire history of Linux, GNU and many, many other free and open-source projects whose foundations were ideological.
Yes, there are some things which systemd does well, and in some cases better than any of its predecessors. Regardless, I live in the hope that I'm never forced to use it.
As highlighted by wpeckham, the price is too great.
I'm interested to see what Debian will do when this happens. Is the frog cooked?
The oldest distributions still in development are Slackware and Debian, in that order. RedHat came a year later, but quickly gained a rather large community following. Debian also had a large community, and a strong relationship with the community. Slackware had a smaller but devout community and continues to have a loyal community to this day. Now Debian has made some social and technical decisions that have cost it some following, and that is important. RedHat has been absorbed into IBM and has totally lost the open-source focus and MOST of the community support! Of note, the Kernel development team is NOT an IBM resource.
The point is that RHEL is not the tail that wags the Linux dog, they are just one very distribution that HAD strong influence. That influence is weakening
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.
