Debian Still Debating Systemd vs. Upstart Init System

[profoundWHALE]

I still have seen very little of this:
Systemd is great for this
RE: Systemd is great for that but bad for this

Compared to:
Systemd is the best and you all suck
RE: Not as much as you and systemd.

So, to get back on topic, does anyone here disagree with the cons that I listed in the previous post? And yes, I know that this decision will go between just systemd and upstart, but it's all about choice and discussions right?

[GreatEmerald]

Yes, I disagree with the cons you listed as well. There is little work involved in switching to systemd, really. Because it already works on Debian. The whole decision is whether to make it the default or not. Meanwhile, there are attempts to port OpenRC to Debian, but they are still experimental. There would be more work to switch to OpenRC than to systemd or Upstart. And speed is not even considered an argument, all systems are fast enough, the important thing is features.

As for the discussions being what they are, well, that's because systemd is truly the best option. As I said before, Upstart only has slightly better portability as the technical argument for it. All other technical arguments are in favour of systemd. And people who don't want to see Debian adopting systemd seem to be motivated by non-technical arguments.

[rohcQaH]

The biggest distribution not using Systemd besides Debian and Debian-based distributions is Gentoo, who is using OpenRC. Why is Gentoo using OpenRC, because it was created by a Gentoo/FreeBSD developer!

Gentoo default installs use OpenRC, but switching to systemd is as easy as checking a box in your kernel configuration, setting "systemd" use flag and re-emerging, and adding a kernel parameter to grub.conf.
Apparently some manual configuration is required, since not all service packages carry systemd init files, but it works and it's supported.

Here's an example of how `simple' Upstart is. Consider the following Situation:

• user Alice has defined three services called A, B and C, respectively
• service A gets started after both B and C have started
• furthermore, service A stops when either B or C stops

Alice now wants to change the configuration of B and then restart it manually. So, here's what happens:

• Alice makes some changes
• Alice restarts B
• B goes down
• A goes down automatically
• B goes up ...

Oh dear, if I wanted to put up with crap like that, I'd stick to sysvinit, thank you very much.


OpenRC has a similar problem. I don't use network manager, so I have a simple script that does:
/etc/init.d/net.eth0 stop
/etc/init.d/net.wlan0 start
or the other way round. Every time I start it, sshd and friends are stopped, but never brought up again.

There once was a configuration option where net.lo would satisfy "net" and stop these issues, but it seems to have disappeared..

[zester]

I can't stomach the bloat in systemd, dbus, udev, ... more than likely someone is going to say "How is systemd bloated?". Its an Init Daemon
and it uses Glib enough said.

You guys/gals can think whatever you wan't but I am telling you give it some time and before long your system is going to be soooo bloated
your going to need 16GB of ram just to get your desktop to function properly.

Linux is the only operating system I have ever installed on my computers, except for the BSD's on occasion, when I first started using linux
My webbrowser was lynx and my file manager was midnight commander, we created /dev and its contents by hand.

You don't want systemd, kill it before it's to late. This isn't an anti-systemd thing this is an anti-bloat thing. Those of use who are lifers
have been down this road before with HAL, PulseAudio, Mono, ....

Use it privately if you wish but don't support it publicly!!! please please please listen.

[89c51]

I can't stomach the bloat in systemd, dbus, udev, ... more than likely someone is going to say "How is systemd bloated?". Its an Init Daemon
and it uses Glib enough said.

You guys/gals can think whatever you wan't but I am telling you give it some time and before long your system is going to be soooo bloated
your going to need 16GB of ram just to get your desktop to function properly.

Linux is the only operating system I have ever installed on my computers, except for the BSD's on occasion, when I first started using linux
My webbrowser was lynx and my file manager was midnight commander, we created /dev and its contents by hand.

You don't want systemd, kill it before it's to late. This isn't an anti-systemd thing this is an anti-bloat thing. Those of use who are lifers
have been down this road before with HAL, PulseAudio, Mono, ....

Use it privately if you wish but don't support it publicly!!! please please please listen.

You sound like those neckbeards that think the command line is good for everyone and use a featureless WM (if at all).

[rvdboom]

What a useless thread.

But this must be answered :

I really thing these distros need to disappear due to them holding back the progression of Linux and negative affliation with BSD:

-Debian (because of their kfreebsd port)
-Gentoo (their support of BSD. There are too many pro-BSD devs to gentoo. OpenRC was created by an ex-NetBSD developer Roy Marples)
-Slackware (refuse to accept modern features in favor of being BSD-like)
-CRUX (same as slackware)
-DracoLinux (same as slackware)

In what way Slackware is holding back Linux progression?
I run slack-current on all my systems with 3.12.6 kernels, with git versions of mesa and open source AMD drivers, KDE 4.12.0, most applications in their very last version if not in recent git pulls.
It's not behind in any aspect. It's just that the Slack devs are reluctant to add Pottering's projects for some reasons, mainly because they seem to change so many things while providing little added value, at least for users since apparently systemd simplifies things for devs.
I tend to respect their choice as it has never impacted my computing experience in a bad way, on the contrary. And I prefer to simply edit a small shell script to enable/disable service start at boot than anything else.

Why don't zaelots on both sides just leave the others alone? Let's each distro/system choose what fits their goals and if you're disatisfied with the decision, choose another distro/system.

[yogi_berra]

You sound like those neckbeards that think the command line is good for everyone and use a featureless WM (if at all).

You sound like one of those stupid millennials that just downloaded Ubuntu and thinks he knows how to administer linux but is lost when faced with actual administrative duties.

[GreatEmerald]

It's not behind in any aspect. It's just that the Slack devs are reluctant to add Pottering's projects for some reasons, mainly because they seem to change so many things while providing little added value, at least for users since apparently systemd simplifies things for devs.
I tend to respect their choice as it has never impacted my computing experience in a bad way, on the contrary. And I prefer to simply edit a small shell script to enable/disable service start at boot than anything else.

Actually, systemd simplifies things for users. One major thing it has is journalctl, which allows for very nice debugging of any and all services (it even filters out messages depending on their priority, so it's easy to get all the errors from all the services you have running listed in one place). Another thing is a unified way of managing services. You are saying that you "prefer to simply edit a small shell script to enable/disable service start at boot"; that's extremely complicated, because with systemd it's just a simple command (systemctl enable/disable service-name). And it works with all services equally and reliably.

Now for distributions like Fedora or openSUSE, systemd isn't as relevant to regular users (they get to deal with it only in case they need to debug things), that's true. But in more bare-bone distributions, like Gentoo, systemd is just extremely convenient, because you get to deal with the init daemon a whole lot during the system setup. With systemd, you get to skip entire pages of the Gentoo handbook, because things work automagically (or require a simple one-line command to adjust things if the defaults happen to not be optimal). And those are just a few examples of how it makes it easier for users; as you mentioned, it also benefits developers (because they get to use advanced features).

So yes, if you don't use modern technologies like that, you don't get impacted in a bad way, but you also miss out on a lot of noteworthy advances. They don't provide little added value, they provide a lot more, but you won't know about it until you try.

[Nobu]

You sound like one of those stupid millennials that just downloaded Ubuntu and thinks he knows how to administer linux but is lost when faced with actual administrative duties.

You both sound like children, making fun of each other just so you feel better about yourself. You don't need to be a rocket scientist to administer a desktop computer. On the other hand, you don't need to be a rocket scientist to administer a large server cluster either, you just need to know the tools used to administer such a system. In that respect, neither sysvinit nor systemd is better--it's the admin which determines how well the tools work, and that varies depending on the administrative person himself.

Squabbling about how great one system or the other is ends up boiling down to "this system is better because I'm so good at using it"...which is great, until you realize there are many other users who have a differing opinion, because they're "so good at using" that system. And then we have a stalemate. Should we continue using the system that takes so long to learn, simply because it works and all these people know how to use it? Or should we move to this other system that is less difficult to learn, and simplifies many tasks, but is (in some ways) more "bloated" than the other? Lets not even get started on the other options, which seem to be there only so that every camp is nearly equally upset...

</rant>

[BlackStar]

Actually, systemd simplifies things for users. One major thing it has is journalctl, which allows for very nice debugging of any and all services (it even filters out messages depending on their priority, so it's easy to get all the errors from all the services you have running listed in one place). Another thing is a unified way of managing services. You are saying that you "prefer to simply edit a small shell script to enable/disable service start at boot"; that's extremely complicated, because with systemd it's just a simple command (systemctl enable/disable service-name). And it works with all services equally and reliably.

Now for distributions like Fedora or openSUSE, systemd isn't as relevant to regular users (they get to deal with it only in case they need to debug things), that's true. But in more bare-bone distributions, like Gentoo, systemd is just extremely convenient, because you get to deal with the init daemon a whole lot during the system setup. With systemd, you get to skip entire pages of the Gentoo handbook, because things work automagically (or require a simple one-line command to adjust things if the defaults happen to not be optimal). And those are just a few examples of how it makes it easier for users; as you mentioned, it also benefits developers (because they get to use advanced features).

So yes, if you don't use modern technologies like that, you don't get impacted in a bad way, but you also miss out on a lot of noteworthy advances. They don't provide little added value, they provide a lot more, but you won't know about it until you try.

Well said!

As an Arch user, I used to be skeptical of systemd due to the amount of negative posts on phoronix and elsewhere.

I was mistaken. Systemd was by far the best thing to happen to Arch in the last few years. Managing the system is now a joy, compared to a freaking pain in the ass that it used to be. It really makes a difference.

A happy systemd user here. I hope Debian doesn't give in to the political pressure from Canonical and makes the right choice here.