I’ve been doing a lot of reading on systemd for the past few days. I honestly didn’t notice it creep into my own Debian Jesse system much until I went to reboot and it hung on stopping smbd – for some reason it felt necessary to wait almost 5 minutes (with a counter) for it to die before doing something about it. I could not break out of it, couldn’t even switch to another console to try and help it along. I felt, for the first time, locked out of my own system.

I then read that systemd has much more functionality than just an init system. This scope-creep seems to bleed into the territory of cron, at, system logging (binary logs? Really?), network management, console logins and others. While I can see benefits of things like binary logs (easily filtering for a specific daemon, for example), I fail to see how it would be so difficult to develop a more ‘traditional’ way of accomplishing the same goals (grep, sed, awk?).

In essence, why does systemd intentionally pull out the rug from underneath what has “just worked” for so many years? Why does systemd feel the need to duplicate tools that already exist?

I won’t take part in the personal bashing of systemd developers. I think that people who do this are simply immature. I do, however, think it is irresponsible for such an elaborate system to trump on what is widely known as *nix culture. I feel that the view of existing tools and methodologies has been disregarded and labeled as “old school”. This is the kind of mentality that really turns me away from accepting it as a replacement.

*nix culture has roots deeply embedded in what I feel is the core of computing as a whole. It has its own traditions. It has its own jargon. It relies on and weaves together single-purpose *nix tools to accomplish goals that other nameless operating systems could only dream of accomplishing with such ease. It is a culture of continually honing one’s skills with an existing toolbox to provide a lifetime of technical mastery. *nix culture is not afraid of a very, very steep learning curve because seasoned sysadmins have learned over the years that once you master a tool, you can be confident that you’ll be able to use it for years to come.

For systemd to truly become part of the Linux ecosystem, I believe that it must work within it and not attempt to create a completely new one. To put it metaphorically, the concept of transportation is MUCH more than how fast you get there. The systemd car might be fast, but will never be able to travel on roads only a bicycle can.

—

LOGGED: 2014/12/01 (irc.freenode.net, #systemd)

<lns> Why is POSIX compliance not a goal with systemd?
<zdzichu> lns maybe becaise Linux surpassed POSIX by leaps and bounds?
<lns> zdzichu, from what I’ve read Linux is 99.5% POSIX compliant.
<zdzichu> lns what’s your point?
<zdzichu> car can achieve 100% speed of bicycle
<zdzichu> but can also go much faster
<lns> zdzichu, you could also say that a bicycle can ride on many roads that a car cannot

—

Tagged with: *nix culture | power grab | systemd | tradition

Categorised as: Blogs | Technical