systemd over engineering

Posted Dec 2, 2014 12:20 UTC (Tue) by mpalamara (guest, #69530)
Parent article: The "Devuan" Debian fork

systemd suffers from a severe case of over engineering. It's solves problems that do not exist and does it in an inelegant way. If systemd was worth using you would see a general consensus. There is no consensus.

systemd is here from the moment. I'm probably not going to use most of it's features. It's time to go back to the drawing board. SysV is old but it works well enough for what I have to do. A better init would be great but systemd is service manager plus a whole lot of baggage. I don't care about most of what systemd is selling. I don't understand why systemd aims to replace services that work fine. It's just a waste of code that could be better spent solving real problems.

Code speaks louder than word. Forks can seal the breach for the moment but a new init is needed.

systemd over engineering

Posted Dec 2, 2014 13:25 UTC (Tue) by pizza (subscriber, #46) [Link]

> systemd suffers from a severe case of over engineering. It's solves problems that do not exist and does it in an inelegant way. If systemd was worth using you would see a general consensus. There is no consensus.

AKA "I don't care about the problems claims to solve so the problems must not exist for anyone"

(And by calling systemd 'inelegant', if you're being intellectually honest, you must consider the old hodgepodge of what it replaces "lord please gouge out my eyes with a rusty spork" because, let's face it, it's about as far from elegant as it can possibly get)

Incidentally, I think the fact that the a clear majority of Linux distributions have willingly (and indeed, happily, for it solves their very real problems far better than what came before) switched over to systemd is a pretty loud consensus.

> Code speaks louder than word. Forks can seal the breach for the moment but a new init is needed.

You're absolutely correct -- Code speaks louder than words, numerous forks of sysvinit and incompatible rc scripts were badly "sealing" the breaches between Linux distributions, and a new, unifying init was definitely needed. Guess what -- some folks spoke with their effort and code, and the result, and current state-of-the-art, is systemd. If you think you can do better, by all means, show us the code. Or heck, even the design. You have to start somewhere.

systemd over engineering

Posted Dec 2, 2014 13:33 UTC (Tue) by mpr22 (subscriber, #60784) [Link] (9 responses)

It's solves problems that do not exist

If you had limited this statement to "it solves problems I don't have", then it would be beyond civilized dispute. However, you have chosen to declare those problems non-existent, and thus set yourself up to be argued with by those who think those problems do exist.

If systemd was worth using you would see a general consensus.

That sentence was nonsense when it referred to emacs or vi in it; I submit that it's just as nonsensical when referring to sysvinit or BSD init or SMF or upstart or launchd or systemd in it.

systemd over engineering

Posted Dec 2, 2014 15:25 UTC (Tue) by mpalamara (guest, #69530) [Link] (8 responses)

True, it doesn't solve my problems therefore I don't want to be bothered. systemd is an imposition. This is a rational response.

You're are not listening to the admins and users. Most of us don't want systemd. The take on systemd varies from annoyance to full rejection. Accepting this is the first step forward.

systemd over engineering

Posted Dec 2, 2014 15:30 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link]

>You're are not listening to the admins and users. Most of us don't want systemd.

Please speak only for yourself. None of us have any real idea what most users or admins want. We just know that major distributions are shipping it by default.

systemd over engineering

Posted Dec 2, 2014 15:45 UTC (Tue) by anselm (subscriber, #2796) [Link] (1 responses)

I teach classes on Linux administration that among other things feature systemd. The people in these classes include all sorts of folks from newbies to people with years of Unix/Linux administration experience (usually not in the same class, thankfully). I have yet to meet anyone who after some demonstrations, explanations, and hands-on experience in the class was not wildly enthusiastic about systemd – especially the ones with previous experience of the traditional Linux init setup, and especially when people get to compare the traditional setup and systemd directly.

I don't know where all “the admins and users” are who detest systemd (other than certain web forums and mailing lists, that is) but I for sure haven't run into one in my professional life. I wonder how much of the “annoyance to full rejection” on the part of “the admins and users” is actually real, and how much is just wishful thinking based on the anti-systemd echo chambers that some people seem to frequent.

systemd over engineering

Posted Dec 2, 2014 18:56 UTC (Tue) by jb.1234abcd (guest, #95827) [Link]

"I don't know where all “the admins and users” are who detest systemd (other than certain web forums and mailing lists, that is) but I for sure haven't run into one in my professional life. I wonder how much of the “annoyance to full rejection” on the part of “the admins and users” is actually real, and how much is just wishful thinking based on the anti-systemd echo chambers that some people seem to frequent."

Yes, the negative reaction to systemd is real.
It comes from many places where IT matters are discussed and experienced
former or current sysadmins.

An example from the echo chamber of the systemd beast:
https://lists.fedoraproject.org/pipermail/users/2014-Dece...

And as time passes on and more "goodies" of and by systemd become apparent,
the true picture emerges. And it is not pretty.
jb

systemd over engineering

Posted Dec 3, 2014 0:13 UTC (Wed) by zlynx (guest, #2285) [Link]

I am a former pro sysadmin, a current amateur sysadmin, and a current developer. I've written my own init scripts and suffered.

I had to learn a few new things but I like systemd rather a lot. I know others who also like it. We are VERY HAPPY with it and therefore we don't bother complaining about distros using it.

systemd over engineering

Posted Dec 3, 2014 7:54 UTC (Wed) by niner (subscriber, #26151) [Link] (3 responses)

I'm responsible for system administration and web development at my company. Our system administrators love systemd for its reliability and because it's so easy to write _correct_ service files. Our developers love systemd because we could get rid of complicated daemonization code and because we don't have to do anything anymore to get great logging features.

systemd over engineering

Posted Dec 3, 2014 8:28 UTC (Wed) by dlang (guest, #313) [Link] (2 responses)

so who exactly is saying that you shouldn't be able to use systemd if you love it?

systemd over engineering

Posted Dec 3, 2014 8:49 UTC (Wed) by niner (subscriber, #26151) [Link]

I was just replying to "You're are not listening to the admins and users. Most of us don't want systemd." by mpalamara who is pretending to speak for "most admins and users" which I quite frankly doubt.

systemd over engineering

Posted Dec 3, 2014 10:14 UTC (Wed) by drago01 (subscriber, #50715) [Link]

No one is saying that. The thing is the technology that works better should be used by default. And that is (by ignoring emotional and irrational nonsense and sticking to technical arguments and facts) systemd not sysvinit.