I have a thread in general linux about privoxy not working in systemd.
But I was about to do some other explorations that are similar. I was intending to use python/flask to log interactions with the web.
And wireshark is similarly close to the kernel. Much of security is similarly close.
If systemd is hosing services that used to work, will I have similar problems with the above?
WTF is systemd doing? Why is it so different? How is this better?
A useful answer would be pointers to some web pages that argue systemd isn't a really bad idea...or even if they DID argue that, I'd know what problems I needed to fix.
This is probably ignorance talking, but so far I hate systemd.
(and if you can answer my other thread, great!)
(and I just found the other threads. Nevertheless, systemd people should understand the breadth of the pain they're causing)
I think I'll be default booting SystemV.

Rants like this should be in a blog rather than a forum post.

Go outside, breathe some fresh air.

I rebooted with sysvinit in Debian jessie. Then I typed and got:
Are Cgroups a systemd concept?
Am I correct in guessing that the jessie "sysvinit" is really a hybrid of systemV and systemd?

Not really but that depends on what you mean by "hybrid".

PID1 should be SysVinit -- check by running:
Lots of programs will not run without systemd-logind and a few other systemd libraries.
https://wiki.debian.org/FAQsFromDebianUser#systemd

Shrug ... SystemD is simply a different way of doing things, which (this being Linux ...) you can choose to use, or not.

I think that the main thing these folks were trying to address is "the niggling number of separate configuration files," but without repeating Windows' mistake of creating "One Registry To Rule Them All" (which turns out to be Windows' most serious and most entrenched architectural weakness, IMHO). I think that they were trying to get a lot of the goodness of Windows' approach (which actually has a lot of goodness! ) without repeating all of its mistakes.

It is a very different strategy, and it's sure to be polarizing: you're gonna like it a lot, or not-like it a lot. It greatly expands the concept and the role of "process #1" from simply being a progenitor, grim reaper and baby-watcher. Maybe you think that's swell. Maybe you don't.

But if you have a lot (read: hundreds) of Linux systems to maintain, e.g. the contents of a large blade-rack, this strategy might (or, might not) be just what you were looking for. I do think that the designers of this project were thinking in this way. It should come as no surprise that Red Hat is a big player in it, because that's the profile of a lot of their customers.

Some people have said that "Linux system administration doesn't scale up well." They might be on to something. To their credit, Microsoft's engineers have paid a lot of attention to "system management in-the-large," and I think they deserve recognition for that. (They've certainly sold a lot of copies for that.)

I think that it's a testament to Linux's power and flexibility that you can actually architect sweeping things like this ... all in open source ... then choose whether to use them on your computer(s) or not. It dramatically changes the system. Therefore, it's a testament to Linux, that Linux can be so changed if one chooses. (Or not, if one chooses "not.")

But let's not make this "a bone of contention." There's no sense in that.

1 members found this post helpful.

The more I read, the more I think systemd is a laudable goal. Is it reachable? counter-shrug. Obviously I'm coming late into this game.
...ROFL
I had wondered why the login-progress icons came up pseudo-sumultaneously (in KDE). I may now know why. Same for the whole boot process.
And I think I understand what "systemd" means...even the system is a daemon. Elegant. Terrifying. Clever. How can all those bloody pieces reliably self-synchronize?
And I don't yet understand how creating lotsa D's makes this net-scalable (I think is the word). Maybe there's a git/VirtualBox/vagrant/docker tie-in here somewhere.I'm really not contending. I didn't realize I was ranting. But my head hurt from beating it on this new harder wall...if that makes sense. Sorry for sharing my headache so affectively.
One last question before I forget. What should I be reading? ...issue list?

I didn't wish to imply that you were "ranting," but some people sure-as-hell do.

Part of the reason might be that the so-called "systemd project" is quite a bit more than just a daemon. It's an entire set of subsystems, tightly interlocked, which touch a great many responsibilities formerly covered by discrete systems such as init, udev, and cron. So, there's an entire "philosophical" aspect to it, and therefore to its acceptance by some. "This ain't your mamma's system-V."

By "one registry to rule them all," of course I meant that, while it was in theory a good idea to coalesce the information that drives an operating system (and/or a network of same) into one place, with a consistent OS-API to secure and to manage it, I think that Microsoft went overboard. If you modify the right keys, you can redefine how the system responds to a .exe file ... and I defy you to even find it.

One thing I have come to know well is that the politix of 'GNUnix' (GNU Linux), in particular, closely parallels the larger society's social politix in general. Therefore you have all the usual pathologies and vested interests, in the GNUnix World... and hypocrisy and crass ignorance to add to it. And screaming and yelling. And holding-of-breath-until-one-turns-blue, etc. I'd say there's almost a one-to-one correlation, here. Unfortunately.

So; when it comes to GNUnix (or FOSS/'Open Source'/'Free Software', et al.), there are de facto conservative vested interests defending what came before (however such interests see themselves).

On the fundamental question of whether 'change is good' in regards some new development, these entrenched forces too often err on the side of it being NOT good (when it in fact could be VERY good) -- simply because they prefer to hold onto the investment they themselves have personally made on what came before (and assume this is also so for most everyone else, too). And in the well-known case of the Linux init system in particular and the Unix 'POSIX standard' in general, the common wisdom has been that 'decentralization is good' -- always -- and therefore 'centralization is BAD'.

However what these conservative 'gatekeeper' vested interests do not seem to grok here, is that this is not really actually a binary choice to be made: and that what in fact is occurring is a movement away from 'each Unix tool does ONE thing well, and ONLY one thing' (at least in the realm of system initialization), to a simplifying, unifying model whereby there is a 'Best Practice' level of balancing the former with the latter, above -- which is actually what makes the most sense, most of the time , in the real World; and does so too, in the case of systemd. Which of course, remains a still-unfinished work-in-progress. And which of course MOVES AWAY from the detailed, arcane, 'initiate' practice of the past which many sysadmins doggedly adhere-to out of mere familiarity, if nothing else.

What is actually happening here is a moving away from a past practice which simply overwhelmed many, out of its arcane complexity, to a NEW SYSTEM with up-to-date thinking on how to UNIFY a systems process which in fact requires a modicum of unification -- if it is not to simply fall apart in the future from sheer complexity (as it was already beginning to).

Not all fundamental, revolutionary change is BAD, eh? However, such change often requires vested interests to give them up. And go back to learning something new, along with everyone else...

Sorry, but you can't call "Linux®" "GNU Linux," because "Linux®" is a registered trademark throughout the world. You have to use the term exactly as the owner proscribes.

The only "GNU operating system" was called "HURD," and I doubt you've ever "hurd" of it.

- - -

And, whaddaya know ... the Free Software Foundation also owns, "GNU®."

- - -

I think that if you give anyone a "bully pulpit," a certain percentage of people are going to stand on it and rant about "how good things used to be." If only because ranting is a great deal easier to do than contributing.

Don't like it? Write something else. Write something better, if you can. Then, promote it, and watch as other people, standing on their bully pulpits, start to rant about you!

If these subsystems were "no damned good," distro publishers wouldn't have adopted it as their new default. And, as for me, I'm glad they did, because it has made my job a helluva lot easier. Behind the scenes of most web sites there is a fairly elaborate system of crontabs, and there is often a homebrew "job system," and a whole lot of logic to detect and resolve various conflicts and to make sure that things actually ran ... most if not all of which can now be discarded. I was very glad to be able to have "jobs," and to define complex dependencies, and a great many other things – all of which I sorely needed in several recent projects.