systemd is destructive

[debiman]

M51 has their mind set.
nothing anybody here says can change it even one iota.
suck it up.

[M51]

M51 has their mind set.
nothing anybody here says can change it even one iota.
suck it up.

About what exactly? That systemd is crap? You would be absolutely right on that point, as I've looked at it in depth, long before this issue arose.

That what I found was a bug in systemd? I haven't seen anything even remotely indicating I am wrong.

Might as well close this thread, since it has long since served its purpose...namely to let people know the issue exists.

[edbarx]

Using Debian with sysvinit will hold me over nicely until I can get everything moved over to my private distro.

Sadly, you will still hit your head against formidable walls attempting to author a private distribution. I assume you know systemd's API is used by many upstream packages. I know, as I actually contributed writing the simple-netaid* packages for Devuan.

Don't take me wrong, I pressed you with questions to indicate what is wrong in systemd's code, as I believe, objective criticism is the best way to discuss issues like systemd is having. If you point to parts in code stating what could go wrong and why, nobody can tell, that you are only stating an opinion without supporting facts. The code will speak to support your claims.

[bdtc1]

the entire root filesystem had been remounted read-only! Attempts to remount it read-write failed. Nothing relevant shows in the logs

Would this call for a bug report with Debian itself? Would it be "release-critical"?

And if such a report were filed, would this keep Debian from using systemd in the next Stable release until it is fixed? (Because, how can this behavior be called "stable"?)

[GarryRicketson]

https://github.com/systemd/systemd/issues/2402

and

https://www.phoronix.com/scan.php?page= ... -directory

It is something users should be aware of, being aware of it is the best
way to avoid making the fatal error.

[M51]

Sadly, you will still hit your head against formidable walls attempting to author a private distribution. I assume you know systemd's API is used by many upstream packages. I know, as I actually contributed writing the simple-netaid* packages for Devuan.

I already have pretty much everything I need, minus one or two things which I should have done in a week or so depending on how much real life intrudes. I'm absolutely not interested in Gnome or any other RedHat-isms, so systemd dependencies aren't a big problem. I will not use any software which has a hard requirement on systemd...it's that simple. I'm not interested in making my distro public beyond a few friends and co-workers, so I don't have to support stupid crap just because some yahoo likes Gnome. My distro is source based, so compile time flags are easy to control. It also has reproducible builds, which I feel are an important and underpublicized thing that I know Debian is working hard on.

Don't take me wrong, I pressed you with questions to indicate what is wrong in systemd's code, as I believe, objective criticism is the best way to discuss issues like systemd is having. If you point to parts in code stating what could go wrong and why, nobody can tell, that you are only stating an opinion without supporting facts. The code will speak to support your claims.

No thanks, If you want me to point out all the places in systemd's code I think could go wrong, we are going to be here forever and it would still just be my opinion to those who disagree.
I'm done talking about systemd. It's wasted too much of my time already.

[tomazzi]

No thanks, If you want me to point out all the places in systemd's code I think could go wrong, we are going to be here forever

Exactly. I've already tried this - pointing out that systemd has a serious flaws is pointless - the devs are ignoring just about everything...

I'm still working on an alternative solution: pid1 should never crash, and if it crashes, then it should not trigger a kernel panic - it should re-execute itself, keeping the run-time data.

Regards.

[golinux]

This is pretty much the way the systemd debate went:

[Danielsan]

This is pretty much the way the systemd debate went

So funny and sooo true!
For me is the first time but I have never seen a lobby behind a gpl software, I have my opinion but it is everything but a technical analysis...

[edbarx]

For me is the first time but I have never seen a lobby behind a gpl software, I have my opinion but it is everything but a technical analysis...

If it is an activity, attitude or feature that shouldn't be there, its nature has nothing to do with it being a problem or not. Yes, undue pressures from the outside are badly shaping how GNU/Linux is developed. The new business-oriented trend is too obvious to ignore. My take on it is, that certain developers joined free software movements not out of their free will, but to work for their employers with the latter's business-oriented motives/agendas.

Puts on an earthed tinfoil hat.

[Danielsan]

You got the point, unfortunately the parable of systemd has been, and still continues to be, pretty miserable from my point of view.

[dryden]

Congratulations, but anecdotal

So long as your N=1, so's yours.

(Just sayin')

Nice thread. Just want to say.

Your N=1 seemed to imply "there are no problems, it's just you." You are trying to make N=1 into the majority of "it works".

Even if that majority turns into 80%, it still sucks, because that is hugely detrimental if a system only works on 80% of installations.

His N=1 was proof of the existence of problems. Even if his N=1 turns into 10%, it is a HUGE problem for a system that has to be stable.

So, the situation is not equal.

Negative N=1's are more valuable than positive ones. They effect more strongly the operation of the machine, because a 50-50 success rate is no good, and a 90-10 success rate is no good either. Only a 100-0 or 99-1 success rate is any good from the viewpoint of a system that has to work as the base thing on ALL systems equipped with it (or suck completely).

So your implication that your N=1 defeats his N=1 is wrong.

If the statement was "There are systems on which SystemD works fine" you would be right.

But the statement was "SystemD sucks on many systems" and then your "n=1" is just self-defeating because it is irrelevant. It does not defeat the argument.

Just saying.

[dryden]

For me is the first time but I have never seen a lobby behind a gpl software, I have my opinion but it is everything but a technical analysis...

If it is an activity, attitude or feature that shouldn't be there, its nature has nothing to do with it being a problem or not.

You mean the problem arises out of politics right.

If it was just a technical problem people would dump it and be rid of the problem.

An atom bomb can't kill anyone if it's just sitting there not exploding.

It becomes an issue if people are going to say "You know, we should really use that bomb."

Advertising creates an impetus to lie and to promote and to put it in place of something else that is better. Then people are no longer free to leave that inferior system be; as an inferior system on its own it could do no harm, who would use it? But as an inferior system that is getting pushed, it can do great harm.

[dryden]

Don't take me wrong, I pressed you with questions to indicate what is wrong in systemd's code, as I believe, objective criticism is the best way to discuss issues like systemd is having. If you point to parts in code stating what could go wrong and why, nobody can tell, that you are only stating an opinion without supporting facts. The code will speak to support your claims.

Code is gold but the weird thing is that when you dive into it (anything) usually the corruption is much worse than you initially expected.

You just had an impression from the outside at first. You knew things had to be ridiculously complicated or certain problems would never arise. You knew the model had to be flawed because otherwise some things would have been possible that now aren't.

But you never looked at the code before. And then you did. And then you are flabbergasted that it is even worse than you expected.

At least this has been my life experience thus far .

[dryden]

The OP has effectively admitted that he is just having a rant, as he doesn't want to file a bug report for his particular problem.

I've been using Jessie on two machines at home since it went stable, and am completely happy with it. However, if others find implementation of systemd doesn't work for their usage case they have options.

Everyone likes progress, but few like change....

Yes, the option is usually "out the door".

"Play by our rules or get out".

Nice atmosphere you know. Great addition to the Debian spirit as well. It is a poison.

It poisons the Debian culture even I believe.

Becaues before it was "It has to work for everyone." Now it is "Get out if you don't like it".

Or "be relegated to the shadows" which is effectively the same thing. Marginalized...

Bug reporting is not the only way to change things. That is like saying voting is the only way to do politics. Also a popular argument. And also an equally faulty one.

If the main channels don't work, you do not have to use them. If I don't vote, does that mean I don't have a say in politics? Of course not, I have an even bigger say, because I oppose the entire system.

Those "democratic" people want to stifle everyone who is against the system by claiming that the only way they may have a say, is through the system. In this way, they stifle all criticism of the system itself.

If you file a bug report, you admit that the system is good, or you profess that you like it enough to want to try to improve it though its dedicated channels. However, if you know there is no hope for that, this becomes pointless.

Some things can only be fixed by drowning and by violent opposition . You demolish a house before you build a new one, if the walls are not stable.

This basic truth that you see in construction everywhere, is left unregarded. Sometimes a fire is the best way to purify something .

[dust hill resident]

Systemd is controversial, but ultimately, it'll make Linux great again.

I just updated my main desktop computer from Debian oldstable to stable a few days ago, so it has systemd now. Everything's working great. Lennart Poettering is the best! I support systemd.

[millpond]

Systemd is controversial, but ultimately, it'll make Linux great again.

I just updated my main desktop computer from Debian oldstable to stable a few days ago, so it has systemd now. Everything's working great. Lennart Poettering is the best! I support systemd.

The right tool for the right job.

As i have elsewhere stated, as a Devuan user, with no great love or respect for Lennart, I do see plenty of useful applications for systemd.
I regard it as a form of busybox, something which assumes the functions of others that may not do *very well* put performs functions adequately enough for special device purposes.

A consumer Linux system designed to discourage users from delving into the cli for mods, and keep them distracted with eyecandy on the desktop - in indeed quite special purpose. It should work great with such users.

For sysadmins and others who need full control of the system down to the raw code systemd is a nightmare - it breaks thing, especially when those things are tried and tested scripts for honing and customizing systems. Binary logs are the ultimate insult to anyone running a server. A disgrace actually.

I would not mind squeezing systemd into a Raspberry Pi fitted with WRT or Tomato to make a small roter. There reboot times are critical, and there are few dependencies to break.

I would not mind putting systemd into a toaster, or a burglar alarm.

I do not even really mind having libsystemd on my Devuan systems. It comes when called and then goes away. There it seems to act like software *should*.

I simply have a problem with ANY software that tries to take over and dominate system functions without being easy to turn off, disable, or remove.

SystemD is not the only one to be singled out for restricting freedom. Others do it at the excuse of security, not realizing that security is inversely proportional to freedom. Ask Ben Franklin. I dislike PAM, and I hate grub.

Systemd is only a chancre sore on the underlying ulcer of 'korporate efficiency' that is affecting and infecting the FOSS world and worldview.

Recently it has been determined that the Win microkernel, and especially svchost is not only vulnerable to booger attacks, it has been since day one of the nt kernel, and it is in its entirety unpatchable. As in unfixable.

Systemd is just another svchost app, and do we really want to discover 20 years from now that all our real attempts at system security is nothing but a joke???

I dont really care if my toaster gets hacked, or if my alarms go off at 3am. I can fix that, with an older version of Linux.

I do care if the swat teams come bashing down my door because someone in the DNC decided to use my system as a bot for their emails.

[bw123]

Systemd is just another svchost app, and do we really want to discover 20 years from now that all our real attempts at system security is nothing but a joke???

I dont really care if my toaster gets hacked, or if my alarms go off at 3am. I can fix that, with an older version of Linux.

I do care if the swat teams come bashing down my door because someone in the DNC decided to use my system as a bot for their emails.

Interesting point of view, but real examples of systemd being hacked this way would be more persuasive.

[ruffwoof]

If you look at all of the derivatives ... massive amounts of duplication/replication http://futurist.se/gldt/wp-content/uplo ... dt1210.png then I see a common kernel and a common systemD convergence as being a good thing. More eyes all looking at the same, rather than every-which-way alternatives.

[millpond]

If you look at all of the derivatives ... massive amounts of duplication/replication http://futurist.se/gldt/wp-content/uplo ... dt1210.png then I see a common kernel and a common systemD convergence as being a good thing. More eyes all looking at the same, rather than every-which-way alternatives.

It is the sheer diversity that defines Linux to many of us. Ideally, a master of Linux creates his own distro. It is crucial to the element of freedom.

Linux is not cheesecake where one would want uniformity between units.

It is more like pizza - where every attempt to standardize\korporatize it with chains\affiliates only degrade the end result.

Centralization of functions is not always a good thing. There are powerful arguments against the Linux monolithic kernel itself. And systemd is set on a path to displace, and even replace that kernel. If the days of the Linux kernel are numbered, I'd much rather see it replaced by a microkernel than a Lennart monstrosity.
But sadly it will probably not really be Linux anymore.