https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=825394
systemd now kills all user processes on logout.
Do you regularly detach screen or tmux sessions over ssh?
that's great! because poetterling literally just broke all server daemons unless they have a hard dependency on systemd apis
ggggggg
That's it senpai, I'm switching to OpenRC
>>54813981
Thread's over here.
>>54811266
>>54813981
>another huge fuckup from systemd
>/g/ doesn't even care anymore
It doesn't surprise anyone anymore, isn't it?
>>54813981
>Do you regularly detach screen or tmux sessions over ssh?
>that's great! because poetterling literally just broke all server daemons unless they have a hard dependency on systemd apis
No, I believe that long-running daemons must be run as services (daemons) and in separate session, not from screen/tmux. And as a desktop user I welcome this change and dont see any harm in it. If somebody uses nohup for over 30 years now then that somebody may as well use UNIX from 30 years back.
>linux
>>54813981
OP, protip: if you run your "daemons" manually in tmux/screen as a regular user, then you're doing it wrong.
If you still want to run your "daemons" like that for whatever reason then look for KillUserProcesses in /etc/systemd/logind.conf and switch it to no
it's easier than preventing a Windows 10 update :)
So why is systemd still a thing? I don't use it.
>THEY FELL FOR THE SYSTEMD MEME
>>54814432
Okay, not OP, but what's with the need to put daemons in quotes?
>>54813981
So it's doing exactly what OS X normally does because "muh desktop OS"
How ironic that BSD, the thing that spawned OS X, is now the only *nix that isn't basically OS X
>>54813981
>systemd
Who cares? Let's talk about real issues
https://bugzilla.mozilla.org/show_bug.cgi?id=1276348
whaaaaaaaaat the fuuuuuuuuuuuuck mozilla
Your daily reminder that BSD doesn't have this problem.
If you're a /v/eddit kid who switched to Linux for /g/ points, then I'd encourage you to lie in the hole you've dug, but anybody else with half a brain is welcome to switch over.
>>54814432
>SYSTEMD: MAKING LINUX EASIER
>MANAGING MY SYSTEM IS SO SIMPLE
>ENJOY WASTING YOUR LIVES LUDDITES
>uh, yeah, so anyways, so you'll need to set up a dedicated tmux session as its own user, and then change these obscure configuration options....come on, the manual is only a few hundred pages, you read the whole thing right?
>man you're just not sysadminning hard enough systemd is 2 patrician uuuu (that means "for you")
xd
>>54814432
Shitty default settings like this are the entire reason distro fragmentation exists.
>>54814297
Next time you want to post something like this, just type "BUT ITS THE CURRENT YEAR" in all caps, you colossal inbred myopic retard (english: you are a fat thoughtless piece of shit and can not see outside perspectives or think beyond a prescribed dogma)
So? I would expect a system to kill all started processes when I log out. That's how it should be. It is also completely obvious that if a process shouldn't be killed when I log out that I need to use some way to declare it as such to the environment.
Are you saying Linux pre-systemd didn't do this?
What a fucking amateur OS.
>>54814826
You are in waaay over your head, you stupid winnigger.
Sit down and let the adults talk.
>>54814864
>my shitty 60s software doesn't work anymore because someone actually enforces sane defaults
>>54814878
Once again, you don't even have a clue about what the rest of us are talking about if you think "logging out" means the same thing on *nix that it does in Windows.
>>54814894
The terminal command is called "logout". Lazy cunts press ^D.
>>54814878
How the fuck is it sane? Because you're autistic and have a seizure any time you perceive "clutter"?
Is it sane in some poorfaggot scenario almost no linux users encounter, like multiple people using the same computer?
>>54814908
>how is cleaning up after the user in order to not have dangling processes on the system sane????????
>>54814919
>dangling
Ah so
>Is it sane in some poorfaggot scenario almost no linux users encounter, like multiple people using the same computer?
>Hellom rajesh? Pajeet called, he said your tmux session is using up too much RAM and emacs is swapping
>do you want to walk down and log in so you can kill it? the internet went out again so no ssh
>>54814804
Nice rebuttal you autist.
>>54814934
>multiuser system
>multiple users running the system is a "poorfaggot scenario"
Grow up.
You autists are why linux will never be viable on the desktop.
>>54813981
What is this tard smoking? I've at least tolerated systemd up until now even though I disagreed with it just because it was a hard GNOME dependency, but screw it. Runit here I come
>>54813981
>systemd now kills all user processes on logout
That's actually a pain in the arse.
>Do you regularly detach screen or tmux sessions over ssh?
And you got the reason in one.
This update essentially kills screen/tmux and almost kills SSH over the Internet entirely. Particularly with a bad connection.
You'll be doing something through screen, etc and SSH and then you might disconnect for some reason or another. Congratulations, you just killed whatever you were doing. Better hope it wasn't a long compile or system updates or some shit.
In fairness, I wouldn't mind this if there was a systemd-screen type thing though, that allowed you to maintain a connection over a patch SSH link though. HINT HINT HARRY POETTERING.
>>54814967
It is.
It's for poorfag families who can't afford a computer per member and for losers with poorfag IT jobs.
If you ran into the rare case where you're on a real multiuser system for a good reason other than your $80k/yr sysadmin job in the land of $79k/yr living expenses, some part of the system would already have been configured (or as lennart would say, INELEGANTLY HACKED) to do this if it were actually an issue. But it probably wouldn't be, as such a system would considerably limit user accounts.
Oh wait, don't tell me you work in IT hell and your retard coworkers/boss actually want to eventually use ganoo/linux for your multi-employee tech support workstations.
>>54813981
>kills all user processes on logout
>not keeping them up in the background like OSX does
Why do people insist on trying to make windows bullshit the standard?
just run the command with a & at the end
:^)
>https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=825394
They are suggesting that screen should be modified to request and handle it's own new user session.
This way when you logout, the screen will continue to run as it would be it's own session.
I'm inclined to agree that this is the best solution.
You get your cleanup on logout, but you also get to run something in screen beyond your standard user session.
>>54815103
Programs shouldn't have to work around one init system's faggotry. Systemd should not be changing how linux works to the point where userspace applications have to work around it.
>>54815169
I'd argue this is a better default behaviour though.
But it certainly shouldn't be dropped without notice in true systemd fashion though.
It needs to be phased in over the period of a year or two with a reasonable amount of publicity, otherwise you will get confused users.
>>54815199
A better default behavior for desktop/workstation systems, sure. I agree with you there. It shouldn't be slammed in there without warning, but it might be an improvement. But it's absolutely retarded behavior for basically anything else. It's retarded that they made this the default at all. And it's especially retarded that they did this just to work around the way that GNOME doesn't shut down cleanly on logout.
https://github.com/tmux/tmux/issues/428#issuecomment-222330478
>>54815006
You can just turn it off, it's a setting. Stop sperging.
>>54815237
You're confusing things. Systemd handles logging out and shutting down user sessions. GNOME just expects that feature to work as advertised. This is not a GNOME bug.
>>54813981
This is exactly why debian testing exists.
Do I have any problems on my debian machine? Nope.
>>54814826
Actually, you are entirely right.
Any user processes should be stopped when the user logs out. That makes perfect sense.
If you look at it from another direction, if you kick a user off your system then all his processes should stop. What is the difference between a user logging out and a user being made to log out? A user should only be allowed to have persistent processes if granted that ability.
>>54815279
GNOME shouldn't expect that, it wasn't the default behavior. Systemd shouldn't have made that the default behavior either just because one program expected it, either. Now tons of other programs are essentially broken.
>>54815359
>GNOME shouldn't expect that
systemd said it does that so gnome should expect that systemd does what it says
>>54815380
systemd never did so by default. You can't assume all possible options are turned on because you feel like it, especially when it's not by default.
OpenRC master race when
>>54815245
A default setting. Are you really so retarded you don't get the difference between a setting and a default setting? Same reason people flipped out over unity adding in Amazon results by default. Or the forced Windows 10 upgrade. Or the advertising tracking in Windows 10. Nobody cares about opt-in, but making it opt-out is totally different.
>>54817900
Advertising and tracking*
>>54814217
/g/ is filled with normies who use Windows. Anyone who actually has half a brain has already moved to *BSD.
>>54817566
Never ever
>>54818954
What if I say please?
>>54814751
Neither does slackware or any other distro that doesn't use systemd by default like gentoo.
https://github.com/gentoo-mirror/gentoo/commit/bcc3e48
Systemd is the Hitler of FOSS.
Thank goodness, they reverted it.
>>54813981
this was a pretty obvious turn of events
alarms should have gone off when linger was added and wasn't turned on by default
>>54814751
>Your daily reminder that BSD doesn't have this problem.
OS X uses a motherfucking xml-based init system not all BSD is good senpai
>>54814432
how should I run minecraft servers now?!
>>54813981
I'm at a loss for words. This breaks the expected behavior on any system with a new systemd install. Which violates the principle of least astonishment. It's practically malicious. I can only hope the maintainers of distros like Debian are intelligent enough to change the default settings back to what they should be.
>Not using openrc
>Not using Debian
>Not using gnome
Omg
>>54823315
You can still run background servers, it will just be killed by default when you log out if you don't instruct systemd to bypass them.
>>54823791
>improving security principles
>fully documented
>malicious
>>54823898
How does it improve security? I'll wait.
>>54824061
prevents malicious programs and processes from continuing to run after the user has logged out, for example bitcoin miners, keyloggers, and botnet software.
does not prevent intended applications from running (when set up by the user to bypass)
bypassing is trivial, check out the docs.
>>54824061
There are processes that remain running when you log out that haven't been explicitly instructed to do so, they've been implicitly doing so because of bad programmer practises. This is the reason why this default has changed - you will explicitly notify the system that you want a certain list to run after you've logged out.
>>54814758
I bet you use "daemons".
>>54823791
>I can only hope the maintainers of distros like Debian are intelligent enough to change the default settings back to what they should be
https://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=c11c9a4601ec0dbfb8a64e2c1c0309a590ab838b
>>54824133
But this also breaks the decades-old mechanism for explicitly instructing programs to remain running. nohup, disown, screen - all of these no longer operate as intended under this change. An abrupt shift in default policy is the wrong way to go about this.
Also, it is not and should not be the init system's responsibility to look after poorly coded programs that don't terminate properly. If a program is broken, file a bug report.
>>54814154
>>54814217
>>54814694
>>54814722
>>54814734
>>54814745
>>54814750
>>54814751
>>54814758
>>54814804
>>54814894
>>54814908
>>54814986
>>54815006
>>54815071
>>54815169
>>54815295
>>54817566
>>54819383
>>54819892
>>54822380
>>54823845
>all these "daemon" users
get out of my /g/ redditors. systemd for the win. i bet you dont have all of the freedesktop programs.
>using systemd unironically.
>>54814154
There are two big problems with OpenRC: first, most if not all OpenRC systems
are completely broken (missing daemons, lingering processes, etc. are just the
tip of the iceberg); secondly, it should be called GNU/OpenRC.
For these reasons, it is impossible to support OpenRC on Linux. Problems
appearing on OpenRC systems will usually simply be ignored unless they can be
reproduced on non-OpenRC systems.
>>54824233
It's not the init system's responsibility. Systemd init is not Systemd cgroups. They are different things.
>>54824233
The very reason that we have cgroups is so that the init system can look after poorly coded daemons.
Why are people arguing in favor of this?
This breaks expected behavior and completely breaks the purpose of ssh, tmux, screen, things you expect from a SERVER OS???
>>54824474
tmux and screen are broken because they don't use pam.
>>54824489
Fuck off.
>>54824523
>b-but my workflow relies on this bug
>>54824474
Because it's easy to use it. Either flip the configuration setting or explicitly tell the system to do something different. Problem solved.
>>54824357
>>54824391
But this isn't about cgroups, this is about systemd's maintainers changing the default behavior in a way that breaks standard *nix job utilities.
>>54824489
What about nohup? Should we start changing process signals up now too?
>>54824531
A bug is unexpected behavior that fucks up your workflow.
This new default setting of of kill-user-processes is the literal definition of a bug.
>>54824559
Apparently you don't know the definition of a bug. Files starting with a . being hidden is a bug. systemd killing processes is not a bug.
>>54824540
It's easy for you, because you know about it.
What happens if someone installs a fresh linux install and suddenly, their tmux session no longer persists between ssh logins.
Their first thought isn't gonna be "oh, let me just dig through systemd documentation".
They're gonna have no fucking clue what caused the new behavior and will be screaming on forum sites.
I approve of this change. Since this got implmented, the IRC bouncer I operate has had an increased clientbase.
They're finally getting what they deserve from running irssi inside tmux and then detaching from their sessions.
>>54824559
A configuration setting that can be fully controlled is now a bug? Terrible!
>>54824611
A malicious default is now being pushed to users.
Fuck off, this is a literal regression.
>our software has bugs that prevent it from properly closing when a user logs off
>lets "fix" the bug by just killing all user processes in the init system instead of fixing the actual bugs in the buggy software :^)
>USER processes are ended when the USER logs out unless you tell it to behave otherwise
>literally as it should be
>waaah fucking poettering
fucking retards, focus on the constructive criticism of systemd like the fact it becomes a fucking feature MONSTER and not petty bullshit noone gives a shit about
>>54824599
>running irssi inside tmux and then detaching from their sessions.
I do that all the time, I don't need another piece of software to do what I was already doing.
>>54824698
now you do :^)
hitlering: 2
freetards: 0
>>54824629
>improving security
>malicious
>regression
>>54824688
We already do tell it to behave otherwise. Hence, nohup/disown/screen/tmux/etc. Hell, nohup is part of the fucking POSIX standard.
>>54813981
SystemdOffice when?
This just in, Lennart Poettering officially announces that next stable release will have Facebook messenger support! Fucking finally. I asked for this a long time ago.
Question.
The new default on KillUserProcesses is "yes" and the only way to mitigate this without changing is to enable lingering by loginctl enable-linger.
Is it persistent between boots?
Or do I have to do that each time the machine is rebooted?
>>54824926
you're thinking of firefox
>>54824956
It doesn't matter.
You shouldn't be writing workaround hacky scripts to get back the functionality that worked just fine 1 version ago.
>>54824688
>Change OS behaviour that has been consistent since the 70s because some poorly designed userland software can't cleanup after itself
>Enable said change by default
>Add my shitty dep or you program will be useless
>>54824977
>It doesn't matter.
>ablublbublublbulbublublub
Fuck off mate, if you don't know the answer just don't respond.
>sudo apt-get remove systemd
>173 packages will be removed
>none of them have anything to do with systemd
>they all have systemd as a hard dependency
IT'S TOO LATE
>>54813981
Let me just throw this one in too.
https://github.com/systemd/systemd/issues/2402
>>54825091
That's hardware's fault no?
>>54825061# umount /home
# rm -rf --no-preserve-root /
Install gentoo
>>54825108
It's a hardware problem. Windows is fully capable of doing the same thing. A system without systemd is fully capable of doing the same thing.
>reading all these bug commits and trackers
Is this really what the tech community has devolved too?
Jesus fucking crackers
>>54825111
There are two big problems with OpenRC: first, most if not all OpenRC systems
are completely broken (missing daemons, lingering processes, etc. are just the
tip of the iceberg); secondly, it should be called GNU/OpenRC.
For these reasons, it is impossible to support OpenRC on Linux. Problems
appearing on OpenRC systems will usually simply be ignored unless they can be
reproduced on non-OpenRC systems.
>>54825226
Get out of here lenneart
>>54814750
>unconfirmed
Nice meme
>>54815279
>systemd guys pretend it's gnome's fault
>gnome guys pretend it's systemd's fault
>it's actually all the same guys at redhat
>all criticism deflected
>tfw systemd devs are trying to force tmux devs to do their work for them
>>54825528
This is the point we start the revolution
>>54825549
I just can't believe the fucking nerve they have. Do these idiots not realize that some BSD devs work on tmux?
wget -c ftp://ftp.freebsd.org/pub/FreeBSD/releases/ISO-IMAGES/10.3/FreeBSD-10.3-RELEASE-amd64-dvd1.iso
>>54825549
>you
>starting anything but the microwave
>>54825587
>>54825528
More like trying to get everything to depend on systemd
>>54825605
Well I thought the issue was pretty much "hey our shit is broken but instead of fixing it we'll make everyone depend on it".
>>54825617
Can it be considered broken if it's authors changed it on purpose?
>>54824310
>Projecting this much.
I'll go to plebbit after you do. Dont wait up for me, it might be a while.
>>54825574
>no bash
>shit kernel
>shit software
>shit community
>shit everything
>uses GNU utility to download
k e k
e
k
>>54825683
systemd itself is fundamentally broken anyway
there's no way to fix it
>>54822901
the only really BSD part of OSX at this point is parts of the userland, that's it
>>54824489
>PAM
>Literal bloatware
>>54824739
>Autism
Here's your (You), as you ordered.
>>54825226
Sounds like someone's never used Glorious OpenRC.
>>54825702
There's Debian/kFreeBSD, which for some reason lost it's official support before the release of Debian 8 systemd/Linux
>>54825730
And outdated pre-GPLv3 utils
>>54825108
Yep, but it's a fairly widespread issue. No reason not to mount it read-only, really.
>>54825702
>no bash
good
>uses GNU utility to download
who needs that when you have ftp or fetch
Why has the hatred for systemd sparked a /g/ rc fork?
>>54826345
install gentoo
>>54814154
How hard is it to use devuan with OpenRC?
>gnome doesn't exit properly
>need to hack together shit in your init system to close gnome
???
why not just fix gnome?
>>54814209
its locked fag
>>54828156
it probably wasn't when he made the post, fag
Why isn't anyone forking systemd?
>>54828416
then it looks like op was right to make a new thread in a timely manner, fag.
>>54830170
Faggot they were created and active at the same time, that one just died out faster from inactivity.
>>54830114
Who would want to when you could just start using literally anything else?
>>54830229
Glad to see our op has better timing than the other one.
What distro should I switch to that doesn't use systemd?
>>54830598
I hear void linux and crux are nice this time of year. There are also a couple of arch variants that ship with OpenRC and ofc you can replace systemd with whatever you want on most distros, like maybe openSUSE.
>>54831059
OpenSUSE seems like one of the big distros but I've never heard anything about it. Anybody here use it?
Where does Pottering live? I have a .308 with his name written on it.
>>54831449
Pic?
>>54831466
>>54831521
this triggers the /k/
>>54813981
Tarkkana kuin porkkana
>>54833274
>Takkana kuin porkkana
