Systemd now kills background processes after user logs out

response from devs "it's not a problem, just change these new obscure configuration options that no one will have heard about until it's too late!"

systemd devs continue to destroy linux from the inside outwards
inb4 claims of systemd hate bandwagon from idiots who ought to go back to windows

Other urls found in this thread:

bugs.debian.org/cgi-bin/bugreport.cgi?bug=825394
devuan.org/
bugs.freedesktop.org/show_bug.cgi?id=70593.
github.com/tmux/tmux/issues/428
voidlinux.eu/
github.com/systemd/systemd/blob/master/NEWS#L29
systemd-free.org/install.php
twitter.com/NSFWRedditVideo

forgot link
bugs.debian.org/cgi-bin/bugreport.cgi?bug=825394

Distro maintainers choose the defaults that suit their distros.

Why the fuck would you need a specific process running after logging out anyways? Just lock your account without logging out

Distro maintainers can choose to change the default behavior by shipping systemd with configuration which is not the systemd default. That does not change the fact that systemd kills background processes by default and the only way to change that is changing the configuration or getting systemd from 3rd party that did that for you.


Again, nothing wrong with using systemd if you like it, it's the they [people who use other inits] tend to be old people who don't like change, every normal person should be using systemd attitude of that cocksucker that I have issue with.

Linux isn't an iphone.

Systemd actually isn't a terrible desktop init. Problem is, it's developed as exactly that and than used on servers.

The default is important because lennartware is filled to the brim with obscure and badly documented options. I remember one issue with PulseAudio where the default configuration would interfere with other (application-specific) sound level control systems such that the audio level increased every restart. Getting randomly earraped was fun until the the distro maintainers found out that this option even existed and had this effect was fun. Come to think of it, that was actually kind of dangerous, ears are sensitive.


Not everything is your desktop.

That's what I get for quickly revising a sentence before sending.

I meant "Getting randomly earraped until the the distro maintainers found out that this option even existed was fun."

Do you know what databases are? Often I leave queries running in the background with nohup dumping the results of hundreds of thousands of registers on files for later reports. If this process is killed, I have to start over, and hours of work are wasted.

Wintoddlers, everyone.
>>>Holla Forums

Really? How is it important?
Do you compile it from git on a LFS system?
No? Then it's not important, and up to your maintainers.

Try reading the post if you want to find out the reason.

If your distro maintainers don't read the NEWS file than I'm sure you're using a shit distro.

Fagioli Fagioli give me the System32dickoli

This happened on Fedora and Debian, but I already know which smartass answer is going to follow. Don't waste your time.

What? This is about the systemd change
There's no reason for this thread to exist anyways, it's a duplicate systemd thread.
Just going to be filled with retards like you chimping out at a change that won't affect you.

Go back to Reading Comprehension 101, this is fucking embarrassing.

Doesn't nohup keep processes running? I thought all user processes were ended on logoff anyways.

Then take the advice in this post

devuan.org/

Not him, but logging in and logging out is fucked on Linux.

As far as I can remember, Switching users used to be implemented as a hack over logging out and logging in. Has that been fixed or will your processes be killed when you switch users?

Also, how does this affect users who SSH into a computer, run tmux, then detach and logout. Is their tmux session killed?

I'm not going to defend systemd, but hasn't it always been the case that logging out of a user session would kill processes tied to that user session?

If you switch user you are not technically logging out, in that case it would not make sense for the processes tied to that user session to be terminated.

The distinction is processes that ignore SIGHUP and tasks that don't.
This change, it won't even be a change because every distro will just keep the previous default, would end tasks that ignore SIGHUP after the session ends.

It's probably the slackware autist that was in the other systemd thread, that bitched and whined about how he was being persecuted and how systemd is a totalitarian conspiracy despite the fact that these things are configurable, and will be configured by your distro maintainers.

Simplistic.

The explanation about backgrounded tasks is overly simplistic, yes, it's not even getting into `huponexit` for BASH for example, and other trivialities.
Unless you're talking about default options from distros, in which case please elaborate.

yall niggers need runit

How is it totalitarian when it's configurable? I'll wait.

Shitty defaults make shitty software

So daemon is fucked?

That doesn't mean it's totalitarian.
Stop using words you don't know the meaning of.

No, nothing will change. Just depends on whether you install systemd from git or use a distro that doesn't configure their own defaults with compile time flags.

Have you read the OP? Holy shit you dumb faggot.

It's a bad idea for servers and a good idea for work stations. As long as the behavior can be changed, I don't see the problem.

How many times does it need to be said?
Linux is primarily a SERVER OS
Not every linux install is a desktop. So no, your desktop solutions do not work in many cases.

Wait why is this bad? If I log out with a process running it's because the process is fucked, killall and pkill aren't working, and I really want to kill that process. If I want to walk away from my computer while a process is running, I lock the screen.

Have you even read the thread?

Someone forgot bugs.freedesktop.org/show_bug.cgi?id=70593. They probably couldn't fix this bug so they just said "kill it, no more waiting".

that feature is part of X11 you moron.

Wait, I just realised that my torrent box (cubieboard 2) running Arch ARM will no longer allow me to ssh-tmux-rtorrent-exit.
Fuck this shit, I'll use net or freebsd.

Just wait till systemd starts trying to consume X11 too.

Hopefully it chokes and they both die.

Just write a systemd unit for it :^)

see

Or you can change the settings. Less work than replacing the operating system.

oh now it's just like everything else linux.

Some day, systemd will hit the point where configuring sane settings is more work than switching to an alternative operating system.

Think long term, user.

That day linux will be called "GNU/MicroApple iSystem32d"

ur waifu a shit

Bleeding edge was a mistake.

Not it's not, but it should be optional, like in Gentoo. And Arch isn't that bleeding edge.

Running linux already is more work than switching to an alternative operating system.
Windows wins again!

I'll say 4/10, because you seem to try.

Just change the settings. That's what makes linux so great isnt it? Paradox right here.

Solution: DON'T LOG OUT OF YOUR SYSTEM IF YOU WANT YOUR ACCOUNT'S TASKS TO KEEP RUNNING

Logging out by definition means kill that user's logged in session.

Jesus christ, for all these claims of superiority over windows users, linuxfags are pretty fucking stupid.

wew

Where did you get that definition?

Or they could just add the one line configuration change to logind.conf. The way people are complaining you would think the change is written in the source code with no configuration options possible without rewriting code and recompiling. Anything to bitch about muh systemd boogeyman and muh lennart boogeyman though.

Nice drawing, well memed my friend.

That's not a problem with systemd, that's a problem of dbus not terminating properly.

You're fucking retarded.

Why?

Another freedesktop abomination, basically. Also, reads the entire bug thread, it happens to people not even running X.

"Add systemd code to your program good citizen, it's your bug not ours"
github.com/tmux/tmux/issues/428

And?
dbus doesn't have to have X running.

You didn't read. dbus itself catches its SIGTERM, it's supposed to be one of its child.
tl;dr dbus, systemd, PAM, polkit and all that shit is an incredible mess that I'm happy to have left

spot the Holla Forums tard in that thread.

TRUUUUUUUUUUUUUUUU

When is system dicks gonna take over X and we can call it XD
who the fuck would use this shit ever its like you WANT backdoors

0/10

It's also a part of screens if you're a retard who only wants to run things through CLI.

Didn't see "SSH'd". Sorry m8o.

How about a problem with these DUBS not terminating properly?

"Hehehe, just add a dependency for our monolithic LennartOS behemoth, goy--I mean guy. Everyone else is doing it!"

...

I have great respect for the tmux guys for holding their ground.

Non-Unix Wizard here, what exactly is bad about background processes being killed when user logs out? and whats good about background processes as the way before? My technical info is awful - so please teach muh.
When will systemd fucking die? I've never seen anyone speak for it that isn't an obvious shill

it isn't bad. in fact this change increases security. the only reason it's "bad" is because it breaks functionality of a few applications which makes people butthurt that they have to make a 1 line configuration file change.

wew

this board is absolutely filled with incompetents

You're right, this board is filled with incompetents.

Yep sure is

The best part is how everyone here refuses to acknowledge how retarded they are

Breaking functionality just because you can isn't a good thing, especially if there is no practical gain whatsoever from this breakage.
Where on earth do you get the idea that the tmux devs are making a "1 line configuration file change"?
[citation needed]
Why don't I just shove reboot in your crontab "to increase security"? It makes about as much sense.


Hm, I guess I missed the compile time option which magically reduces the maintenance cost of #ifdef blocks to zero.


Okay, so can you tell us what problem in particular systemd is solving here which hasn't already been solved for decades by SIGHUP?
Have we all collectively thrown out the idea that kill -9 is a bad thing?

More importantly, you have no argument other than "trust them, they're professionals and you're not". Fuck you, I'm a full time linux sysadmin and I think the decision is bullshit. systemd is reinventing the wheel out of lack of understanding of the systems they work with, as they have done in the past and will continue to do for the foreseeable future.
As for who's "allowed" an opinion (lol), the difference between a professional and a hobbyist should be irrelevant, since both are affected by the decision. Don't shout "muh professionals" as if that makes their use of the operating system somehow more legitimate than anyone else's. It's not their operating system alone, and hobbyists count as users too.

Managing a file server from your moms basement to store anime on does not make you a ``full time linux sysadmin´´

No, but getting paid to do 8 hours a day plus on-call does.

...

...

This thread is getting awfully shilly. Hello Lennart.

The problem with systemd doing this is that they've changed default behavior. If they added the ability to terminate all user's running processes on logout, and left the option to turn it on with it defaulting to off, no one would really care. They also wouldn't care if systemd was using existing, standard protocols and software to make it happen.

This thing with tmux is typical embrace, extend, extinguish. Red Hat and Poettering are trying to force their code to be a part of tmux. It doesn't matter if it's optional or a compile time option. The goal is to have Red Hat and Poettering's code in a project that would normally not have anything to do with them.

Notice how their solution is to start bringing in even more Red Hat software to solve the problem they created?

People fucking hate it when they are forced to change. It's the same idea behind why desktop users were so pissed off about Windows 8.

Now tmux is in a position where their software won't work properly on systemd. And Red Hat's solution is to add more Red Hat software to tmux. It's a hostile way for Red Hat to gain influence over third party software they normally wouldn't have control over.

Red Hat wants control of GNU/Linux. They want it to become their corporate tool to compete better in professional markets. And they're going to keep doing these sorts of things now that systemd has reached the point where it's on enough systems that Red Hat can bully developers into supporting systemd.

Reminder that systemd doesn't have readahead anymore because "it's the CURRENT YEAR!!! Everyone has SSDs! GOSH."

You might also want to try
voidlinux.eu/

alias tmux="systemd-run --scope blah blah tmux"
wow that was hard

so the actual command seems to be
systemd-run --scope --user tmux

For a "sysadmin" you are a fucking retard. The configuration file change is to logind.conf to turn this new feature off in systemd if you have such problems with it, you absolute fucking mong. Do some research before spouting off shit you have no idea about. You know the shit you get paid to do as a "full time sysadmin".

Except it works fine with systemd if you make the change to turn this new feature off in logind.conf. Which is what distro maintainers will do if they don't feel this change is appropriate for their distro. Do you retards know anything about maintaining packages or distros?

That's not what is being asked of the tmux developers though. That is a workaround a sysadmin can make, or a packaging decision a distribution can make.
If the systemd guys had just said "recommend that people change some systemd setting when they package tmux", then that would be stupid but nothing to rave about. Instead they are recommending that changes are made to the tmux source code to accomodate their poor solution-in-search-of-a-problem, which is something to rave about.


Now have fun with that only working when launched from the shell, and even then only when called exactly as "tmux". That's not a solution.


This post covers a lot of my concerns nicely
Especially this. Why the fuck are they reinventing the wheel so much?


And then you have two different ways of doing things, and upstream projects need to support both to avoid fucking up on one setup or the other.
Not to mention, given the "gentle push" ideology at redhat, they're likely to deprecate and remove that option in the future once more software has succumbed to systemd integration.
And that option is still missing the point that they're being dumbasses and reimplementing something that already works, in a stupid and overly complex way.

Nope. Thank god there is somebody who actually reads the documents to know what configs to change in my system who is not me. There is good documentations for systemd and for this feature too after all?

I'm new to tmux so this might be a silly question... but when else would you launch tmux? Given that it's a terminal multiplexer, it seems logical that it'd be launched from a shell

ssh -t machine tmux
which, while it does technically launch in a shell, launches before aliases are processed, so won't do what you want.

Other times it won't fire:
thing=tmux$thing"tmux"/usr/bin/tmuxenv TERM=xterm tmux
Or anywhere in scripts you write to manipulate tmux. I've done this before as a sort of poor man's GNU parallel or cssh, letting me check session manually to see how it's getting on.

I've also set up keybinds before to open a terminal with tmux inside, but that was short-lived because it turned out to be a little impractical for me. Those would also not be affected by the alias.

Interesting, I never knew that.

I guess if you were determined to do something like then you could do
#!/bin/bashsystemd-run --scope --user /usr/bin/tmux
save the above as 'tmux' somewhere in your $PATH before the location of the real tmux.

That just seems way too hacky though, definitely not a thing that the user should have to do.

linux is a desktop OS, you use *BSD for servers

in other words Lennard fucks something and everyone else is the one who needs to fix it, how uncommon seriously when is this faggot going to die on AIDS, he must love some serious dicking just from the way he looks

DON'T LOG OUT IF YOU WANT PROGRAMS TO CONTINUE TO RUN
It's that simple.

Guys they already backed out the change.

They tried to silence their critics with the same tired "this is a bug tracker not a discussion forum" tactic.

This didn't work as the change they did was so obviously retarded that they were forced to listen to reason.

This has happened before when systemd thought it was a good idea to flood the kernel debug log with so much data that it froze the whole system.

That was also "not a bug, change your workflow" until Torvalds berated them into changing it.

It's amazing how much extra bullshit their putting sysadmins through just to create a better desktop init.

I've switched to SmartOS and FreeBSD for my webservers and routers.

Enjoy your pain.

My entire point was it should be off by default. You shouldn't be changing default behaviors like this. Maybe I should write a program for you systemd guys that changes your shortcut keys in all your programs each login. And when you bitch, I'll just tell you that you can change it back manually.

Changing defaults is completely fucking retarded. Imagine someone who doesn't know this change happened. They need to run tmux/screen/etc because they're sshed into a VPS or something and they need to shut down their main computer they're using to connect, because they're starting something that will take a long time.

Now imagine that user comes back, expecting their work to be completed, only to find that systemd silently shut down tmux and nothing got finished.

The fact that lennart is changing default behavior is absolute proof of his incompetence. This is Windows-tier level of faggotry and fucking over users.


Confirmed for never using Linux beyond ricing your desktop. Maybe you retards with Arch Linux who are ogling your low memory usage and your riced as fuck window managers don't give a shit, but the adults who use Linux to do actual work have every right to be pissed about this. Maybe when you get out of puberty and use Linux for actual work you will understand.

wew what a whole bunch of retardation in that post.

What do you expect from underage cancertards.

Throughout this entire thread people have had the chance to explain why or how this is useful. Nobody has yet even attempted.

This is a terrible idea. Not because there aren't casual users who might benefit from the system cleaning up after their messes, but because systemd, supposedly an init replacement, is doing this.

Why is systemd doing this? It has nothing to do with init functions. Are they just looking for features to add now? Let's put the window manager in there while we're at it.

When are you guys ever going to stop with this fucking meme?

from systemd's man page:

nice meme

How is breaking existing applications a useful change? What is the purpose of terminating all processes on logout? To be more like Windows?

ssh bby is okay

stallman still loves you, even if you're wrong


His point is that this won't necessarily break existing applications, because different distros will turn this option on or off based on their needs. Desktop-focused distros like Ubuntu will probably leave it on, but server-focused distros like RHEL are able to turn it off

And my point is why not just add the feature and let each distro turn it on if they see a reason. Instead they change the default and barge into incompatible projects and demand changes are made to those projects.

And I'm still waiting on a valid reason why programs should be terminated on logout. Can you not imagine a single use case where this is helpful?

I already stated that this breaks any sort of remote work you do when you want to run screen or tmux in the background and close your ssh connection.

I'd agree with you, I'm not trying to defend systemd. Nonetheless, has a valid point that it's not really that bad because distro maintainers will change this option to suit the distro's needs.

To avoid needless resource usage, obviously

Because only a leet haxor could install and operate Ubuntu.

DON'T LOG OUT IF YOU WANT PROGRAMS TO CONTINUE TO RUN
It's so fucking simple. For all the shit you niggers try to talk about HURR DURR EVERYONE THAT ISN'T LINUX PRO LIKE ME IS RETARDED, you have a lesser understanding of basic concepts than even the most basic of windows users.

Even retards that don't know the difference between their Excel document being fucked and corrupted and the online service that's supposed to run a powerpivot refresh know "log out = your shit isn't running anymore".

You are really dumb, for real.

I hope you're trolling.

So I have a server somewhere in The Cloud(TM). I ssh into the server and start a process. Maybe it's a web server, or a game server. Could be anything. After a couple of minutes, I decide that I'm done starting/configuring the server and I close the ssh session. This logs me out of the server.

Under the old way of doing things, this isn't an issue. As long as I started the webserver/gameserver process from tmux/screen, or any other method of doing it, then it will continue to run while I'm logged out. This is desired behaviour. With the new systemd change, even if I start the process with tmux/screen, the process gets killed when I close the ssh session and log out. This is not desired behaviour.

Not that user, but I fail to see why logging out is necessary, why is it so hard to just...not?

tis official
github.com/systemd/systemd/blob/master/NEWS#L29

The alternative is to keep the ssh connection open indefinitely, which means you can't reboot your local computer or disconnect from the network.

If your internet connection goes down for a second, the ssh connection is dropped, triggering a logout, killing your processes.

What said, and also because it's just practical. What's the point of keeping the SSH connection open if you're done interacting with the server? It seems silly for me to keep a terminal window constantly open on my desktop PC, not doing anything interactive, its only purpose to keep the SSH connection going. The whole point of a remote server is that it's remote - if it has to be constantly tied to by local machine, it defeats the purpose.

It's a process running under a uid that does not login and has no working password. You can't see why because it's outside of your experience. But the solution is for you to try out different ways of doing things, not to disable them.

systemd-free.org/install.php
It actually takes like 10 minutes, just fix your shit, goys.