Scheduled Maintenance: We are aware of an issue with Google, AOL, and Yahoo services as email providers which are blocking new registrations. We are trying to fix the issue and we have several internal and external support tickets in process to resolve the issue. Please see: viewtopic.php?t=158230

 

 

 

What should we do about systemd?

Here you can discuss every aspect of Debian. Note: not for support requests!
Post Reply

What should we do about systemd?

Poll ended at 2014-11-26 08:34

a) give up Debian to use another distribution which respects the *nix tradition
21
24%
b) concentrate on systemd's fork (uselessd) to port it to Debian
10
11%
c) use sysvinit (INIT) irrespective of its limitations with respect to modern software requirements
14
16%
d) use another initialisation system like runit
5
6%
e) accept systemd and continue using Debian
37
43%
 
Total votes: 87

Message
Author
User avatar
keithpeter
Posts: 502
Joined: 2009-06-14 08:06
Location: 5230n 0155w

Re: What should we do about systemd?

#136 Post by keithpeter »

thanatos_incarnate wrote:
keithpeter wrote: 2) I *think* it may be possible and technically desirable to deliver a 'just works' desktop or server a) without using abstraction upon abstraction and b) using core components with well defined interfaces that c) will work with other core components of differing versions thus giving distributions freedom to mix and match.
Forgive me, I'm still trying to wrap my head around this, but Isn't that the systemd approach?
A unification of things like all those daemons (hal, etc.) into the init system and sending those
calls directly to the kernel which has already evolved into being able to cover all those needs?
And it's also relatively modular, i.e. components can be enabled/disabled at compilation.
The systemd wider project calls for components to be lockstepped in each version and each version in future be tied to a specific version of the kernel. Upgrade one piece, upgrade the lot. My understanding is that the systemd api is also still under rapid development, so interfaces between components changing with each version with little in the way of a roadmap available.

I'm sure it will always work, and work well, within a given version, so Jessie release, and Jessie+N release will always work. May be implications for what you can put in a release though. And when you can make a release. See where that is going? Not intentional, just an inevitable side effect of the architechture.

twoflowers

Re: What should we do about systemd?

#137 Post by twoflowers »

LOL ... good point :mrgreen:
I can just speak for myself: It's a nuicance to get functionality that A provides wrapped in a layer of B and then be told that's cool. It's not, it simply bad design. Same goes for the "debig" kernelcommand. Hijacking functionality is not cool, it's stupid. And from there on a tree of consequences branches, including "breaks unix philosophy", "violates KISS" and probably "tries to cirvumvent GPL" are the easiest.
And "providing buildingblocks for an OS" is the worst of all: I already have an OS, what would i want virusd for? Playing LEGO with my OS?

User avatar
fleabus
Posts: 98
Joined: 2013-08-28 18:36
Location: Winchester, VA, USA

Re: What should we do about systemd?

#138 Post by fleabus »

[removed by fleabus, redundant]
Last edited by fleabus on 2014-11-24 23:23, edited 1 time in total.
Lenovo Legion 5-15IMH05 32GB Ram, 2TB Samsung 970 EVO Plus NVMe SSD

User avatar
buntunub
Posts: 591
Joined: 2011-02-11 05:23

Re: What should we do about systemd?

#139 Post by buntunub »

thanatos_incarnate wrote:But feel free to keep me in the dark, that way I will surely be convinced of your way. :lol:
Nobody convinced me one way or the other in deciding to fight against Debians decision to foist Systemd on the next stable release. I came to that decision after a long bout of searching the net and reading the Systemd docs, as well as sifting through all the various blogs and forum posts about it on a whole host of websites and Distro specific entries. This is how it works. You have A LOT of reading to catch up on!

In the beginning I was as incrudulous as anyone that a piece of software could get anyone up in arms. Education has taught me the why, and reading the tech news and email lists on Systemd keeps me informed. While it is easy enough to find outraged posts, it is harder to find educated threads that really discuss the matter deliberately and calmly. You will find the same common threads:

0. Systemd breaks POSIX and is incompatible with the BSDs. That is not good in regards to Debian, which claims to be the Universal OS, and runs counter to the open arms ideology Debian has always espoused.

1. Systemd has become monolithic. Too many dependancies create lockin. The code has become obfuscated and only a core cadre of devs really know whats going on. Sure, it is open source. Feel free to read through the thousands of lines of code to try to figure it out!

2. The most egregious is that one cannot run an alternate init system, such as SysV, OpenRC, or Upstart. So what? This runs counter to Debians Social Contract, which clearly states that the needs of its USERS come first. What if I don't want to run Systemd? A whole lot of folks don't want to run Systemd. So its tough luck for us! Use it or leave! Debian is an exclusive club now, or so the Debian Devs say on the mailing list, if you read it. But for Jessie, and probably only Jessie, alternate init may be possible thanks to a small handful of Devs who care about this stuff.

3. Systemd functions as a Windows-style svchost. Give a google on that to find out the obvious attack potential and exploitability of such a massive attack surface. Microsoft has been dealing with that for the last decade or more. Now, it's our turn!

4. The Systemd devs are "my way or F!ck Y$u"! Due to the dependancy lockin of a larger and larger core system component spread that are now tied to Systemd, nobody has a choice anymore. We all are now at the mercy of Lennart Poetering and his ilk. He has a growing number of major Distros all at his mercy. He has virtually all major upstream projects as a result by the throat. Why is that? As I said before, the fact that all major distros are locked into Systemd, and that Systemd has a huge number of dependant core functions, virtually every piece of software will have to tie-in to Systemd to work at all.

Above are just a tiny few reasons. I did not mention the UNIX model, because I am not that interested in it, although it does make a whole lot of sense. There are many more reasons. The only real way to find out which side you are on is to search and find out for yourself.

User avatar
Linadian
Posts: 490
Joined: 2013-12-20 15:25
Location: In a systemd free distro

Re: What should we do about systemd?

#140 Post by Linadian »

buntunub wrote:
thanatos_incarnate wrote:But feel free to keep me in the dark, that way I will surely be convinced of your way. :lol:
Nobody convinced me one way or the other in deciding to fight against Debians decision to foist Systemd on the next stable release. I came to that decision after a long bout of searching the net and reading the Systemd docs, as well as sifting through all the various blogs and forum posts about it on a whole host of websites and Distro specific entries. This is how it works. You have A LOT of reading to catch up on!

In the beginning I was as incrudulous as anyone that a piece of software could get anyone up in arms. Education has taught me the why, and reading the tech news and email lists on Systemd keeps me informed. While it is easy enough to find outraged posts, it is harder to find educated threads that really discuss the matter deliberately and calmly. You will find the same common threads:

0. Systemd breaks POSIX and is incompatible with the BSDs. That is not good in regards to Debian, which claims to be the Universal OS, and runs counter to the open arms ideology Debian has always espoused.

1. Systemd has become monolithic. Too many dependancies create lockin. The code has become obfuscated and only a core cadre of devs really know whats going on. Sure, it is open source. Feel free to read through the thousands of lines of code to try to figure it out!

2. The most egregious is that one cannot run an alternate init system, such as SysV, OpenRC, or Upstart. So what? This runs counter to Debians Social Contract, which clearly states that the needs of its USERS come first. What if I don't want to run Systemd? A whole lot of folks don't want to run Systemd. So its tough luck for us! Use it or leave! Debian is an exclusive club now, or so the Debian Devs say on the mailing list, if you read it. But for Jessie, and probably only Jessie, alternate init may be possible thanks to a small handful of Devs who care about this stuff.

3. Systemd functions as a Windows-style svchost. Give a google on that to find out the obvious attack potential and exploitability of such a massive attack surface. Microsoft has been dealing with that for the last decade or more. Now, it's our turn!

4. The Systemd devs are "my way or F!ck Y$u"! Due to the dependancy lockin of a larger and larger core system component spread that are now tied to Systemd, nobody has a choice anymore. We all are now at the mercy of Lennart Poetering and his ilk. He has a growing number of major Distros all at his mercy. He has virtually all major upstream projects as a result by the throat. Why is that? As I said before, the fact that all major distros are locked into Systemd, and that Systemd has a huge number of dependant core functions, virtually every piece of software will have to tie-in to Systemd to work at all.

Above are just a tiny few reasons. I did not mention the UNIX model, because I am not that interested in it, although it does make a whole lot of sense. There are many more reasons. The only real way to find out which side you are on is to search and find out for yourself.
*standing applause*, and lots of it. :!: 8) I already suggested Debdows as a new name for Debian, rather fitting. My personal fave is Debhat.

Edit: Has anybody seen this yet (General Resolution: init system coupling)? There may be hope afterall, I'm not holding my breath though, the Poetthat oligarchy is like a slime blob 'B' movie, smothering everything.
Linux Registered User 533946

kereberos
Posts: 67
Joined: 2011-05-23 09:12

Re: What should we do about systemd?

#141 Post by kereberos »

As long as there is another option to use for me personally C is the solution, on my own servers and workstations of course.
I administrate over 200 debian servers which range from version 3 to the latest wheezy. The clients couldn't care less about what type of linux, kernel or init system is running on all these but I'm glad to be working with Debians only not like RHEL, Suse, Centos and their kind of windows mutant systems running on linux kernels.

Although I love the phylosophy behind Slack and Gentoo, they are just way too time consuming to manage (even for personal use) and you will never encounter them in work anyway. If I need a samba server, sogo, tomcat, dns, apache just apt-get install em.


As RMS said open source is a mixture of captitalsit ideas and anarchist ideas. This is why regardless so many people like me are against systemd, none of us can stop this cancer from spreading in the unix community. Pottering should've been put to work in the mines in Africa for the rest of his life instead of causing more damage to the open source community.

As much as I would love if systemd would never happend, I pretty much just going to handle machines I encounter with systemd like the previously mentioned "clicking crap" OS-es which require you to be RHCE monkey certified for clicking. Log in there, do the job with the ugliest and laziest workaround you can and then log out and don't touch that system again.

Uselessd probably going to be the long term solution because even if we have a choice of the init system, more and more gnome apps will depend on systemd's libraries. Even now regardless that I use fluxbox (20mb), I have to endure that 1.2Gb of crap libraries pulled in from gnome just to have some gtk apps running.

Code: Select all

debi64 test # /etc/init.d/samba stop << Nothing happens ...
debi64 test # ps aux |grep smb
root      3004  0.0  0.3 272700  7772 ?        Ss   10:16   0:00 smbd -F
root      3006  0.0  0.1 272700  3240 ?        S    10:16   0:00 smbd -F
root      3023  0.0  0.0  12000   924 pts/4    S+   10:16   0:00 grep --colour=auto smb
debi64 test # killall -9 smbd
debi64 test # dmesg -T | tail -5
[Die Nov 18 08:54:47 2014] hda-intel: IRQ timing workaround is activated for card #0. Suggest a bigger bdl_pos_adj.
[Die Nov 18 10:16:22 2014] init: smbd main process (772) killed by KILL signal
[Die Nov 18 10:16:22 2014] init: smbd main process ended, respawning
A big middle finger for h. potter, but I'm sure he will never come to this forum because such a cretin cannot come from the Debian community, he must be a SuSe fan.

User avatar
keithpeter
Posts: 502
Joined: 2009-06-14 08:06
Location: 5230n 0155w

Re: What should we do about systemd?

#142 Post by keithpeter »

kereberos wrote:I administrate over 200 debian servers which range from version 3 to the latest wheezy. The clients couldn't care less about what type of linux, kernel or init system is running on all these...
Red Hat have probably worked out the same thing. Hence their strategy involving a more monolithic 'middleware' (the bit between the kernel and userland). As user mor might say, your clients are not really interested in the *freedom* of what they run, so they get what Red Hat, Oracle and Suse chuck over the wall.

http://www.microlinux.fr/mled.php

This chap does slackware based desktop/servers for small businesses in France. He posts quite often in the slackware forums.
kereberos wrote:Even now regardless that I use fluxbox (20mb), I have to endure that 1.2Gb of crap libraries pulled in from gnome just to have some gtk apps running.
Which gtk apps bring in what sounds like most of Gnome? So I can avoid them :twisted:. I use IceWM, XDM, see sig.

PS: the language used in the last few posts could be quoted by those in favour of the more radical parts of the systemd project and used to paint you all as tinfoil hat wearing old greybeards grumbling into your coffee. Think about it.


User avatar
Linadian
Posts: 490
Joined: 2013-12-20 15:25
Location: In a systemd free distro

Re: What should we do about systemd?

#144 Post by Linadian »

Linux Registered User 533946

tomazzi
Posts: 730
Joined: 2013-08-02 21:33

Re: What should we do about systemd?

#145 Post by tomazzi »

Linadian wrote: The ships are leaving the rats, lol. :shock: :wink: :lol:
I can't see anything funny in those links - for me, it's a sign of incoming disaster...
Odi profanum vulgus

User avatar
fleabus
Posts: 98
Joined: 2013-08-28 18:36
Location: Winchester, VA, USA

Re: What should we do about systemd?

#146 Post by fleabus »

Yes... No matter how I may personally feel, those folks who manage and maintain the project have a pretty rough road all the way around these days, and I can't help but feel for them all, and for the project as a whole.
Last edited by fleabus on 2014-11-18 22:27, edited 1 time in total.
Lenovo Legion 5-15IMH05 32GB Ram, 2TB Samsung 970 EVO Plus NVMe SSD

User avatar
golinux
Posts: 1579
Joined: 2010-12-09 00:56
Location: not a 'buntard!
Been thanked: 1 time

Re: What should we do about systemd?

#147 Post by golinux »

tomazzi wrote:
Linadian wrote: The ships are leaving the rats, lol. :shock: :wink: :lol:
I can't see anything funny in those links - for me, it's a sign of incoming disaster...
I prefer to think of it as a wakeup call.
May the FORK be with you!

User avatar
fleabus
Posts: 98
Joined: 2013-08-28 18:36
Location: Winchester, VA, USA

Re: What should we do about systemd?

#148 Post by fleabus »

[removed by fleabus]
Last edited by fleabus on 2014-11-24 23:40, edited 1 time in total.
Lenovo Legion 5-15IMH05 32GB Ram, 2TB Samsung 970 EVO Plus NVMe SSD

User avatar
edbarx
Posts: 5401
Joined: 2007-07-18 06:19
Location: 35° 50 N, 14 º 35 E
Been thanked: 2 times

Re: What should we do about systemd?

#149 Post by edbarx »

I have the bad feeling that the DDs who are leaving are those who support choice, and that would be bad.
Debian == { > 30, 000 packages }; Debian != systemd
The worst infection of all, is a false sense of security!
It is hard to get away from CLI tools.

User avatar
golinux
Posts: 1579
Joined: 2010-12-09 00:56
Location: not a 'buntard!
Been thanked: 1 time

Re: What should we do about systemd?

#150 Post by golinux »

Debiapocalypse
Countdown timer from debianfork IRC . . .
May the FORK be with you!

User avatar
fleabus
Posts: 98
Joined: 2013-08-28 18:36
Location: Winchester, VA, USA

Re: What should we do about systemd?

#151 Post by fleabus »

Well.. Soon we could all be running Lennux, along with three or four simultaneous malware scanners.... :wink:
Last edited by fleabus on 2014-11-19 13:14, edited 1 time in total.
Lenovo Legion 5-15IMH05 32GB Ram, 2TB Samsung 970 EVO Plus NVMe SSD

tomazzi
Posts: 730
Joined: 2013-08-02 21:33

Re: What should we do about systemd?

#152 Post by tomazzi »

fleabus wrote:Well.. Soon we could all be running Lennux, along with three or four simultaneous malware scanners.... :wink:
It's only my suspiction, which can be entirely wrong, but I'm going to say a few sentences here:
First, RedHat is fighting with Oracle for years. Oracle have started the war by offering support services for the (then exclusively RedHat) OS. RedHad responded in a very predictable way: "Hiding" patches and new features for Linux kernel...
That however showed up to be insufficient, because at some point, sooner or later, the changes would have to land as an "upstream" patches, which are easy to grab/take over.

And I think, that with systemd, RedHat is going to "take the leadership" of linux again, by enforcing it's own solution. Since all the systemd developers are paid RedHat employees, then they can take over the service market (in all of the aspects).

Unfortunately, new solutions needs testing, before they can be accepted by the customers - so they've tried Fedora as a testing ground, and Arch have joined as a volounteer. This however was not sufficient to convince serious customers which were running hundreds of servers. RedHat needed some really wide deplyment to prove that their solution is even worth attention.

How to do this? Buy Debian - the most widely deployed Linux distro...

So, in that aspect we are victims of the war between 2 corporations - and RedHat is going to win at least this single battle...

RedHat and Debian are both the most frequently used distros (on servers) - and the systemd code is crap (because corporations are leaded by bookkepers, not by engineers). This may lead to two kinds of disasters:
First: Oracle wins by prooving that RedHat is now unable to provide rock-solid solutions - and if this will happen, then we are all in the big crap.
Second: Oracle will give up and accept leadership of RedHat: we are all going to use a software resulting from a microshit-like development model, full of security holes.

systemd has to be abandoned or re-written from ground-up. We need reliable software, which is not created for or used as a weapon in inter-corporation war, which is commanded by bookkeeprs...

Regards.
Odi profanum vulgus

User avatar
fleabus
Posts: 98
Joined: 2013-08-28 18:36
Location: Winchester, VA, USA

Re: What should we do about systemd?

#153 Post by fleabus »

[removed by fleabus -- redundant]
Last edited by fleabus on 2014-11-24 23:09, edited 1 time in total.
Lenovo Legion 5-15IMH05 32GB Ram, 2TB Samsung 970 EVO Plus NVMe SSD

User avatar
Linadian
Posts: 490
Joined: 2013-12-20 15:25
Location: In a systemd free distro

Re: What should we do about systemd?

#154 Post by Linadian »

tomazzi wrote:
Linadian wrote: The ships are leaving the rats, lol. :shock: :wink: :lol:
I can't see anything funny in those links - for me, it's a sign of incoming disaster...
I don't find it funny either, it was my play on words joke that was supposed to be funny. Truth be told, this will turn my world upside down, no more dual SSD fakeraid Raid 0, Brother supplies a printer driver that works in Debian (the CUPS driver sucks, it stopped working properly many kernels ago) and I'll miss the huge repos. I was going to make a modest donation, but sys-choke-d kyboshed that. So now, if I can't get my printer working (properly, not buggy) in Salix, PCLinuxOS or PC-BSD, I'll either have to buy a new Linux compatible printer or be forced back to Windhose (that OS that vacuums money and your rights), I'd rather stick red hot torch heated nails in my eyes before going back to Windhose. The loss of Raid I can deal with, but a printer and good functionality, stability, are a must. Debian can shove spywared up their keester, I won't be a Debdows guinea pig. :evil:

Edit: Forgot one more thing, I was able to get my Via Technologies USB 3.0 controller working too, see this post. AMD's IOMMU and a proprietary USB 3.0 chip are no easy task in other distros, oddly enough, the USB 2.0 port mouse works and so does the 3.0 ports with IOMMU disabled (default) in i686 PCLinuxOS Trinity Live, strange that, same with the printer, when I switched over to 64-bit anything, it's been a bitch to get working. So again, all those fix hunts and troubleshooting FOR F#%KING NOTHING. :twisted:
Last edited by Linadian on 2014-11-19 03:23, edited 1 time in total.
Linux Registered User 533946

s2pido
Posts: 18
Joined: 2011-12-20 00:08
Has thanked: 1 time

Re: What should we do about systemd?

#155 Post by s2pido »

For desktop computing, debian isn't relevant. Regardless of distro, the entirety of Linux desktop computing isn't relevant.
Oracle vs RedHat might be meaningful (relevant) in the context of corporate workstations, but not in the context of PC purchasing end users.

Yes, printer support kinda sucks. In a corporate workplace, workstation user can (probably usually do) send their documents to networked print servers.
At home, you can chuck CUPS ('cept for the bit apparently needed to support PDF viewing) and print your documents from another PC.
Who sez (not me) that a "proper desktop PC" must have ability to print? Following that train of thought, who the HELL (not me) thinks it's a bright idea to distribute a "desktop o/s" preinstalled with ssh _server_ component? Hey, for extra "Duh!" ...let's preconfigure it with remote root login enabled. Seriously, lotta stuff preinstalled by Debian isn't sane (nor safe) for inclusion on a PC which will (only ever) serve as a desktop computer.

We users think of it as being "our" distro, but clearly it's not. We're facing a trend of "doacrats" repeatedly foisting ill-advised changes upon us, and NOT just at the o/s level. Refer to my earlier post to this forum (I haven't many) grousing about apps which are preinstalled, and /or preconfigured (within .deb package) to silently call out to various webservers. Naw, RedHat isn't "buying" Debian -- the annual revenue of Software in the Public Interest is (still) only about US 150k. PARTNERING, and bootlicking (yeah, and probably swag, and "gifts")... all things considered, RedHat is getting Debian in their pocket relatively cheaply. This corporatization, is it the inevitable product of human nature (ego, greed, politicking)?

Post Reply