sadly, a lot of hot air...
|
| Author | Content |
|---|---|
| mbaehrlxer Aug 14, 2014 8:49 AM EDT |
...and no actual substance in that rant.
i wish the author could explain what the actual problems with systemd are, instead of just venting. as it is i am forced to dismiss this posting because there is nothing i can learn from it. greetings, eMBee. |
| TxtEdMacs Aug 14, 2014 9:59 AM EDT |
You miss its inherent value of the exercise. First it raises the atmospheric pressure. That means clear days in the future. So some rant might clearly state the shortcomings of systemd, if and where they exist. Have hope, and sunny days. YBT |
| Bob_Robertson Aug 14, 2014 10:17 AM EDT |
I believe that the author's objection is the same as mine, given the information that I have: systemd uses compiled binaries rather than plain text scripts. If I am wrong about this, I will remove my objection to systemd. I'm not wedded to the spelling of the thing assigned PID 1, I am a rabid advocate of plain text rather than binary blobs, even if it makes it so my machine requires 15 seconds more to boot and the the job number of my first user application is in the 4,000s rather than 100s. Because, seriously, those two things have been held up to me as the benefits of systemd and binary boot files. |
| albinard Aug 14, 2014 11:43 AM EDT |
Waal, I'll tell ye, Sonny, it was a sad day when they put in them ee-lectric starters for cars. Why, in the old days ye could tell just how each and ever' cylinder was holdin' compression... I will probably hate systemd with a bloodthirsty passion until I get used to using it, after which it will vanish into invisibility like the fuel injection system in my car (What? You can't just tweak the idle? Terrible). I will confess that although I am of the generation born before even the Bletchley Bombe, and although in 1960 I briefly wrote scientific programs in Fortran, I have never tweaked the Linux kernel and have never felt a desire to do so. I learned some time ago that my ability to master the details of new processes dwindles with time, so I pick and choose the tasks I must master anew. No, I don't like being forced to master a new task just so I can continue to shape a distro my own way, but it sure beats hand-cranking a car and writing programs in Fortran! |
| GDStewart Aug 14, 2014 12:11 PM EDT |
The biggest problem I have with systemd was clearly stated in the "rant". It has inserted itself so deeply into or taken over functionality for many other important programs so that you no longer have any options, it is what you MUST use for all Linux systems. To me this is simply not acceptable. |
| Bob_Robertson Aug 14, 2014 1:32 PM EDT |
Albinard, I do not object to "new". I object to "shiny" and "squirrel". |
| notbob Aug 14, 2014 4:05 PM EDT |
> ...and no actual substance in that rant. Is there any substance to any rant? And should there be? I'm sure one can find specifics on systemd online, if one so wishes. No, this is a personal rant --as all rants are-- and reveals the feelings and reasons for disliking systemd and why the author thinks it is against all things *nix and the general openness of Linux. Myself, I have no desire to adopt it and am quite dismayed over the top distros having done so. Debian? Arch? Redhat? Say it isn't so! I'm no developer. I'm no code boy. It's difficult enough for me to safely edit plain txt script files, but at least, with a little effort, I can. If I wanted a Windows OS or an Apple OS, where everything is done for me, I'd have chosen one of them. I want total control. It's why I adopted Linux in the first place. Remove my control and you must insert "your" control. Been there, rejected that. As for Albinard's analogy of electric starters, it's perfect. Yes, crank starters made everyone an expert on the health of one's powerplant, but electric starters are what everyone now prefers. Looked at yer new car model's electric starter system, lately? Of course not! Gotta take it to the shop cuz the days of fixing it yerself are long gone. Can you customize how yer starting system works? Can you remove those event data recorders (EDR)? Can you shut off the vehicle's GPS device when you want to? I didn't think so. ;) |
| JaseP Aug 15, 2014 2:29 AM EDT |
Quoting:..., I do not object to "new". I object to "shiny" and "squirrel". It's not about "new," it's about control, as the author said. The second part of your comment is mildly insulting to those of us who have ADHD... The problem with systemd isn't the "pop" or "whiz-bang" trendy nature of it (shiny and squirrel, as you derogatorily refer to it),... The problem is that it takes away control. I too am under the impression that it will be all binary, with binary logs and with difficulty editing or altering it's "plan" for booting my machines. That means, if you have a quirky piece of hardware, and you need to tweak the boot process to make it work (and I have some), you're S.O.L. So, if you're in that camp, as I mentioned, you're stuck with an old distro with little or no support in a few years. Better hope the hardware gives out before then ... And for what??? Better boot times (by a few seconds,... and how that helps a server I don't know), standardization, and hope and pray that your hardware knows to power up and initialize all devices prior to the OS loading, because otherwise it won't be found (things an init tweak can easily accommodate). Systemd helps exactly who? I think I've even heard of Linus ranting about it. I think the author has it right ... It's the "CamelCaps" (read lazy, MS type thinking) crowd trying to take over. |
| Bob_Robertson Aug 15, 2014 8:51 AM EDT |
Jase, > It's not about "new," it's about control, as the author said. If you go back and look, you'll notice that it was I who was being accused of being against it because it was new. If you go back and look, you'll notice that my objection is about loss of control, just as the article's author said. Has been since I first learned of it. We are on the same side, here. |
| flufferbeer Aug 15, 2014 11:00 AM EDT |
@notbob >> Myself, I have no desire to adopt it and am quite dismayed over the top distros having done so. Debian? Arch? Redhat? Say it isn't so! Seems to me that developers with these and other top distros should (and will!) eventually bring some kid of workaround for systemd to make it VASTLY easier for sysadmins and end-users to appropriately gain back control of their systems!! 2c |
| Bob_Robertson Aug 15, 2014 11:26 AM EDT |
Fluffer, Fingers, toes, and eyes, crossed, in hope that you are correct. |
| notbob Aug 15, 2014 12:25 PM EDT |
> VASTLY easier for sysadmins and end-users to appropriately gain back control of their systems!! Wait! So we gotta endure this detour of making it harder, then hope upon hope some other ppl somehow graciously return it? Good luck with that. More than likely, the same ppl that have contributed to Linux being totally FOSS will revolt and create a totally new Linux init or at least a completely new alternative FOSS OS. Otherwise, I'll jes dump Linux for a BSD. |
| flufferbeer Aug 15, 2014 2:30 PM EDT |
@notbob, >> More than likely, the same ppl that have contributed to Linux being totally FOSS will revolt and create a totally new Linux init or at least a completely new alternative FOSS OS. I hope so too! Another good idear is for some ppl to write more good articles on how to easily revert systemdevil back to init for those like ourselves. And here's my hope against hope that maybe some NON top distros will manage to avoid systemd and put pressure on those toppies.... 2 more c's |
| me1010 Aug 15, 2014 7:31 PM EDT |
After reading the whole thing... systemd will definitely not be the end of the GNU/BSD/Linux system. Anyone tried to boot a Linux system without the kernel lately? I guess that all controlling binary blob is OK. But this other one over there --> not OK... hummm... Perhaps Stallman's GNU version would be acceptable to those who claim that systemd is a terrible idea... I mean Stallman really started the whole thing rolling... http://www.gnu.org/software/dmd/ Oops! It looks like systemd from a conceptual standpoint passes the Stallman test. That's good enough for me! |
| BernardSwiss Aug 15, 2014 7:37 PM EDT |
You seem to be confusing the question of whether Systemd is good engineering with the question of whether Systemd is good design with the question of whether Systemd is Free Software. Everyone agrees that Systemd is Free, Open Source Software. It's the other two questions that are problematic. |
| me1010 Aug 15, 2014 7:52 PM EDT |
@BernardSwiss I'm definitely not confused. I'm just not afraid of typing "journalctl -option" instead of "cat /log/ | grep 'regex' " I truly believe reasonable people will not have a problem with systemd. |
| BernardSwiss Aug 15, 2014 8:41 PM EDT |
BernardSwiss: "Blue is a colour, not a flavour, nor a shape." me1010: "But, I still think basketball is better than tennis." |
| me1010 Aug 15, 2014 8:53 PM EDT |
@BernardSwiss I've got two nickles that say "In ten years, there will be a new systemE which will have loads of naysayers who moan endlessly and pointlessly about how the world is going to end when systemd is replaced by systemE" |
| JaseP Aug 15, 2014 9:38 PM EDT |
Quoting: I truly believe reasonable people will not have a problem with systemd. I consider myself a reasonable person,... and I have a problem with Systemd. I'm not saying that the existing init system didn't need an overhaul,... But systemd strikes the "old guard" Linux users as going too far,... in the same way that KDE did with the release of 4.0, Gnome did with the release of 3.0, the Xorg people with eliminating manual configurations (at least without a lot of hassle) and the way Canonical did with Unity (regardless of version). Systemd is taking something that "ain't broke" (at least not by much) and trying to fix it (and in the process, making it completely SNAFU). It's another example of developers changing things for the sake of change, regardless of what the users want. Read the Wikipedia entry for a reference to how people (in the know) feel about the project... http://en.wikipedia.org/wiki/Systemd#Controversy |
| me1010 Aug 15, 2014 10:38 PM EDT |
@JaseP I don't disagree with the premise of not messing with what's not broken... and I've jumped the Gnome ship for Openbox to avoid the large overhead imposed. However, the generalized user base of a GNU/BSD/Linux system is larger and less technical than in prior years. And the code base is simply reflecting this change. The 'thing' still works. And it is still configurable to your heart's content in even the most minute ways... systemd does not change that... If you believe otherwise, you are not looking hard enough at what has been built and what possibilities for configuration exist. New systems and new methods will always be controversial - thus it has always been, and thus it will always be... However, I, for one, will continue to use FOSS as my desktop -- systemd or not... |
| flufferbeer Aug 15, 2014 10:51 PM EDT |
@JaseP, >> Systemd is taking something that "ain't broke" (at least not by much) and trying to fix it (and in the process, making it completely SNAFU). It's another example of developers changing things for the sake of change, regardless of what the users want. Read the Wikipedia entry for a reference to how people (in the know) feel about the project... http://en.wikipedia.org/wiki/Systemd#Controversy Double ++1 goes to you for the reminder "If it ain't broke..." and for the direct link (from the opinions of MANY reasonable people!!) 2 more c's |
| BernardSwiss Aug 16, 2014 12:24 AM EDT |
The thing is, the goals of Systemd are quite reasonable -- it's the design and implementation that people have a problem with. Couldn't it be argued that the fact, that systemd is designed to only work in Linux, but not other *nix's, is a very strong indication that it's design philosophy is misguided? |
| krisum Aug 18, 2014 1:44 AM EDT |
@Bob_Robertson > systemd uses compiled binaries rather than plain text scripts. While that is true for what systemd itself ships, but it can certainly launch shell scripts or any other executable. |
| Bob_Robertson Aug 18, 2014 8:43 AM EDT |
Thanks, Krisum. Indeed, that is obvious once stated. It remains to be seen if it will be used that way. |
| CFWhitman Aug 18, 2014 1:24 PM EDT |
One of my biggest problems with Systemd is the way the developers dismiss concerns about it rather than really addressing them. |
| jdixon Aug 18, 2014 2:19 PM EDT |
> One of my biggest problems with Systemd is the way the developers dismiss concerns about it rather than really addressing them. KDE4, Gnome3, Systemd, ... This does seem to be becoming a pattern, doesn't it. |
| mrider Aug 18, 2014 2:46 PM EDT |
What bugs me is that as soon as someone brings up an objection - ANY objection - the devs do a Jedi hand wave and say "you just hate change". If I tried that in my day job I'd be fired nearly instantly. |
| 750 Aug 20, 2014 10:53 PM EDT |
If systemd absorbes polkit, i suspect we have a RH funded attempt at building the Linux equivalent of Active Directory on our hands. Systemd has clearly gone beyond init and into centralized administration. For example there are plans for it to have the ability to protect configurations so that a reboot will reset any changes. Never mind that RH is the one Linux distro that many corporate computer suppliers offers support for alongside MS Windows. This clearly seen for instance with Lenovo, where UEFI only recognize a Windows or RH labeled boot. Label it Ubuntu or anything else and nope, not gonna boot that. |
| Koriel Aug 24, 2014 9:48 PM EDT |
Lets see Systemd brought to you by the makers of that fine piece of software called Pulseaudio (sarcasm). And you thought they were going to listen to anybody hahahahahahahha. Ive seen how PA turned my 3 linux boxes into nothing but sophisticated noise generators at least it helps with my tinnitus and I personally cant wait to find out what effect Systemd is going to have on them. For the moment im going to reserve judgement on systemd but the PA experience does not bode well for it. |
| Ridcully Aug 25, 2014 12:25 AM EDT |
If a person has to resort to filthy language to get a point across, he's lost my interest immediately and demeans his stance. PS.....no, I'm not a wowser on bad language.......but there are lines in the sand and I find it offensive personally, to cross them. |
| jezuch Aug 25, 2014 1:08 AM EDT |
I'm using systemd and I'm loving it. And there are lots of people like me. I thought I'd provide a counterbalance to this anti-systemd hatred in this thread. Thank you. |
| gus3 Aug 25, 2014 11:43 AM EDT |
@Ridcully, are you suggesting we consider "systemd" and "PulseAudio" to be profanities? [/sarc] |
| Koriel Aug 25, 2014 1:09 PM EDT |
@gus3 I think we can safely use PA as a profanity as it justly deserves it lets wait on Systemd maybe it will turn out great but not holding my breath on that one though. But the one thing I did know was that soon as I knew SystemD was coming from the PA guys i knew that they had no intention of listening to others. I believe in giving all things a chance to prove themselves PA had its chance and blew it IMO. |
| Ridcully Aug 25, 2014 6:24 PM EDT |
[/serious] @gus3.......Good heavens gus3, you are hurting my delicate sensibilities....Me ? Suggest "systemd" and "PulseAudio" to be profanities.......well, I might....if I knew what the darn things were.....I think Koriel might have saved me a proper response though. :-) [serious] PS.......Honestly, I think we may need to have words with TxtEdMacs......there's more of us using his technique of turning serious on and off......Is this the deliberate introduction of a type of user malware by TxtEd that infects the user rather than the computer ? Or very clever social engineering ? How does one get infected ? LOL. PPS........there is a Jim Lynch article linking to more about systemd in the queue.....I'll try to get it up as soon as I can. |
| TxtEdMacs Aug 25, 2014 7:24 PM EDT |
My Dearest Serious Mr./Dr./Professor Emeritus Ridiculous, Ah, you caught on ... you are no fun at all. Well some of the time you are. I will wait until this is forgotten and then spring another stealthy attack. As always, YBT |
| Ridcully Aug 25, 2014 11:09 PM EDT |
I've just had a look at Paul Venezia's article on InfoWorld......Without even knowing what systemd or SysVinit do, I can still understand the points he makes: First, that SysVinit wasn't broken and still functioned perfectly; second that it was easily "manipulated" so that it was easy and effective for a developer to use/apply when it was needed to work with other software; third (and this is my interpretation) the new binary blobs of systemd break the established rules of Unix/Linux which allow developers much fuller freedom with code (?). I'd like to know if I've got that right too.....I'm not a programmer but this sort of thing interests me, and if I am right, then I'd firmly fall into the group which suggests systemd is a very retrograde step. [deadly serious] @TxtEd......aharrhhh !!......so me hearty, a deliberate attempt to sabotage the peace and good will of this elderly gentleman. I'd better hire someone from the NSA to guard me 24/7 - or perhaps James Bond might still be around at MI5..... be warned. [unbelievably deadly serious]........ :-) |
| Bob_Robertson Aug 26, 2014 8:47 AM EDT |
Ridcully, Your First, Second, Third, fit exactly with my understanding of the situation. |
| 750 Aug 26, 2014 9:29 AM EDT |
Pretty much, Ridcully. But say so in "polite" company and you will be labeled a conspiracy nut before you know it. It don't have a problem with it existing, but i have a problem with it being a opt out rather than opt in issue. And if i opt out i have to pretty much return to 90s Linux across the board, because such things as udev are now part of Systemd. Take Consolekit vs Logind for instance. Yes, Consolekit has gone unmaintained. But it could be used with any init out there. Hell, it didn't depend on a init at all. It depended on Poliykit/polkit. But rather than pick up maintenance of Consolekit and allow it to optionally link up with Systemd or whatever, they instead replace it wholesale with Logind and make it part of the Systemd "package". Another is that they find that existing network managers are too slow. But rather than working on one of them to speed it up, they make a new one within the Systemd package. The prime example may be that Pottering says to toss/ignore the chapter on Posix in Programming Linux. In contrast, even tho he said it was borked, Torvalds left a Posix compliant API in there even while putting a Linux only improvement in alongside it. In essence the Systemd team is willfully or ignorantly burning bridges behind them, when leaving them standing would have gained them much goodwill. |
| me1010 Aug 27, 2014 8:46 AM EDT |
@Ridcully You've nailed the opposition argument. However, the opposition argument is wrong. 1. systemd works entirely with text file configuration, just like sysvinit. 2. systemd outputs text log lines, just like sysvinit. 2a. you can even direct the logging output to a particular tty console. 3. systemd is very easy to use, implement, and configure. 4. systemd enables better transparency between applications, daemons, and other services by creating a uniform platform for configuration items and logging. This is really useful for servers, especially. For example, it is no longer necessary to have separate parsing scripts to find dated information in separate log files. All information is uniform in format, which means parsing the data is /much/ easier. 5. systemd has several experimental components which may be very useful for easily creating secured chrooted type 'jails' for virtual environments -- this is extremely useful for servers as well as mobile devices. The Infoworld article mentioned contains a massive amount of misleading and incorrect information. The most blatant misleading part is that the Internet is text. There are very very few websites that are only html. Any actual interactivity over the web requires the use of programming on the server. And whether the language chosen is an interpreted language like php, ruby, or perl ... or a complied language like C... doesn't matter. The author's basic statement is incorrect. The web runs largely on programming whose source code is text... but that's a meaningless statement, since /all/ programming has text source code... Furthermore, unless a website is using insecure connections, SSL will send binary data over the connection - not text data... The way I see the systemd versus sysvinit argument is: Those who don't wish to learn something different, and those who see the advantage of systemd due to its uniform configuration across GNU/Linux distros, its inherently better data security through binary storage system, its ability to mesh very well with desired server functionality currently found only in paid software or difficult to configure FOSS visualization, its easy to configure services configuration files which can be understood clearly even by non-programmers, as well as many other benefits too long to list... |
| Ridcully Aug 27, 2014 9:15 AM EDT |
@me1010.......I really cannot comment any more than I have but thanks for the confirmation that I did at least understand the points as made by the article. I suffer from the disadvantage of not being a programmer/developer actively involved in using the software. |
| jdixon Aug 27, 2014 9:43 AM EDT |
> Those who don't wish to learn something different, and those who see the advantage of systemd due to its uniform configuration across GNU/Linux distros... The configuration isn't going to be uniform across the different distros. > ...its inherently better data security through binary storage system... Binary storage is not inherently more secure. > ...its ability to mesh very well with desired server functionality currently found only in paid software or difficult to configure FOSS visualization... Which undoubtedly benefits corporations which use Linux greatly. Desktop users? Not so much so. > ...its easy to configure services configuration files which can be understood clearly even by non-programmers, As if init scripts can't be easily understood. > ...as well as many other benefits too long to list... Well, maybe they'd be more convincing, but I doubt it. |
| flufferbeer Aug 27, 2014 10:24 AM EDT |
@you1010,
>> Those who don't wish to learn something different, and those who see the advantage of systemd due to its uniform configuration across GNU/Linux distros, its inherently better data security through binary storage system, its ability to mesh very well with desired server functionality currently found only in paid software or difficult to configure FOSS visualization, its easy to configure services configuration files which can be understood clearly even by non-programmers, as well as many other benefits too long to list... Start off why don't you by re-doing the old Blame We Hapless Victims of systemdevil gambit (just as above!), then continuing with assertions as tho YOU'RE actually one of those secluded systemd devs :( +yet 2 MORE c's |
| notbob Aug 27, 2014 12:40 PM EDT |
As a longtime Linux user, I keep looking for a choice. I'm not seeing one. What I'm seeing is, a buncha developers attempting to create their own standard and imposing it on everyone, regardless of what the user may prefer. Bottom line: I don't like it! |
| Bob_Robertson Aug 27, 2014 1:25 PM EDT |
> its inherently better data security through binary storage system Ah, so it is not using text, just as all the accusations say it's not using text. This puts your items 1, 2, 3, and 4, into question. > Those who don't wish to learn something different Ad-hominem. Address the issues raised, not the character of the questioner. |
| notbob Aug 27, 2014 2:21 PM EDT |
> Ah, so it is not using text, just as all the accusations say it's not using text. Pretty much the bottom line for this non-developer. Not every Linux user has a dev looking over his/her shoulder to provide guidance. How is a binary-based system gonna benefit me, a 13 yr "intermediate" user of Linux, on the "desktop"? |
| me1010 Aug 27, 2014 2:43 PM EDT |
@Bob_Robertson
and
@notbob good luck running your non-binary, non-compiled collection of source code. I'm sure the kernel will work just fine without all that binary compiled code [oh the horror of compiled code!] as will of course init, pam, et al. |
| jdixon Aug 27, 2014 3:00 PM EDT |
> ...good luck running your non-binary, non-compiled collection of source code... Thanks for demonstrating the honesty and integrity of your position so thoroughly. I was thinking I should give the systemd devs the benefit of a doubt until I saw the results, but I'd say your comments (which, from what I've seen, seem to be representative of the those of the systemd developers, just as they were with the KDE4 and Gnome3 developers before them) have made that completely unnecessary. I'm thinking that if Red Hat does succeed in pushing systemd, it may be time for a full fork. And if that doesn't happen, well, there's always the BSD's. |
| gus3 Aug 27, 2014 4:01 PM EDT |
@notbob, after 13 years of using Linux, your knowledge of the system probably runs deeper than you realize. Okay, so I finally broke down and did some research on systemd. AFAICT, it has one major point in its favor: using cgroups instead of PID's for management. Beyond that, its supposed "advantages" accomplish nothing (that I've found) that can't also be done in SysV or BSD init. Parallel execution? Check. Logging? Check. Snapshot/restore? Already have tools for that. But the one stated goal I found laughably ignorant was the intention to "reduce the computational overhead of the shell" (from the Wikipedia article). This effort is nullified by "implements an elaborate transactional dependency-based service control logic" (from the systemd homepage). This is nothing more than replacing one computationally complex system with another. If you don't understand what I just said, read it again, and again, until you do. There are real minimum levels of energy required to accomplish any calculation. Even 1+1=2 requires a certain energy expenditure in the human brain. Booting a Linux system to a particular configuration is no different. One goal which seems to be guiding some systemd tool development is "exposing more knobs and gauges for the sysadmin," but sadly, I don't see this with other stated goals. My conclusion on the whole affair: Efforts invested in systemd are merely attempts to "fix" what isn't broken. |
| me1010 Aug 27, 2014 4:16 PM EDT |
@all Read this post: http://techspear.com/blogs/2011/11/systemd-journal-an-altern... You will find it much more enlightening than a rant around 'This is just not what I want." sysvinit is broken. systemd may also be broken... however, it is undeniable that sysvinit is, in fact, broken. |
| jdixon Aug 27, 2014 4:36 PM EDT |
> sysvinit is broken Not on my system it isn't. It may be broken from the perspective of multi-billion dollar companies that want to use Linux on servers, but that isn't my concern. They can develop and run whatever they want. Forcing it on everyone else, otoh... |
| JaseP Aug 27, 2014 4:48 PM EDT |
@ me1010: Couple of major issues,... First, many distros do not use sysvinit as opposed to Upstart or an Upstart derivative. Second,... Most of the core issues cited as problems with sysvinit could be fixed without changing the system completely. Many others are just plain opinion, such as the O(n) complexity of log files... O(n) complexity in log files is not an issue for searching them with anything approaching a simple text editor with a search function. We are not talking about multi-gigabyte files here... Log files are what? Maybe a few megabytes?!?! There are cron jobs that limit the sizes of the files, and can back them up for later searching.. AND,...Doing a search on one,... hell, even manually skimming one,... wouldn't take more than a few seconds with a search tool, to probably under an hour manually (which is what many recommend for tackling problems, anyway)... Log files are serial in nature after all... Plus it wouldn't take much (of a patch) to output the log files as an indexed binary tree data structure which would drop search complexity to O(log n)... IMHO, that's a case of "ain't broke." It also still doesn't address the fact that the systemd developers have expanded the scope of their "mission." That is one of the core complaints of the "people in the know" ... ya know,... like Linus Torvalds,... |
| gus3 Aug 27, 2014 4:56 PM EDT |
"Forcing it on everyone else" ain't gonna happen in Linux-land. Wasn't it just a few weeks ago that we saw a spate of articles asking if there was too much choice offered by FOSS? I think we can now answer that with a definitive "NO!". And if SysV init were broken, we would have known it over a decade ago. Same with BSD init. The assertion that they're "broken" is just that: an assertion. |
| CFWhitman Aug 27, 2014 5:01 PM EDT |
I'm not going to try and claim that there could be no improvements to SysV init, BSD init, or even the much newer Upstart. However, anything so completely non-modular and non-portable as Systemd is at least going to give me pause when I see it on a Unix-like system. If you are going to depart from what made Unix so great in the first place, then you'd better be prepared to explain exactly why it can't be modular and it can't be portable or it won't work. I don't care about all the great benefits of it until it's explained clearly why it couldn't manifest those benefits without making it non-modular and non-portable. Tell me that there are many shortcomings to current init systems, and I'm prepared to believe that. What I want to know is whether Systemd is really the best approach to fixing them, or something else could be created that fixes them without tossing out the Unix principles of the last forty years. |
| JaseP Aug 27, 2014 5:11 PM EDT |
Quoting:Tell me that there are many shortcomings to current init systems, and I'm prepared to believe that. What I want to know is whether Systemd is really the best approach to fixing them, or something else could be created that fixes them without tossing out the Unix principles of the last forty years. Amen! ...(emphasis supplied in quote) |
| gus3 Aug 27, 2014 5:18 PM EDT |
+1 CFWhitman, and +1 again for saying what I wasn't able to gel in my mind. |
| mbaehrlxer Aug 28, 2014 12:56 AM EDT |
here is a nice summary of the advantages of systemd, which also refutes some of the points made against it:
https://bbs.archlinux.org/viewtopic.php?pid=1149530#p1149530 and a response to most complaints against it: http://0pointer.de/blog/projects/the-biggest-myths.html i am still on the fence, but the more i read about systemd, the more i like it. greetings, eMBee. |
| Bob_Robertson Aug 28, 2014 8:56 AM EDT |
Agreed, CF. Well said. "without tossing out the Unix principles of the last forty years" That's not fear of change. That's a recognition that those principles exist for a reason, and it is those principles which have made F/OSS work. |
| BernardSwiss Aug 28, 2014 1:44 PM EDT |
Quoting: Agreed, CF. Well said. Seconded. (Thirded?) |
| gus3 Aug 28, 2014 6:03 PM EDT |
"Conservative: a statesman enamored of existing evils, as opposed to the Liberal, who wishes to replace them with new ones." -- A. Bierce (Edit: added link) |
| DrGeoffrey Aug 28, 2014 6:37 PM EDT |
Ambrose certainly nailed those definitions. |
| 750 Aug 29, 2014 3:36 AM EDT |
The Arch forum posting seems to have a bunch of entries that make marginal sense outside of the server room, and one or two (hotplug for instance) that makes me go "why is Systemd even dealing with that?!". |
| jdixon Aug 29, 2014 7:19 AM EDT |
> The Arch forum posting seems to have a bunch of entries that make marginal sense outside of the server room That does seem to be the problem. Systemd makes a lot of sense for server farms run by large corporations. For desktop systems or one off mom and pop operations? Not so much so. |
| me1010 Aug 29, 2014 7:29 AM EDT |
@gus3... sysvinit /was/ known to be broken more than a decade ago. Replacements for sysvinit, including dependency type replacements are not new. It's just that most people don't have enough time to know everything about everything all the time. So, while sysvinit is and was broken - and this has been a known 'thing' for a long time [probably a few months after it first materialized] - the generalized user doesn't know that sysvinit was broken... For your reading pleasure... note the year [2002] ... this doc is 14 years old... and there are many others like it and older too... https://www.kernel.org/doc/ols/2002/ols2002-pages-176-182.pd... init systems are always contentious... ha! that's a joke you know... anyway... sysvinit vs systemd -- systemd is definitely not going to sink the *nix ship. And it's definitely much much better than sysvinit. Read the docs... after your eyes go blurry -- maybe you'll say, 'Hey! Here's something I always wished I could do!' http://www.freedesktop.org/wiki/Software/systemd/ |
| jdixon Aug 29, 2014 11:41 AM EDT |
> So, while sysvinit is and was broken - and this has been a known 'thing' for a long time [probably a few months after it first materialized] All software is broken. Systemd will be just as broken as sysvinit is currently. Just in different ways. |
| Bob_Robertson Aug 29, 2014 11:42 AM EDT |
> Systemd will be just as broken as sysvinit is currently. Just in different ways. Exactly. Which is why sticking to the principles of openness and readability are important. |
| gus3 Aug 29, 2014 1:48 PM EDT |
@me1010, that paper describes the default Gentoo boot manager (or at least did back then). Been there, done that. Again, it's merely shifting the complexity to a different location. His chief complaint about SysV seems to be how much more complex the interplay is among scripts, not about the implementation of the system itself. Big whoop. Symlinks in /etc/rc?.d/ are explicitly numbered to show their execution order, a point he admits earlier. His complaint is based on personal taste, not on any actual deficiencies in the system. And this might be a shocker to Mr. Gooch: Slackware's BSD-based init has (and had, even back then) explicit support for modular SysV init scripts. Who'da thought that was possible? I've been a sysadmin, on multiple Linux and Unix platforms. I've also created an installer CD. The #1 headache to making an installation CD work was getting past pivot_root. It had nothing to do with any alleged deficiencies of BSD or SysV initialization. "Do one thing, and do it well." The sole exception to that in *nix history has been the X Window System. Every alternative that springs up intends to address at least two deficiencies in X, one of them always being the modular complex brought on by time and hardware advances. No such cruft build-up has occurred with either BSD or SysV. |
| gus3 Aug 29, 2014 2:06 PM EDT |
And another point: SysV was "good enough" for proprietary Unix makers (Sun Microsystems, Hewlett-Packard, IBM that I know of). If SysV were really as broken as so may systemd fans are asserting, would not IBM have proffered an alternative a long, long time ago? Remember, IBM's #1 rule of programming is, "Correctness first, speed second." Their endorsement and use of SysV carries far more weight with me than Poettering's rejection of it. |
| me1010 Aug 29, 2014 2:55 PM EDT |
@gus3... You do know that you can run systemd as something other then PID 1... Therefore passing all systemd input to sysvinit as well as satisfying nearly all dependency issues on systemd -- without actually running it? Yes? Of course, the above scenario may seem pointless to some people... but there is definitely not a requirement to use systemd and only systemd - and most, if not all, of systemd's purported 'scope creep' is optional. The installation size of systemd and other its other components will barely be noticeable on any desktop system... But, I suppose there are purists who attempt to scourge their multiple 3TB drives of the last few kilobytes of unnecessarily occupied magnetic field. Anyway - I've had a hard day dealing with work and kids --- good luck holding on to your sanity... me1010 - out. |
| Bob_Robertson Aug 29, 2014 3:16 PM EDT |
> good luck holding on to your sanity As much as I like a good personally insulting rant, as many here will testify, why not address the technical issues of this technical issue? |
| jdixon Aug 29, 2014 3:34 PM EDT |
> ...why not address the technical issues of this technical issue? The answer to that seems obvious, at least to me. |
| BernardSwiss Aug 29, 2014 5:20 PM EDT |
There appears to be two sets of issues: 1) the actual "technical" issues, about how to best accomplish the technical goals, how important it is to maintain classic UNIX design principles, etc, 2) the "chokepoint" / development-nexus issues, where the small, insulated group developing Systemd, and Gnome and what-not else as well, is proving to have enough real-world power to pressure or effectively force the rest of the Linux developer world to follow along, and convert to using systemd as the standard init system for their own Linux distributions, regardless of whether they approve or disapprove of the changes. They are both significant issues -- but people would be much less troubled by the first, if they weren't having to cope with the second, as well. |
| mbaehrlxer Aug 29, 2014 11:04 PM EDT |
750 wrote:The Arch forum posting seems to have a bunch of entries that make marginal sense outside of the server roomi have to admit that i treat every computer i have access to as a server. even my laptops. one is sitting in another corner of my home but i use it from my workstation to read email. the workstation in turn serves files and DNS for the whole house. half way across the globe my grandmothers workstation does backup duties. so you can see i am all about servers and thus maybe i just don't see how systemd really mainly just benefits the server room. please elaborate. also, (asking everyone) in particular i don't get this: what are the drawbacks of systemd? so maybe the benefits are marginal for some, but just because systemd doesn't improve things for everyone, that shouldn't mean that everybody else should forgo those benefits. the complaints about modularity, configurability, text-based config and log files, all of which are issues for the server room too, btw, have been refuted. what remains is issues like learning a new system, feature creep, change for changes sake, NIH syndrome, don't fix what's not broken, devs ignoring users, etc, all of which are very subjective (new learning perhaps less so) and thus can't be answered objectively. i just went through the whole thread again and i could not find a single complaint where systemd actually breaks something or removes functionality that is available now. i am not claiming that this shows that no problems exist, i am actually sure they do exist, no major change like this can go by without causing problems somewhere. but i am still waiting to see any such problem with systemd. until then it is all huff and puff... greetings, eMBee. |
| tuxchick Aug 30, 2014 9:12 PM EDT |
This is a good summary of the problems with systemd:
http://www.infoworld.com/d/data-center/systemd-harbinger-of-... Quoting: systemd is growing, like wildfire, well outside the bounds of enhancing the Linux boot experience. systemd wants to control most, if not all, of the fundamental functional aspects of a Linux system -- from authentication to mounting shares to network configuration to syslog to cron. It wants to do so as essentially a monolithic entity that obscures what's happening behind the scenes. |
| Ridcully Aug 30, 2014 11:26 PM EDT |
@tuxchick....thankyou for that excellent link. I read the last paragraph, again, and again, and again..........and thought about KDE and semantic desktop. There is an eerie resemblance. |
| mbaehrlxer Aug 31, 2014 1:38 PM EDT |
and i read the same link and learned nothing. more complaining without references or proofs. again, i still don't know what's wrong. i'd like to see an actual case of a problem being obscured by systemd. examples please... greetings, eMBee. |
| vainrveenr Aug 31, 2014 2:10 PM EDT |
Quoting:and i read the same link and learned nothing. more complaining without references or proofs. again, i still don't know what's wrong. i'd like to see an actual case of a problem being obscured by systemd. Disregarding possible Straw man baiting, if "i still don't know what's wrong" is truly the case, then perhaps several of the comments following the more recent piece by the same author Choose your side on the Linux divide will be most enlightening, especially those comments that successfully manage to point out the specific drawbacks of systemd. |
| jdixon Aug 31, 2014 3:54 PM EDT |
> ...and i read the same link and learned nothing. more complaining without references or proofs. Well, if at least half a dozen restatements of "It violates the basic design principles of Unix and Linux" doesn't get the idea across, I don't know what will. |
| mbaehrlxer Aug 31, 2014 11:37 PM EDT |
vainrveenr: thank you for the link. i am reading through the comments now. unfortunately i have a hard time finding comments that satisfy my expectations. so far i found one comment about switching away from systemd because they found attack vectors. but no specifics on what these vectors are, so i can't decide whether they affect me. i find more complaints about violating unix philosophy, but no example, some comments about how little systemd achieves with so much code, there are some examples of how some goals of systemd could be achieved by improving existing applications. ok, that's nice. but while we are talking about unix philosophy i could not see a rule that disallows rewriting a utility from scratch. so, i am sorry, but i am not satisfied. maybe you could point me to one comment that you believe should answer my questions. it could be that i missed something. there was a lot to read and i may have skimmed a few comments to briefly. jdixon: i am looking for an actual example of a real problem. or a demonstration of how systemd violates basic design principles unix. stating that it does, is not enough. show me one program that is part of systemd and explain how it violates unix philosophy. the only contender i am aware of could be journaling with its binary logging. but since it is also able to log to syslog, the problem is reduced to it being an intermediary, instead of letting applications log to syslog directly. is there another one? greetings, eMBee. |
| mbaehrlxer Sep 01, 2014 12:04 AM EDT |
how about going through the points mentioned in Choose your side on the Linux divide and analyzing how systemd violates them:Quoting:1. Small is beautiful.systemd is more than 60 individual binaries which all do different things. we'd have to look at each one of them to decide if they are small enough. Quoting:2. Make each program do one thing well.which of the more than 60 binaries does more than one thing that could better be done by separate programs. (edit:) or are there cases where multiple binaries depend on each other to do one thing together, which would be better done if it was one combined program? Quoting:3. Build a prototype as soon as possible.not sure if this was violated. but since the code is released it is no longer relevant. Quoting:4. Choose portability over efficiency.here we have a candidate. systemd is not portable. though i don't know if that relates to efficiency. as far as i understand it is not portable because it depends on other non-ported tools or libraries. making it portable would require to replace their functionality and potentially violate point 2. Quoting:5. Store data in flat text files.journaling fits this point. but it can send logs to syslog which produces text files. as mentioned above, the only problem here is that it adds an additional intermediate program, not that it denies us access to flat text logs. Quoting:6. Use software leverage to your advantage.i don't know how that should apply. Quoting:7. Use shell scripts to increase leverage and portability.the question here is where to draw the line. this point is highly debatable. Quoting:8. Avoid captive user interfaces.i don't think this is violated. Quoting:9. Make every program a filter.no clue here. but i don't think this really means everything must be a filter, unless you want to restrict all your editing to sed. so the question again is: where to draw the line? what should be a filter, and what shouldn't be? greetings, eMBee. |
| 750 Sep 01, 2014 4:33 AM EDT |
@ridcully: With KDE it is optional. If you don't like it, don't use KDE. But with the tentacles Systemd keeps sprouting, soon enough you will be running Systemd if you are running the Linux kernel... |
| Bob_Robertson Sep 02, 2014 8:41 AM EDT |
I don't think 60 binaries is in any way "small". |
| JaseP Sep 02, 2014 10:21 AM EDT |
Quoting: @ridcully: With KDE it is optional. If you don't like it, don't use KDE. With KDE, you can (by and large) turn off (most of) the semantic desktop features, or, as you say, don't use it. And as you say,... the problem with systemd is that the developers are swallowing up additional stuff,... not just the init process, but also mount, cron, power management... the list keeps growing. The bloat problem isn't size or number of files it takes to run,... it's in the way it has become viral, trying to take domain over other services. Pretty soon the OS could be described as GNU/systemd (featuring Linux) instead of (the politically correct) GNU/Linux. |
| mbaehrlxer Sep 02, 2014 11:53 AM EDT |
Bob_Robertson: depends on what you expect the 60 binaries to do. if all that systemd was doing is to replace init, then 60 binaries would be ridiculous. but that's not the case, so before we can judge that we need to look at the purpose of each of those binaries. i do expect that each binary provides in independent function that does something useful, not only for systemd itself. (i remember to have read somewhere that some of the things that come with systemd are useful even without it) compare that to fish vs bash for example. bash provides a lot of built-in functions. fish instead comes with dozens of separate executables and has only a few functions built in, namely those that wouldn't work if they were running as a separate process, like cd. now which one is closer to unix philosophy of keeping things small and only doing one thing? fish, i would think. the next question is whether it makes sense for all these programs to be delivered together with systemd. i can't tell, because i haven't yet looked at what they are. but as they are developed by the same team, and probably have related functionality, i don't see why not. a suitable comparison might be the coreutils package. it comes with more than 100 executables. does that make coreutils violate unix principles? would you rather have each of those 100 programs distributed separately? would you expect each of them to be developed and maintained by a different person or group? i think not. so just because systemd is aggregating a lot of functionality in separate programs, does not seem a violation of keeping things small. greetings, eMBee. |
| flufferbeer Sep 02, 2014 12:40 PM EDT |
@750, @JaseP >> But with the tentacles Systemd keeps sprouting, soon enough you will be running Systemd if you are running the Linux kernel... >> And as you say,... the problem with systemd is that the developers are swallowing up additional stuff,... not just the init process, but also mount, cron, power management... the list keeps growing. The bloat problem isn't size or number of files it takes to run,... it's in the way it has become viral, trying to take domain over other services. Pretty soon the OS could be described as GNU/systemd (featuring Linux) instead of (the politically correct) GNU/Linux. ++2 for both of you! Do any of you realize that there's also a boycott systemd site that shows what you describe ACTUALLY HAPPENING??? No hot air here at all, really!! Just do an online search, visit the site, and you'll definitely see what I mean. -fb |
| gus3 Sep 02, 2014 12:52 PM EDT |
eMBee, portability is specifically rejected in systemd development. It is using Linux-specific interfaces for its work. EDIT: yup, just like it says on that boycott page. |
| Ridcully Sep 02, 2014 5:18 PM EDT |
Gus3......thankyou for the link.......I've just read it and I am both staggered and dismayed. I think it is time Linus bought into this, if he hasn't already. And I take your point JaseP, as well....I think it is most definitely a takeover to produce GNU/systemd - even if that wasn't the original intention. |
| vainrveenr Sep 02, 2014 7:42 PM EDT |
Quoting:thankyou for the link.......I've just read it and I am both staggered and dismayed.Link now available on LXer, via http://lxer.com/module/newswire/view/205483. Quoting:I think it is time Linus bought into this, if he hasn't already.He already very much was brought into this exactly five months ago, as can be seen in the Linux Kernel Archives List post linked to at LXer as 'Linus Torvalds vs. systemd developers' via http://lxer.com/module/newswire/view/200597/index.html. AAMOF, later on in that very thread, Linus Torvalds tells off in no uncertain terms (perhaps even bans?) one of the key systemd developers. The 'Linus Torvalds vs. systemd developers' piece happened to spawn a reaction in the LXer comments thread Linus Geekfight! Systemd banned!. |
| gus3 Sep 02, 2014 7:55 PM EDT |
Paul Venezia follows up with an epic smackdown. The punchline is excellent. |
| TxtEdMacs Sep 02, 2014 8:30 PM EDT |
My Dearest August Gus III, [serious] First I read the link in your last post. Now I am confused, correct me if I am wrong. In the above thread, the thrust of most of the argument was that systemd fit the server needs not the desktop where it was a complicated distraction. Well if anything, the author of that post you showed said the opposite, i.e. those really interested in servers were likely to move to FreeBSD where systemd does not rule. The one argument explicitly stated was it does away with the complexity of runlevels. What? Even I have an idea what they are. Moreover, I knew in greater detail when I first started to use Linux. One of the first things I had to do was modify a bash script in one of the startup routines, where I changed some code in what I thought might be reasonable to get my system to run. This was back in '98 or '99, where I think the change allowed me to get my network working. At the time I still could not get a printer working, but that was solved with an early version of Mandrake. Even for me, if these are the benefits of systemd, I can forgo it. [/serious] As always, YBT |
| mbaehrlxer Sep 02, 2014 10:39 PM EDT |
JaseP wrote:the problem with systemd is that the developers are swallowing up additional stuff,... not just the init process, but also mount, cron, power management... the list keeps growing. The bloat problem isn't size or number of files it takes to run,... it's in the way it has become viral, trying to take domain over other services. that's a good point. and it stands on its own as criticism imho. i am not sure i agree that this is a problem though, at least, i don't see it as a technical problem. i consider this a social problem. technically, what matters to me is the functionality. i don't care who wrote it, whether it was rewritten, or what not. i care that it works, and is stable, fixes bugs and doesn't introduce new problems. those are the things that systemd needs to be measured by. does systemd introduce problems that the current system doesn't have? JaseP wrote:Pretty soon the OS could be described as GNU/systemd (featuring Linux) instead of (the politically correct) GNU/Linux.as long as GNU is in the name, it would still be politically correct. and if systemd does take over the core userland of linux, i am fine with that. that's really no different than BSD core userland which is developed by the same team as the kernel. greetings, eMBee. |
| mbaehrlxer Sep 02, 2014 10:54 PM EDT |
http://boycottsystemd.org/ has some interesting points, in particular this link is interesting: http://www.openbsdfoundation.org/gsoc2014.html#systemd it proves beyond a doubt, that lack of portability of systemd or at the least, of the APIs that other applications use, is a real problem. maybe not for linux, but for everyone else. on the other hand, i consider having multiple implementations of an API a good thing. it pushes standardization and stability. greetings, eMBee. |
| kikinovak Sep 03, 2014 3:07 AM EDT |
Systemd proponents remind me of those Jehovah's Witnesses who come knocking at my door on a regular basis. Utterly ignorant of all rational arguments, they keep coming back with their same tiresome rants until one day hopefully I will see the light. Or until I yell at them to just bugger off and get a life. |
| Ridcully Sep 03, 2014 3:30 AM EDT |
For the record, would someone please give me a definitive answer on the structures of systemd and sysvinit in the following areas: My present understanding of systemd is that it has the form of a series of "interconnected binary blobs". If so, is the underlying source code freely available ? If it isn't, then it flies in the face of FOSS. Is sysvinit a "binary blob" or is it an open text file ? My present understanding is that the source code for sysvinit is freely available.....well, if it is a text file, it has to be. I'm trying to understand some fundamentals here that are outside my normal areas of interest, so I do beg the patience of the more knowledgeable readers if my questions seem pretty basic. However, I do get hostile if I think the fundamental tenets of Linux and FOSS are being undermined. |
| CFWhitman Sep 03, 2014 9:13 AM EDT |
The underlying source code for systemd is freely available. It would never have gotten anywhere at all if it weren't. The complaint about binary blobs is in reference to the fact that a lot of things that were handled by scripts and logged in plain text are in systemd handled by binaries and logged in code. |
| Bob_Robertson Sep 03, 2014 11:10 AM EDT |
CF, "handled by scripts", yes, _plain_text_ scripts. Plain text, as in, "Gee, I wonder what options are available for this script? Let's read it and see." |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!