untrue
|
| Author | Content |
|---|---|
| set May 01, 2013 9:14 PM EDT |
Currently udev is in the same tree as the rest of systemd. Systemd depends on udev, but
udev does not depend on systemd to run. No maintainer uses systemd in their distribution because they
were forced. They chose it because they liked it. Others happily just use the udev component,
just as before. And if anyone is worried that the evil systemd people will break their udev, they can fork it. (Actually some Gentoo people did fork to eudev. ) Gentoo, however, defaults to the standard udev paired with their openrc init system, while allowing you to chose systemd as an option. I think Debian supports three or more init systems. If systemd takes over the world, it will be because it is desireable, not because Poettering tricked everyone and is laughing maniacly, whilst stoking a white persion cat in his mountain fortress.... Someone somewhere might have to learn something new though. That would be a tragedy. (I dont run systemd on my main system, but I have it working on an Arch box, and its pretty slick.) |
| linux4567 May 02, 2013 12:57 AM EDT |
>If systemd takes over the world, it will be because it is desireable, not because Poettering tricked everyone and is laughing maniacly That's not quite true, if systemd takes over that because of the dominant position of Redhat in the Linux world, Poettering only 'tricked' Redhat, which is bad enough. systemd is only slick as long as it works, when you have to start troubleshooting it you will change your mind. |
| caitlyn May 02, 2013 3:42 AM EDT |
Nope, you can troubleshoot it a lot more easily than a huge pile of random scripts. systemd is a change that was long overdue IMNSHO. |
| set May 02, 2013 4:50 AM EDT |
@linux4567 What ciatlyn says. Ive been mucking with linux alone almost 20 (*sigh*) years, and various unix init systems. Sysv init sucks @ss. BSD systems are nice and simple, but very limited. Openrc puts lipstick and and a prom dress on that sysv pig... The only aspect of systemd that I did not care for was the journal, for a personal system its kind of clunky. Fortunately, its optional. Besides, who the hell even deals with the inner workings of their init system besides sysadmins? Is all this controversy really about some sysadmins cowering before the prospect of having to learn something new? Somewhere a BOFH is crying. |
| slacker_mike May 02, 2013 2:27 PM EDT |
Quoting:It concerns me greatly that there was what appears to be a total lack of concensus building around even attempting to have agreement that systemd was worth pursuing before getting the project underway. Am I wrong to think that systemd was proposed for inclusion/adoption within the Fedora community just like any other proposed feature? From there other Linux community distros have begun to adopt systemd after discussion within their communities. Now openSUSE, Mageia, Arch, Sabayon, ad ROSA have all adopted systemd. Isn't that the definition of community consensus building? |
| mbaehrlxer May 02, 2013 10:16 PM EDT |
the best argument against systemd i read was that because it is written in C you'd have to mess with compilers etc if there is something wrong with it, making problems in systemd almost as hard to work with as kernel problems.
shell scripts are much farther removed from that. on the other hand i think it is also a bad argument, because really, in which language should systemd be written instead? a huge shellscript? i don't think so. python? maybe. but then you potentially lock the python version for everyone, or require multiple python versions to be installed. to be a better choice than C it would have to be a language that can generate self-standing executables but is easy to debug at runtime. the only language capable of that i know of is common lisp. (although i hope by now there are others) greetings, eMBee. |
| caitlyn May 03, 2013 10:54 PM EDT |
Quoting:Am I wrong to think that systemd was proposed for inclusion/adoption within the Fedora community just like any other proposed feature? From there other Linux community distros have begun to adopt systemd after discussion within their communities. Now openSUSE, Mageia, Arch, Sabayon, ad ROSA have all adopted systemd. Isn't that the definition of community consensus building?You're not wrong at all. The actions you described all happened precisely as you describe them. Of course, there are those who disagree with the various distro developers and communities listed above and they do find other ways to attack a project and a change they dislike, as is their right. However, this isn't something being foisted on the Linux community by Red Hat. This is something Fedora had adopted and that Red Hat will adopt in RHEL 7. Other distributions have decided to follow suit going through their own decision making processes. Similarly, at least for now, Patrick Volkerding of Slackware has decided to put off or opt out of implementing systemd. As always there are choices. Quoting: Besides, who the hell even deals with the inner workings of their init system besides sysadmins? Is all this controversy really about some sysadmins cowering before the prospect of having to learn something new?That pretty much sums it up. I see the same thing all the time regarding SELinux. You know what, SELinux isn't hard at all and it is very powerful in terms of being able to control filesystem/data security in a very granular way not offered by standard UNIX/Linux permissions. It's also not hard to generate a sane working policy. You simply run SELinux in permissive mode, take the log output, and generate your policy based on the real world results on the system in question. You know up front precisely how SELinux will respond in enforcing mode that way. Despite all of this, some system administrators continue to rail against SELinux and argue for disabling it entirely. If a sysadmin working for me disabled SELinux because they didn't bother to learn or understand it I'd fire them on the spot. You don't disable security because you're too lazy to learn. systemd is another case where it really isn't difficult to learn and, in the end, it simplifies the job of systems administration immensely. It doesn't lessen your ability to control the boot process. Quite the contrary: it increases control and lets you do it in one place without having to fiddle with various scripts. Regarding the choice of language: most if not all core Linux internals are written in C. I would think that's the only choice that makes sense if for nothing else than consistency. |
| mbaehrlxer May 04, 2013 12:39 AM EDT |
thanks for that insight, especially about SELinux. i admit that i took the lazy route on my home machines until now too. however, i wanted to share one experience i just made (and probably made before) which is that learning something new can be very hard when you are under pressure. in this case i was tasked with completing the replacement of an apache setup with nginx which was started by someone else which should not be hard, but the setup was complex and i was new to nginx and also only had passing familiarity with apache. so at the beginning i looked at a complex puzzle and could not immediately figure out what was needed to make it work. nginx is not hard to learn, and i eventually got the hang of it, but it is one thing to start from scratch, build one site after another slowly, learning along the way, and another to dive into the deep end and deal with a complex system from the get-go. it appears both systemd and SELinux fall into that category. there is no building up a configuration step by step. you need to get all of the system running at once, and it's broken until you do so. so the initial learning curve appears very high, just like a wall appears very high when you stand right in front of it, but a gently sloped hill of the same height appears small from any viewpoint. so i can understand working sysadmins balking at the idea. you don't add a new complex system until you can understand the need for it and see the benefit. and as long as the benefit is questioned the attitude is probably more like "i simply can't afford to spend time with this". greetings, eMBee. |
| caitlyn May 05, 2013 5:36 PM EDT |
eMBee, I can't disagree with anything you've written. It's always easier to build something from scratch that to figure out what someone else has done if it isn't well documented. Sadly, documentation went out with the personnel cuts after the dot com bubble burst and hasn't made much of a comeback since. Never mind the amount of productive time lost due to lack of documentation. As a systems architect/admin and also a security consultant I've seen the need for both SELinux and systemd and understand how they, in fairly short order, can be real improvements. If someone can't even take the time to understand that they are, or course, going to fight the changes. Sadly, they then write articles full of FUD based on their own unfounded fears. I'm reminded of a Windows admin I worked with years ago. He was a bright guy with a very good understanding of networking. He learned quickly. The company decided to shift to Linux in the server room and he swore he'd rather quit than learn Linux. From what I understand he eventually did just that after I had moved on. It was stupid because he certainly could have picked up Linux easily and would have had excellent support from both management and his peers. "Fear is the mind killer..." |
| gus3 May 06, 2013 12:35 PM EDT |
But what if the shift had been from Linux to Windows? |
| BernardSwiss May 06, 2013 7:04 PM EDT |
@caitlyn -- as far as I caan tell, the argument isn't about the need for or usefulness of something like systemd, but rather about the implementation. |
| caitlyn May 07, 2013 2:47 PM EDT |
Again, I don't think anything wrong was done in the implementation. BTW, systemd made it into SLES 11 SP3. It will be in RHEL 7. It is becoming a de facto standard in enterprise environments. Also, note that SUSE implemented it before Red Hat. @gus3: I do Windows work now and again. I would have done what my employer asked and then considered my options. I certainly wouldn't make blanket statements. |
| Dietrich May 17, 2013 7:52 PM EDT |
I've looked at the 'flip side' on systemd and posted this follow-up story: http://www.linuxadvocates.com/2013/05/cgroups-big-win-for-sy... Regards, --Dietrich |
| caitlyn May 17, 2013 7:57 PM EDT |
Thanks for considering both sides. To me that shows a lot of integrity. |
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!