LinuxQuestions.org
Help answer threads with 0 replies.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Distributions > Slackware
User Name
Password
Slackware This Forum is for the discussion of Slackware Linux.

Notices



Closed Thread
 
Search this Thread
Old 06-02-2013, 11:06 AM   #301
commandlinegamer
Member
 
Registered: Dec 2007
Posts: 98

Rep: Reputation: 17

I must say I'm really impressed by the civility, maturity and willingness to back up argument with solid evidence, in this thread.
 
Old 06-02-2013, 11:19 AM   #302
the3dfxdude
Member
 
Registered: May 2007
Posts: 332

Rep: Reputation: 100Reputation: 100
Quote:
Originally Posted by elvis4526
]I've tried really hard to tell you why everthing that you said since the begginning of this thread is FALSE and TOTALLY WRONG with actual proof that support my statement.
I've read this entire thread, and you got some things wrong about how systemd works. Sorry. You do not have convincing proof.

Quote:
Originally Posted by Martinus2u View Post
Then do it and start contributing. So far all you have done is defaming a developer as drug dealer and calling for grievous bodily harm against him using a baseball bat.
What makes you think he is not doing something? That he has not done research? Why does he need to be a contributor, if something he has already works?

I've been at the point of doing something some time. I've studied systemd code. I've tried coding to test the userspace concepts. This, and other discussions don't interest me because they don't go anywhere. I'm interested in preventing the userspace from turning to crap. So I will be willing to engage people technically to solve problems. Reaper has provided some good information that I didn't know about. That shows he really is doing homework on this.

Why do you want to defend Poettering? He defamed BSD for no good reason when questioned why he doesn't support it. His attitude is childish, but we are supposed to just give in to him; we have no choice and can't do anything? You severely underestimate what a maintainer with a bad attitude can cause people to do. I never have used BSD, but Poettering showed himself very poorly to the whole community. The BSD people don't deserve that. Reaper may be mad too, but please don't provoke him by saying "you aren't doing anything". You do not know that.
 
9 members found this post helpful.
Old 06-02-2013, 01:58 PM   #303
ttk
Member
 
Registered: May 2012
Location: Sebastopol, CA
Distribution: Slackware
Posts: 267
Blog Entries: 13

Rep: Reputation: 220Reputation: 220Reputation: 220
Getting back to the original topic .. is systemd inevitable?

On one hand, almost every major distribution is switching to it, udev will be inextricably entangled in it, and udev has provided the best support for device management.

On the other hand, partially due to backlash against systemd, several alternatives to udev have warmed up and switched into high gear. Also, Patrick himself is wary of systemd (for the obvious reasons):

http://www.linuxquestions.org/questi...ml#post4727299
http://www.linuxquestions.org/questi...ml#post4756034
http://www.linuxquestions.org/questi...ml#post4896422

On the gripping hand, Slackware is ultimately a practical distribution, and Patrick is pragmatic about its development. If past behavior is any guide to the future, he will adopt whatever system serves users' needs without crippling the system or imposing an undue maintenance burden on the Slackware development team.

If I had to predict anything, it's that he will hold off on adopting a systemd-dependent udev for as long as possible while kicking the tires on some udev alternatives (and perhaps assessing prospects for assisting in their development) and keeping an eye on systemd to see if it shapes up. "As long as possible" may be years -- udev-182 isn't intolerably bad. Eventually he will choose udev/systemd or one of the udev alternatives, based on what his investigations teach him.

I hope Slackware comes out of the process unburdened by systemd, but trust that Patrick will make his decision rationally. If that means holding our noses and learning to live with systemd, so be it.
 
2 members found this post helpful.
Old 06-02-2013, 02:39 PM   #304
hitest
Senior Member
 
Registered: Mar 2004
Location: Prince Rupert, B.C., Canada
Distribution: Slackware, OpenBSD
Posts: 4,282

Rep: Reputation: 588Reputation: 588Reputation: 588Reputation: 588Reputation: 588Reputation: 588
Quote:
Originally Posted by ttk View Post
I hope Slackware comes out of the process unburdened by systemd, but trust that Patrick will make his decision rationally. If that means holding our noses and learning to live with systemd, so be it.
I've had the pleasure of running Slackware since 2004(version 10.0). From my perspective Pat always chooses a rational, sane development pathway for Slackware. We are in good hands.
 
Old 06-02-2013, 02:52 PM   #305
elvis4526
Member
 
Registered: Aug 2011
Posts: 114

Rep: Reputation: Disabled
Quote:
Originally Posted by the3dfxdude View Post
I've read this entire thread, and you got some things wrong about how systemd works. Sorry. You do not have convincing proof.



What makes you think he is not doing something? That he has not done research? Why does he need to be a contributor, if something he has already works?

I've been at the point of doing something some time. I've studied systemd code. I've tried coding to test the userspace concepts. This, and other discussions don't interest me because they don't go anywhere. I'm interested in preventing the userspace from turning to crap. So I will be willing to engage people technically to solve problems. Reaper has provided some good information that I didn't know about. That shows he really is doing homework on this.

Why do you want to defend Poettering? He defamed BSD for no good reason when questioned why he doesn't support it. His attitude is childish, but we are supposed to just give in to him; we have no choice and can't do anything? You severely underestimate what a maintainer with a bad attitude can cause people to do. I never have used BSD, but Poettering showed himself very poorly to the whole community. The BSD people don't deserve that. Reaper may be mad too, but please don't provoke him by saying "you aren't doing anything". You do not know that.


I do not have convincing proof?
Man pages and articles from one of the main systemd developper aren't convincing?
ReaperX7 did his homework?

If he's claiming that systemd is one single bloated binary that try to do everything, he clearly did not inspect a systemd source tarball and therefore he didn't do his homework.

Last edited by elvis4526; 06-02-2013 at 02:54 PM.
 
Old 06-02-2013, 03:00 PM   #306
Erik_FL
Member
 
Registered: Sep 2005
Location: Boynton Beach, FL
Distribution: Slackware
Posts: 801

Rep: Reputation: 247Reputation: 247Reputation: 247
Linux systems could benefit from an optimized loader system similar to the one in Windows. What I'm referring to are the features that track the files and disk blocks that are loaded. The next time that the system is started or a particular program is loaded, the required modules and files are preloaded. That has the effect of turning a lot of dis-contiguous disk I/O into contiguous I/O and loading information before it is needed by programs. It greatly speeds up launching applications and system initialization.

Some operating systems achieve this anticipatory loading by saving a snapshot of a booted system and then loading the entire snapshot prior to initializing the OS. The normal drivers, modules and tasks are all loaded from a contiguous disk snapshot that was saved. If a system administrator wants to add things, they simply boot the system, start the additional programs and save a new snapshot. Some operating systems also provide a "virtual shell" that allows administrators to directly load or unload tasks, modules and drivers in the snapshot.

Parallelism in initialization sounds like a good idea until one realizes that the disk access is a major bottleneck. In reality, most of the advantage to "parallelism" is by having the disk I/O occur in parallel with other initialization tasks. That can mostly be achieved by anticipatory loading without the need for excessive parallelism.

I would really be interested in some benchmarks of "systemd" versus other initialization systems to see how much (if any) speed up "systemd" provides. Also, if there are any Linux extensions to provide anticipatory loading, how much benefit do those provide versus "systemd". Does "systemd" provide anticipatory loading, and does it provide that for loading normal applications as well as system initialization?

Whether one likes "systemd" or not, the key question is whether "systemd" provides significant benefits over other initialization systems. Complexity by itself isn't a bad thing if that produces enough benefits to justify the complexity. Some problems require complex solutions. Sometimes the complexity can become a worse problem than the ones it tries to solve.
 
Old 06-02-2013, 03:46 PM   #307
elvis4526
Member
 
Registered: Aug 2011
Posts: 114

Rep: Reputation: Disabled
Quote:
I would really be interested in some benchmarks of "systemd" versus other initialization systems to see how much (if any) speed up "systemd" provides.
On Arch Linux my total boot time is exactly 3.782 seconds (kernel + userspace), on a ssd of course.
 
1 members found this post helpful.
Old 06-02-2013, 04:01 PM   #308
astrogeek
Senior Member
 
Registered: Oct 2008
Distribution: Slackware [64]X{.0|.1|.2|-current} ::X>=12<=14, FreeBSD_10{.0|.1}
Posts: 2,182

Rep: Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859
Quote:
Originally Posted by elvis4526 View Post
On Arch Linux my total boot time is exactly 3.782 seconds (kernel + userspace), on a ssd of course.
But that tells us nothing without a comparison to non-systemd boot on same system.

A useful benchmark is a comparison (i.e., context), not a number.

Last edited by astrogeek; 06-02-2013 at 04:10 PM. Reason: Added context parenth
 
Old 06-02-2013, 04:12 PM   #309
Ser Olmy
Senior Member
 
Registered: Jan 2012
Distribution: Slackware
Posts: 2,129

Rep: Reputation: Disabled
Quote:
Originally Posted by elvis4526 View Post
On Arch Linux my total boot time is exactly 3.782 seconds (kernel + userspace), on a ssd of course.
How long did it take with a non-systemd distro?

Regular desktop and laptop computers use between 5 an 15 seconds for POST, which brings the total boot time down to between 9 and 19 seconds, which is not bad at all. Servers, on the other hand, typically have a POST process that takes between 2 and 5 minutes. Saving 30 seconds on boot time becomes a lot less relevant in those cases.

Is your system fully usable after 4 seconds? On Windows, a typical server can take as little as 30 seconds from completed POST to the desktop appears (provided you can type your password quickly enough), but you can't actually do much until another 60-90 seconds have passed, as a number of important services are still starting in the background. Actually, you can save a quite few seconds by NOT logging in immediately, but instead wait until the drive activity light settles down a bit.
 
2 members found this post helpful.
Old 06-02-2013, 04:20 PM   #310
elvis4526
Member
 
Registered: Aug 2011
Posts: 114

Rep: Reputation: Disabled
Quote:
Originally Posted by Ser Olmy View Post
How long did it take with a non-systemd distro?

Regular desktop and laptop computers use between 5 an 15 seconds for POST, which brings the total boot time down to between 9 and 19 seconds, which is not bad at all. Servers, on the other hand, typically have a POST process that takes between 2 and 5 minutes. Saving 30 seconds on boot time becomes a lot less relevant in those cases.

Is your system fully usable after 4 seconds? On Windows, a typical server can take as little as 30 seconds from completed POST to the desktop appears (provided you can type your password quickly enough), but you can't actually do much until another 60-90 seconds have passed, as a number of important services are still starting in the background. Actually, you can save a quite few seconds by NOT logging in immediately, but instead wait until the drive activity light settles down a bit.
Yes, my system is fully usuable after ~4 seconds since it is the time for the kernel and all the userspace to boot up.
Actually, userspace alone boot in 1.021 second.

I didn't check exactly the time that it took to boot on Slackware for instance, but it was at least 2x slower for sure.
The boot time on Gentoo are quite nice too.
OpenRC is the fastest thing i know next to systemd actually but it doesn't have service supervision unfortunately.

Last edited by elvis4526; 06-02-2013 at 04:24 PM.
 
1 members found this post helpful.
Old 06-02-2013, 04:29 PM   #311
astrogeek
Senior Member
 
Registered: Oct 2008
Distribution: Slackware [64]X{.0|.1|.2|-current} ::X>=12<=14, FreeBSD_10{.0|.1}
Posts: 2,182

Rep: Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859
Quote:
Originally Posted by elvis4526 View Post
Yes, my system is fully usuable after ~4 seconds since it is the time for the kernel and all the userspace to boot up.
Actually, userspace alone boot in 1.021 second.

I didn't check exactly the time that it took to boot on Slackware for instance, but it was at least 2x slower for sure.
The boot time on Gentoo are quite nice too.
OpenRC is the fastest thing i know next to systemd actually but it doesn't have service supervision unfortunately.
I have been reading and trying to not comment any more on this thread, but it has become too strange...

Well, to re-quote from an earlier post, "HOLY COW I'M TOTALLY GOING SO FAST OH F***"!
 
Old 06-02-2013, 04:30 PM   #312
elvis4526
Member
 
Registered: Aug 2011
Posts: 114

Rep: Reputation: Disabled
Quote:
Originally Posted by astrogeek View Post
I have been reading and trying to not comment any more on this thread, but it has become too strange...

Well, to re-quote from an earlier post, "HOLY COW I'M TOTALLY GOING SO FAST OH F***"!
Don't you have something better to say?
 
Old 06-02-2013, 04:35 PM   #313
astrogeek
Senior Member
 
Registered: Oct 2008
Distribution: Slackware [64]X{.0|.1|.2|-current} ::X>=12<=14, FreeBSD_10{.0|.1}
Posts: 2,182

Rep: Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859Reputation: 859
Quote:
Originally Posted by elvis4526 View Post
Don't you have something better to say?
No, I've expressed my thoughts already.

And following earlier posts, you would probably ignore it anyway, accuse me of not contributing anything useful and then post another useless number to demonstrate how smart you are... like I said, this thread has become very strange.
 
Old 06-02-2013, 04:42 PM   #314
Ser Olmy
Senior Member
 
Registered: Jan 2012
Distribution: Slackware
Posts: 2,129

Rep: Reputation: Disabled
Quote:
Originally Posted by elvis4526 View Post
Yes, my system is fully usuable after ~4 seconds since it is the time for the kernel and all the userspace to boot up.
Actually, userspace alone boot in 1.021 second.
Well, that just can't be done with servers, I'm afraid. A number of processes takes longer to initialize than 2 seconds, and some depend on other services which also take a few seconds to start. For instance, I've found that I need to wait at least 2 seconds after starting libvirtd before I can start a VM.

But then again, systemd is supposedly not really about reducing boot time. I have to admit that after reading some of the documentation, I still don't know what it's supposed to do. It certainly makes Linux look a lot more like Windows, binary logfiles and all, but surely that's not exactly an improvement?
 
Old 06-02-2013, 04:46 PM   #315
elvis4526
Member
 
Registered: Aug 2011
Posts: 114

Rep: Reputation: Disabled
Quote:
Originally Posted by astrogeek View Post
No, I've expressed my thoughts already.

And following earlier posts, you would probably ignore it anyway, accuse me of not contributing anything useful and then post another useless number to demonstrate how smart you are... like I said, this thread has become very strange.
Well number are really useful to compare the boot speed against sysvinit since this thread turned in a "Let's say total crap about systemd without any proof".
I won't speak for sysvinit, but all i can say is that systemd is capable of booting an entier userspace in 1s. Do you want to know this statement is based on what? It's based on a nice utility that comes with systemd called systemd-analyze.
It analyze and save how much time the last bootup took.

So here you have my argument and my source.

But anyways, you must be right, Lennart is just an evil drug dealer.
 
  


Closed Thread

Tags
cgroups


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
LXer: Slackware: Is Systemd Inevitable? LXer Syndicated Linux News 5 07-22-2013 05:54 AM
[SOLVED] slackware and systemd fl0 Slackware 512 08-29-2012 12:07 PM
slackware and systemd (OT) eloi Slackware 44 08-24-2012 05:36 PM
[SOLVED] systemd and Slackware's future kikinovak Slackware 95 07-14-2012 12:40 PM
Boot Delay 30min: systemd-analyze blame systemd-tmpfiles-setup.service BGHolmes Fedora 0 07-27-2011 10:02 AM


All times are GMT -5. The time now is 01:42 AM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
identi.ca: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration