| View previous topic :: View next topic |
| Author |
Message |
ycUygB1
Apprentice
Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon
|
 Posted: Fri Oct 25, 2013 10:06 am Post subject: systemd or OpenRC, which way is it moving? |
 |
|
So I thought I would try out Gnome 3, which forced me to move from OpenRC to systemd. I didn't like Gnome 3, and I installed mate instead, which is based on Gnome 2. In addition, I found little documentation on systemd. It throws bizarre errors and seems to behave inconsistently. When I get an error like
| Code: | Failed to issue method call: Unit nscd.service failed to load: No such file or directory. See system logs and 'systemctl status nscd.service' for details.
/lib/resolvconf/libc: line 191: /etc/resolv.conf: Permission denied
Failed to issue method call: Unit nscd.service failed to load: No such file or directory. See system logs and 'systemctl status nscd.service' for details.
Failed to issue method call: Unit nscd.service failed to load: No such file or directory. See system logs and 'systemctl status nscd.service' for details.
localhost# systemctl status nscd.service
nscd.service
Loaded: error (Reason: No such file or directory)
Active: inactive (dead)
|
It is hard to know where to turn. So I thought I might move back to the comfortable world of OpenRC, which is apparently supported by Gentoo equally as well as systemd.
Before I do this, I wonder if I am being a luddite. Which direction are things going? Is systemd the future? Will OpenRC cease to be supported at some point? |
|
| Back to top |
|
 |
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1483
|
| Posted: Fri Oct 25, 2013 10:45 am Post subject: |
|
|
Nobody can see the future but trend is Systemd. (A trend is the slightest movement in one direction and Gnome-3.8 had a hard dependency on Systemd lately). If you want to learn fundamentally about init system
Openrc
is easy and transparent to begin with! Later on you will understand Systemd the better you know about it: Openrc already implements some of Systemds dependency features.
The problem with documentation rather is too much of it, for example issue the command: apropos systemd |
|
| Back to top |
|
|
wuzzerd
Guru
Joined: 05 Jan 2005
Posts: 467
Location: New Mexico
|
| Posted: Fri Oct 25, 2013 2:09 pm Post subject: |
|
|
| ulenrich wrote: | | The problem with documentation rather is too much of it, for example issue the command: apropos systemd |
The documentaton, like the commands are a bit logorrheic. |
|
| Back to top |
|
|
ddriver
n00b
Joined: 24 Feb 2005
Posts: 67
|
| Posted: Fri Oct 25, 2013 2:23 pm Post subject: |
|
|
| I will never use systemd with Gentoo. |
|
| Back to top |
|
|
Fitzcarraldo
Advocate
Joined: 30 Aug 2008
Posts: 2052
Location: United Kingdom
|
| Posted: Fri Oct 25, 2013 3:07 pm Post subject: |
|
|
OpenRC works perfectly on my machines, and I have no desire whatsoever to learn or have the hassle of configuring another init system. I have enough hassle to deal with as it is, without having to mess around with yet another change to something that isn't broken. I boot a couple of times per day, so saving a few seconds during boot is the absolute least of my concerns. Now, if systemd magically halved the disk access time or doubled the CPU speed then I'd be interested! Maybe in a few years when systemd is fully debugged (by a bunch of other people prepared to put up with learning something new and filing bug reports) then I'll take a look at it, but for now I have zero interest in changing from something that already works perfectly for me.
Apart from a couple of laptops running Gentoo with OpenRC, I have another couple of laptops running Sabayon (a Gentoo-based distribution) with OpenRC. Just to see the supposed benefit of using systemd, I temporarily switched one of the latter to systemd but the saving in boot time was negligible: on a 2009 Acer Aspire 5536-643G25Mn laptop (see link in my signature for full spec.) the times to boot from the GRUB 2 menu to the DM login screen for a fully-updated 64-bit Xfce installation were as follows:
| Code: | OpenRC with rc_parallel="NO" 33 seconds
OpenRC with rc_parallel="YES" 31 seconds
systemd 29 seconds |
The differences between OpenRC and systemd would not be dramatically different on any of my machines. systemd is only going to make a big difference in boot time on machines with e.g. a SSD and massively multi-core CPUs (quad core or more). For the machines I'm using (and will be using for some time to come), the difference is chicken feed. I'm not bothered about saving a few seconds in boot time (as I mentioned previously I don't reboot often). OpenRC works fine on all my machines, I know how to use it, and I do not particularly want to have to learn how to use another init system when what I use now works perfectly well for me. Each to his own, as the saying goes. I just hope the Gentoo developers don't abandon OpenRC just because several other distributions have adopted systemd.
_________________
Clevo W230SS: amd64, VIDEO_CARDS="intel modesetting nvidia".
Compal NBLB2: ~amd64, xf86-video-ati. Dual boot Win 7 Pro 64-bit.
OpenRC systemd-utils[udev] elogind KDE on both.
My blog |
|
| Back to top |
|
|
LoTeK
Apprentice
Joined: 26 Jul 2012
Posts: 270
|
| Posted: Fri Oct 25, 2013 3:54 pm Post subject: |
|
|
I see it as Fitzcarraldo, although on my yeeloong without an SSD / single core, the boot time was about 10 seconds longer with openRC (gentoo) than with systemd (parabola). But maybe I have configured something the wrong way or there are some other reasons.
So I would and will stay with openRC, but it seems that the only distributions that still have sysVinit or openRC as default init system are gentoo and debian. Lately I've read a thread or article that was about switching to systemd in debian. So if you have to work with a RedHat system or any other of those distributions, then you probably have to know other init systems, don't you? especially when you want to do sysadmin stuff. And it seems that most people support the switch to systemd, upstart etc, despite all the hate threads... Or am I wrong (hopefully)?
edit:
what disturbs me is that for example gnome3 depends on systemd. I mean, why should a window manager / desktop be dependent on any init system? this is seriously wrong, isn't it?
I would like to see a new init system, with faster boot time, etc, but definitely not in this way. In the debian thread someone compared openrc and systemd to a microkernel and a monolitic one. He said that openrc is simple like a microkernel, but moves the complexity to the userspace. (if I recall correctly and if this makes any sense, since openrc is simple from a user point of view  )
_________________
"I want to see gamma rays! I want to hear X-rays! Do you see the absurdity of what I am? I can't even express these things properly because I have to conceptualize complex ideas in this stupid limiting spoken language!" |
|
| Back to top |
|
|
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1483
|
| Posted: Fri Oct 25, 2013 8:02 pm Post subject: |
|
|
| LoTeK wrote: | | what disturbs me is that for example gnome3 depends on systemd. I mean, why should a window manager / desktop be dependent on any init system? this is seriously wrong, isn't it? |
If you have one machine for a classroom with three workgroups/seats having each some sessions for the children. What to do if you don't want to use the old unmaintained and broken consolekit?
| Quote: | | He said that openrc is simple like a microkernel, but moves the complexity to the userspace. (if I recall correctly and if this makes any sense, since openrc is simple from a user point of view ) |
Yes,
but simple only for the simple use case. See above. But I still advise Openrc to people who want to learn about init (I expect a new Gentoo user he wants to), because it is simple. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Fri Oct 25, 2013 8:20 pm Post subject: |
|
|
| LoTeK wrote: | | what disturbs me is that for example gnome3 depends on systemd. I mean, why should a window manager / desktop be dependent on any init system? this is seriously wrong, isn't it? |
Yes. It's vertical integration, which just means your software only works with one "back-end"; the reason to do that is usually efficiency, but that does not apply when talking about a DE and an init system. They're too far apart; by the time the DE is starting, you're at the end of the whole init phase. OFC, if a few seconds on boot really matters that much to you, eg with VMs, to the extent that it overrides any experience you might have had in the past with "One True Way" and longer-term consequences, and the continual breakages don't bother you, then as with everything else, that's your choice, and no real Gentooista will argue, nor try to impose their method on you.
It's just not done, AFAIC. I simply wish that were acknowledged on all sides.
| Quote: | | I would like to see a new init system, with faster boot time, etc, but definitely not in this way. In the debian thread someone compared openrc and systemd to a microkernel and a monolitic one. He said that openrc is simple like a microkernel, but moves the complexity to the userspace. (if I recall correctly and if this makes any sense, since openrc is simple from a user point of view :) ) |
Well it works as an analogy until you try to extend it and use it as a design-metaphor. OpenRC does as little as possible and stays out of the way, so that you're really just interacting with the base software. Both however are user-land apps, so the idea of moving complexity, which in the case of micro-kernels can lead to performance issues as well as over-complex code, since encapsulation is by definition not as complete, does not really apply. Systemd is actually far more complex, and breaks encapsulation, unless one lumps the whole project into one, where anyone with a few years experience knows it's not just one project, and incorporates several into one codebase.
Unfortunately each of those things it encompasses has been the subject of wide orthogonal development, so effectively systemd starts as a jack-of-all-trades and a master-of-none. Personally I'd rather employ the work of experts in each, who are used to fitting in with what the user tells them to do, not what they hope is the right thing to do, "because" it is for the developer's machine (or "use-cases" which are a dirty word where I work: I get told off if I use it, tho "wayttd" is fine, and told to work on simple tools that do one thing well, perhaps with "options consonant with the main algorithm implemented", such as -q with grep, the "canonical example" according to my boss, since I "cannot be present when the software is in use, and must thus accept that control is with the user." I've heard it too many times now, to pretend I don't remember/understand the point.) |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Fri Oct 25, 2013 8:44 pm Post subject: |
|
|
| ulenrich wrote: | | If you have one machine for a classroom with three workgroups/seats having each some sessions for the children. What to do if you don't want to use the old unmaintained and broken consolekit? |
Well I don't use nubkit with kde-4.10 currently, and if one has to maintain several machines from the same binpkgs, as I do, I would not advise it.
"Fast console switching" would be even worse in a classroom environment, ime, and we don't use it on our LAN either. You'd just end up with lots of "sessions" on each machine, as kids logged in to each other's seats, each wasting resource, and with security implications. A simple timed lockout is better in both situations: you want kids to learn a bit of discipline (ie stick to my desk for my work, and shared stuff is on the server) and the best way to get them used to using the correct method is to make it the only feasible method as they learn the main work, which this is not. It also cuts off avenues to bullying, since it enforces boundaries and respect for one another's space (by first having a sense of one's own space.)
This isn't "hot-desking" and even proper "hot-desking" does not require it: for a start you want your phone-calls routed to the correct extension, so it's never as simple as people make out. Logout out of the machine when you're done, otherwise it's your machine for the day/as long as your calls are ending up at this desk. An admin can always override remotely if required, and you typically have other automated things going on in a corporate environment, so there is plenty of scope to tweak. Scriptlets can be run on X login/logout for example, and that doesn't need anything special at all. It's almost as-if the people who designed unix and X worked over networks..
I've got a lot of time for pam, from my experience of interfacing with AD. Yes I know it's not the same thing, but people tend to be snobby about it, and I really don't understand why. If you're in an organisational environment, it rocks. It works great at home as well, it's just one rarely needs to tweak it, so you don't tend to interact with it apart from seeing it in the occasional etc-update. |
|
| Back to top |
|
|
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1483
|
| Posted: Sat Oct 26, 2013 12:17 am Post subject: |
|
|
@steveL, excuse me I answer to this one, it's more fundamental:
| steveL wrote: | | LoTeK wrote: | | what disturbs me is that for example gnome3 depends on systemd. I mean, why should a window manager / desktop be dependent on any init system? this is seriously wrong, isn't it? |
Yes. It's vertical integration, which just means your software only works with one "back-end"; |
Not necessarily one backend, you can provide different backends:
- dbus or kbus
- dhcp-client or dhcpd
- journald or rsyslog
etc. Systemd is not about "one" backend, but a proper definition of such backend. But there are not many developed yet 
| Quote: | | the reason to do that is usually efficiency, |
in case of Systemd it is more about proper configuration of the whole system in my mind. Though Poettering argues with efficency using Linux specific methods of communication between layers and services.
I have a Debian~unstable installation: Sometimes it runs better than my self compiled Gentoo system: Then I am aware of the amount of experience Debian maintainers put into the system! Such an accumulation of experience and expertise I expect from Systemd.
| Quote: | | but that does not apply when talking about a DE and an init system. They're too far apart; |
How do you define too far apart: By the time a subsystem needs to startup: What about some industrial jboss server, which needs some 20 minutes to come up?
| Quote: | | by the time the DE is starting, you're at the end of the whole init phase. |
No, the init phase is ended when the systemd reaches the shutdown.target.
Systemd acknowledges sober programming principles, for example a stack:
mount a
mount b
umount b
umount a
A proper service management cannot end with some arbitrarily decided end of some phase. It should be possible to start a very different kind of activity (isolate target) at any moment. Some of these activities we are not aware yet they will exist in the future. Hopefully none of them will enslave us human beings ...
| Quote: | | OFC, if a few seconds on boot really matters |
No, seconds really don't matter
| Quote: | | Personally I'd rather employ the work of experts in each, who are used to fitting in with what the user tells them to do |
You are free to develop yourself to such an expert on an issue you are interested in, just like David Herrmann implements his session_vt ideas. |
|
| Back to top |
|
|
ycUygB1
Apprentice
Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon
|
| Posted: Sat Oct 26, 2013 6:52 am Post subject: |
|
|
Systemd boots a lot faster, but half the services don't get started and there is no comprehensive documentation.
So after every boot, I have to enter the error messages into Google and try to figure out how to get networking and other services started. So all in all, systemd takes a lot longer to boot, if you include the Googling to fix the services that didn't start because of the lack of documentation.
I wonder if it is possible to get back to OpenRC, given that Gnome 3 is installed on my system. Even though I am really using Mate. |
|
| Back to top |
|
|
DaggyStyle
Watchman
Joined: 22 Mar 2006
Posts: 5924
|
| Posted: Sat Oct 26, 2013 7:32 am Post subject: |
|
|
| ulenrich wrote: | | LoTeK wrote: | | what disturbs me is that for example gnome3 depends on systemd. I mean, why should a window manager / desktop be dependent on any init system? this is seriously wrong, isn't it? |
If you have one machine for a classroom with three workgroups/seats having each some sessions for the children. What to do if you don't want to use the old unmaintained and broken consolekit?
|
got a multiseat setup with openrc and I never got any problems, what's your point?
_________________
Only two things are infinite, the universe and human stupidity and I'm not sure about the former - Albert Einstein |
|
| Back to top |
|
|
TomWij
Retired Dev
Joined: 04 Jul 2012
Posts: 1553
|
| Posted: Sat Oct 26, 2013 12:38 pm Post subject: Re: systemd or OpenRC, which way is it moving? |
|
|
| hnaparst wrote: | | Code: | | /lib/resolvconf/libc: line 191: /etc/resolv.conf: Permission denied |
It is hard to know where to turn. |
No, it isn't; the permissions on /etc/resolv.conf are set incorrectly, correcting them is the simple solution here. A lot of problems regardless of init system are caused by bad permissions; and this doesn't only happen on Linux, on Windows you get to see the same behavior when you troubleshoot and/or debug using tools like Process Monitor. An installer bailing out, a printer not printing; been there, done that, discovering it is due to a mistake by a random third party developer or something you set yourself for one or another vague reason isn't fun...
Because yeah, permissions aren't only a common cause of errors; they're also a common attempt to fix errors as well, ...
| hnaparst wrote: | | Before I do this, I wonder if I am being a luddite. Which direction are things going? Is systemd the future? Will OpenRC cease to be supported at some point? |
They will continue to exist side by side for a long while; there is no plan to change the default and as far as I know there is no known plan for the OpenRC developers to cease development, the only idea that has been mentioned but not yet revisited is the possibility that we might end up having more specific builds / stage3 to allow people to jump start a systemd installation. Now that there is more support, providing systemd stage3 files next to OpenRC stage3 files is a possibility; yet, there is no concrete plan to do that yet, it's just an idea that has came up once or twice...
Gentoo is a meta distribution; so, one of its main goals is to support as much separate options as possible and where possible such that users can use their system as they see fit.
http://www.gentoo.org/main/en/about.xml
http://www.gentoo.org/main/en/philosophy.xml
http://www.gentoo.org/doc/en/handbook/handbook-amd64.xml?part=1&chap=1
I don't see one init system rule out the other any time soon... |
|
| Back to top |
|
|
ycUygB1
Apprentice
Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon
|
| Posted: Sat Oct 26, 2013 1:16 pm Post subject: Re: systemd or OpenRC, which way is it moving? |
|
|
| TomWij wrote: | | hnaparst wrote: | | Code: | | /lib/resolvconf/libc: line 191: /etc/resolv.conf: Permission denied |
It is hard to know where to turn. |
No, it isn't; the permissions on /etc/resolv.conf are set incorrectly, correcting them is the simple solution here. |
The quote given by the developer is incomplete and out of context, and the solution is wrong. Obviously, the permissions on /etc/resolv.conf are not set according to the desires of dhcpcd, but this was my choice. I got tired of resolv.conf getting overwritten all the time, so I set it as immutable with .
This hardly was the root cause of the complete error message in my original post.
When I said that I didn't know where to turn, I meant that the nscd.service error is not not discussed anywhere that I could see, in the Gentoo docs or anywhere else. Gentoo does not give comprehensive setup advice for systemd, but rather just an intro document. That really doesn't get you too far when the error messages start flying.
I appreciate the answer to my original question, which was whether OpenRC is earmarked for retirement. Perhaps there should be another migration document back to OpenRC and Gnome 2 from systemd/Gnome 3. That path seems to be more difficult, if not impossible. |
|
| Back to top |
|
|
TomWij
Retired Dev
Joined: 04 Jul 2012
Posts: 1553
|
| Posted: Sat Oct 26, 2013 1:45 pm Post subject: |
|
|
Well, I think if you resolve it it would be able to find the file; as I can't imagine fixing it would result in the same output without the line I quoted.
On a side note, have you looked at `journalctl -rb` to see if mentions something as to why it fails?
On another side note, you can prevent dhcpcd from touching /etc/resolf.conf by passing -C resolv.conf to dhcpcd. |
|
| Back to top |
|
|
wuzzerd
Guru
Joined: 05 Jan 2005
Posts: 467
Location: New Mexico
|
| Posted: Sat Oct 26, 2013 2:05 pm Post subject: Re: systemd or OpenRC, which way is it moving? |
|
|
| hnaparst wrote: | | ...I got tired of resolv.conf getting overwritten all the time... |
After pulling out my remaining hair I found two files you can provide:
resolv.conf.head
resolv.conf.tail
dhcpcd will put it's opinion in the middle. |
|
| Back to top |
|
|
666threesixes666
Veteran
Joined: 31 May 2011
Posts: 1248
Location: 42.68n 85.41w
|
| Posted: Sat Oct 26, 2013 2:31 pm Post subject: |
|
|
a bare bones machine running gentoo boots to the point of firefox loaded quicker than windows xp to the point of firefox loaded on the same exact machine.
systemd is sexy, but its clunky. openrc is SOLID. thats why im running it on both of my machines regardless of the boot time. i mean its linux, i have uptimes of over 200 days on embedded machines, who cares about time, i care about performance. no apache on systemd months ago killed it for me. i guess thats been fixed but its not as easy to hop between the two init systems as it previously was. before it was just changing the init=/sbin/systemd to not having that line and back.
my $$$ is on openrc, heading towards proper systemd support when its stable and working with everything out of the box. |
|
| Back to top |
|
|
TomWij
Retired Dev
Joined: 04 Jul 2012
Posts: 1553
|
| Posted: Sat Oct 26, 2013 3:39 pm Post subject: |
|
|
| 666threesixes666 wrote: | | Before it was just changing the init=/sbin/systemd to not having that line and back. |
That's supposed to still work, please clarify what has regressed.
As for Apache, there is apache2.service provided and from what I remember that works; so, if you want to give it another shot... |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Sat Oct 26, 2013 4:27 pm Post subject: |
|
|
| ulenrich wrote: | @steveL, excuse me I answer to this one, it's more fundamental:
| steveL wrote: | | LoTeK wrote: | | what disturbs me is that for example gnome3 depends on systemd. I mean, why should a window manager / desktop be dependent on any init system? this is seriously wrong, isn't it? |
Yes. It's vertical integration, which just means your software only works with one "back-end"; |
Not necessarily one backend, you can provide different backends:
- dbus or kbus
- dhcp-client or dhcpd
- journald or rsyslog
etc. Systemd is not about "one" backend, but a proper definition of such backend. But there are not many developed yet :( |
I don't see what's so fundamental about that: I've never heard of kbus, and dbus is still very much imposed on my system using KDE. dhcpd is simply a service afaic, and journald not being required is also new to me, but again you're just talking about which services systemd supports, not which backend Gnome3 works with.
Although from recent reading on debian-devel, Gnome3 works on BSD currently, they didn't sound that hopeful of continuing with that, and it does effectively enforce a requirement for systemd by stating that everything it needs for a "modern system" must be provided using the systemd APIs. Certainly there seems a move to state that on Linux at least, it will not work without systemd as pid 1.
So I hope you can see that Gnome3 is vertically-integrated with systemd, and not get side-tracked into whether an init system supports more than one service, which is irrelevant.
| Quote: | | Quote: | | the reason to do that is usually efficiency, |
in case of Systemd it is more about proper configuration of the whole system in my mind. Though Poettering argues with efficency using Linux specific methods of communication between layers and services. |
You're off on a tangent, since the integration point was not about systemd, but about Gnome, but it's interesting that Poettering argues the efficiency viewpoint.
| Quote: | | I have a Debian~unstable installation: Sometimes it runs better than my self compiled Gentoo system: Then I am aware of the amount of experience Debian maintainers put into the system! Such an accumulation of experience and expertise I expect from Systemd. |
An expectation I do not share. Their behaviour can only be described as amateurish, afaic. You're welcome to your opinion, and I'm sure its users have enough collective experience to make it work, but it seems to me like it will take a lot of man-hours, for a flawed design.
| Quote: | | Quote: | | but that does not apply when talking about a DE and an init system. They're too far apart; |
How do you define too far apart: |
If you didn't cut my sentence in half you'd see how. This really is disingenuous and verges on trolling afaic.
| Quote: | | By the time a subsystem needs to startup: What about some industrial jboss server, which needs some 20 minutes to come up? |
What about it? It's hardly the same as my laptop or desktop.
| Quote: | | Quote: | | by the time the DE is starting, you're at the end of the whole init phase. |
No, the init phase is ended when the systemd reaches the shutdown.target. |
LMAO.
And oh wow, it uses a stack! Such amazing craftmanship. As for confusing service management with system init, words fail me.
| Quote: | | Quote: | | OFC, if a few seconds on boot really matters |
No, seconds really don't matter :) |
Glad we agree: please admit that it has been used as a major USP of systemd, when most people on these forums basically don't care, since their machines really don't need to reboot that often.
| Quote: | | Quote: | | Personally I'd rather employ the work of experts in each, who are used to fitting in with what the user tells them to do |
You are free to develop yourself to such an expert on an issue you are interested in, just like David Herrmann implements his session_vt ideas. |
You appear to be confused, as well as patronising: please re-read my sentence. I don't need to be an expert in any particular field, to employ the services of an expert in that field. That's the whole point of placing trust in someone else's software. And like it or not, respect for the craft, and an ability to work multi-platform, as well as not breaking userspace with constant idiocy, are indicators of whom to trust, if we are not talking about tried and tested software that is still around after 20 years. The latter indicates it is good, not that it is bad: it survived the review (or QA) process we are currently applying to "modern" software, most of which will fall by the wayside and be forgotten in another 20 years.
Although I have a feeling Poetteringware will be remembered as the first true exemplar of "N + 1 True Way".
Outside Windows that is.
_________________
| creaker wrote: | | systemd. It is a really ass pain |
update - "a most excellent portage wrapper"
#friendly-coders -- We're still here for you™ ;) |
|
| Back to top |
|
|
mrbassie
l33t
Joined: 31 May 2013
Posts: 815
Location: over here
|
| Posted: Sat Oct 26, 2013 5:15 pm Post subject: |
|
|
| hnaparst wrote: | Systemd boots a lot faster, but half the services don't get started and there is no comprehensive documentation.
So after every boot, I have to enter the error messages into Google and try to figure out how to get networking and other services started. So all in all, systemd takes a lot longer to boot, if you include the Googling to fix the services that didn't start because of the lack of documentation.
I wonder if it is possible to get back to OpenRC, given that Gnome 3 is installed on my system. Even though I am really using Mate. |
When I tried it out a couple of times it only knocked about a second off my boot time. Maybe it's different if you've got tons of services starting at boot, I don't know. Now shutting down, that was noticably faster.
I also found it a massive pain, couldn't get wifi working, couldn't get help on here, was pretty much told to bugger off when I asked on the Arch forum and gave up. I was just having a look at it really, didn't like it, it seems unnecessarily complicated to configure for what seemed to me no real benefit. |
|
| Back to top |
|
|
666threesixes666
Veteran
Joined: 31 May 2011
Posts: 1248
Location: 42.68n 85.41w
|
| Posted: Sat Oct 26, 2013 5:19 pm Post subject: |
|
|
i don't care what anyone has to say, poettering is a genius. @wij im too busy building wiki pages to re hash that at the moment.
from what i remember removal of udev was stumping me, and i know that systemd is a full on drop in replacement for udev now. ill let others work on that stuff for a bit before i have another go at it, but for the time being i see undocumented things as a priority. i need to figure out like 10 things before ill return to systemd's glory. 1 is wrapping gentoo for mass deployment & pixie mass deployment of said images. for work stations & kiosk machines... & better lockdown documentation for these situations....
i want to get gentoo corporate feasible & ready. more godlike linux admins looking at our systems is a good thing from my perspective.
@mrbassie wifi worked for me under network manager but i had to load xfce through gdm / xdm / kdm. |
|
| Back to top |
|
|
kurly
Apprentice
Joined: 02 Apr 2012
Posts: 260
|
| Posted: Sat Oct 26, 2013 9:55 pm Post subject: |
|
|
| 666threesixes666 wrote: | | i don't care what anyone has to say, poettering is a genius. |
Many people, myself included, disagree with you for a wide variety of reasons. It is unclear why you do not care. Why even post in a public forum if you don't care what others think?
It has been demonstrated repeatedly through the forums and elsewhere that his ideas are well-intended but fall apart on closer examination.
| 666threesixes666 wrote: | | @wij im too busy building wiki pages to re hash that at the moment. |
An actual developer is asking you an actual question, and you are dodging it so that you can go throw more words on the wiki. Lovely. (Remember the talk we had about quantity vs quality? You have much potential: Use it.)
| 666threesixes666 wrote: | i need to figure out like 10 things before ill return to systemd's glory. 1 is wrapping gentoo for mass deployment & pixie mass deployment of said images. for work stations & kiosk machines... & better lockdown documentation for these situations....
i want to get gentoo corporate feasible & ready. more godlike linux admins looking at our systems is a good thing from my perspective. |
The people who need to do these things are hopefully competent individuals who will not need to refer to the guesswork that often makes its way into your articles. Mass deployment of Gentoo is already a solved problem, and so is PXE deployment (I assume that's what you meant?) of said images.
It is clear that you want to help, but documenting that which you do not understand is NOT HELPFUL. I am sorry to be the bearer of bad news. |
|
| Back to top |
|
|
ycUygB1
Apprentice
Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon
|
| Posted: Sun Oct 27, 2013 11:45 am Post subject: |
|
|
| mrbassie wrote: |
I also found it a massive pain, couldn't get wifi working, couldn't get help on here, was pretty much told to bugger off when I asked on the Arch forum and gave up. I was just having a look at it really, didn't like it, it seems unnecessarily complicated to configure for what seemed to me no real benefit. |
Exactly my point. Maybe in some theoretical sense, systemd is better. But if it isn't documented and no one is supporting it, and if you didn't attend the meetings at MIT where they made the design decisions, then one really has no hope of using it.
Personally, after eight years of using Gentoo, I feel like I am finally ready to contribute something to the community. I am thinking of writing a migration document: from Systemd/Gnome 3 back to OpenRC/Gnome2. I feel so much better now. It is amazing what you can accomplish beginning with
|
|
| Back to top |
|
|
nix213
n00b
Joined: 08 Feb 2013
Posts: 23
Location: Illinois
|
| Posted: Sun Oct 27, 2013 12:54 pm Post subject: |
|
|
systemd and OpenRC aren't mutually exclusive. According to OpenRC's wiki:
| Quote: | | It is not a replacement for /sbin/init |
although, on systemd's wiki:
| Quote: | In April 2012, the source tree for udev (a device manager that interoperates with the Linux kernel) was merged into systemd.[6]
Systemd obsoletes ConsoleKit[7] and provides a replacement for sysvinit, pm-utils, inetd, acpid, syslog, watchdog, cron and atd. |
Note what wikipedia has to say about OpenRC as well, "Its creator is a NetBSD developer, who started the Gentoo/FreeBSD project."
_________________
"Wherever you go, go with all your heart." -Confucius |
|
| Back to top |
|
|
_______0
Guru
Joined: 15 Oct 2012
Posts: 521
|
| Posted: Sun Oct 27, 2013 3:17 pm Post subject: |
|
|
one time I managed to boot gentoo in 3 seconds, in a vm though. The only thing keeping it from sub-second boot was net.lo. Kept waiting for I dunno what but I remember it was the only problem.
I forget whether this was before openrc or with openrc. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Portage & Programming
