View previous topic :: View next topic |
Author |
Message |
Naib Watchman
Joined: 21 May 2004 Posts: 6069 Location: Removed by Neddy
|
Posted: Mon Sep 23, 2019 8:54 am Post subject: |
|
|
steve_v wrote: | erm67 wrote: | Actually the idea of a "home-on-a-stick" (maybe ro) that doesn't break stuff and is automatically handled is not bad ..... | I'm pretty certain I could implement that without systemd or JSON user data if I could be bothered... | you probably could. I probably could... it would just be a squashfs and a loopback file.
The problem is it hasn't been done, it hasn't been accepted, it hasn't been deployed. I am sure there are people who have their own bespoke method which works for them BUT its not formalized and more importantly not released.
Out of everything systemd has done, this is one thing that is directly visible to joe-average. All the "low level" stuff is exactly that, low-level. If someone wants to they can tinker and see but the vast majority of linux users are USERS and thus the specifics of systemd are moot, as long as it works.
Systemd has been accepted and deployed by soo many distro's now it is ridiculous, for them to provide means to easily move /home/foo would be a major plus point for end-users as it is something immediately visible and usable. With systemd on soo many distro's this is then the common platform/implementation that would be accepted.
This is something that needs to replicated as a separate service for non-systemd systems as very quickly migration from say ArchLinux to Gentoo would become less convenient because ArchLinux -> Ubuntu would be easy but Archlinux -> Gentoo would require "archaic" steps
Even if you were or someone else were to release their own migratable loopback home service, how and why would other distro's adopt it? especially as it will come for "free" with systemd.
The cynic in me is seeing a nasty aim to box non-systemd out. The monoculture will be hard to unpick and will take more resource to maintain. eudev was the 1st, elogind the second... how many more "critical" system type services need to be de-systemd to ensure a reasonable amount of compatibility across distributions? Even if you do not use Systemd, it is influencing the system architecture (without any system engineering) _________________ #define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54744 Location: 56N 3W
|
Posted: Mon Sep 23, 2019 9:03 am Post subject: |
|
|
Naib,
systemd has been about vendor lock in from day one.
As soon as you have lock in, lock out comes for free.
Go back to your Apollo workstation, or any other pre PC era workstation. _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
Naib Watchman
Joined: 21 May 2004 Posts: 6069 Location: Removed by Neddy
|
Posted: Mon Sep 23, 2019 9:21 am Post subject: |
|
|
NeddySeagoon wrote: | Naib,
systemd has been about vendor lock in from day one.
As soon as you have lock in, lock out comes for free.
Go back to your Apollo workstation, or any other pre PC era workstation. | I agree, but now it is something that the end-user would request NOT just system admins. Completely new front _________________ #define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54744 Location: 56N 3W
|
Posted: Mon Sep 23, 2019 9:35 am Post subject: |
|
|
Naib,
Its something corporate would enforce on end users.
Most home users don't care about security and leaving laptops lying around. Just like they don't care about systemd.
Just as long as it works.
Look at the pushback to the introduction of usernames and passwords on Windows a while ago. _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
steve_v Guru
Joined: 20 Jun 2004 Posts: 416 Location: New Zealand
|
Posted: Mon Sep 23, 2019 9:42 am Post subject: |
|
|
Naib wrote: | The problem is it hasn't been done, it hasn't been accepted, it hasn't been deployed. I am sure there are people who have their own bespoke method which works for them BUT its not formalized and more importantly not released. | True, but I'd be inclined to say that the reason it hasn't been implemented is because there is little interest in such a feature.
Naib wrote: | This is something that needs to replicated as a separate service for non-systemd systems as very quickly migration from say ArchLinux to Gentoo would become less convenient because ArchLinux -> Ubuntu would be easy but Archlinux -> Gentoo would require "archaic" steps | People have been keeping their home directories during a distro move or reinstall since more than one distro to choose from became a thing. Copying over the files and chowning them or giving yourself the same UID isn't difficult, and it doesn't sound particularly archaic either.
What's wrong with just putting /home on it's own partition anyway? Easy-mode GUI installers are a thing, so joe average barely needs to know what a partition table is.
Naib wrote: | The cynic in me is seeing a nasty aim to box non-systemd out. The monoculture will be hard to unpick and will take more resource to maintain. eudev was the 1st, elogind the second... how many more "critical" system type services need to be de-systemd to ensure a reasonable amount of compatibility across distributions? Even if you do not use Systemd, it is influencing the system architecture (without any system engineering) | I don't know so I won't speculate, but I have a feeling this will go exactly the same way as the rest of systemd went, regardless of intent. _________________ Once is happenstance. Twice is coincidence. Three times is enemy action. Four times is Official GNOME Policy. |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
Posted: Mon Sep 23, 2019 9:48 am Post subject: |
|
|
erm67 wrote: | Actually the idea of a "home-on-a-stick" (maybe ro) that doesn't break stuff |
That's the key, unfortunately from what we've seen of other aspects of LP's coding/ideas, things will get broken.
But if you want to run it, be our guest, let us know how it works out, the next systemd should have it in it.
Edit to add: the more they try and make systemd like windows the less inclined I am to ever run it on any system that I have a bit of control over.
What other do is no concern of mine though, so I'm not arguing against it for anyone that wants to run it. _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland |
|
Back to top |
|
|
NeddySeagoon Administrator
Joined: 05 Jul 2003 Posts: 54744 Location: 56N 3W
|
Posted: Mon Sep 23, 2019 10:00 am Post subject: |
|
|
Anon-E-moose,
You should be concerned. Our problems will start when the rest of the ecosystem requires systemd.
We have already had a taste of that with Gnome. It will only get worse.
There are other examples too. _________________ Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
Back to top |
|
|
krinn Watchman
Joined: 02 May 2003 Posts: 7470
|
Posted: Mon Sep 23, 2019 10:28 am Post subject: |
|
|
Naib wrote: | This is something that needs to replicated as a separate service for non-systemd systems as very quickly migration from say ArchLinux to Gentoo would become less convenient because ArchLinux -> Ubuntu would be easy but Archlinux -> Gentoo would require "archaic" steps |
I'm even less sure than you that migration from ArchLinux -> Ubuntu would be easy, with a distro specific fingerprint, you could prevent this to only allow migration from one OS to another version of itself only.
Remember bios secureboot was made for "security".... |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
|
Back to top |
|
|
Naib Watchman
Joined: 21 May 2004 Posts: 6069 Location: Removed by Neddy
|
Posted: Mon Sep 23, 2019 11:52 am Post subject: |
|
|
krinn wrote: | Naib wrote: | This is something that needs to replicated as a separate service for non-systemd systems as very quickly migration from say ArchLinux to Gentoo would become less convenient because ArchLinux -> Ubuntu would be easy but Archlinux -> Gentoo would require "archaic" steps |
I'm even less sure than you that migration from ArchLinux -> Ubuntu would be easy, with a distro specific fingerprint, you could prevent this to only allow migration from one OS to another version of itself only.
Remember bios secureboot was made for "security".... | you need to drink moar kool-aid. Zipping/mounting/rsync a /home directory is the simplest action and the real problem comes from different distributions having different versions of different software such that the associated dotfiles are incompatible.
This won't change the PR nor the blame when it doesn't work
Interesting read and not really about INIT
Quote: | This is a DPL problem because we can't get the right people together to
make progress. It's not an easy problem. Developers don't have to do
any work: if the systemd maintainers are emotionally exhausted and don't
want to deal with this, they don't have to. (Although if the project is
committed to init diversity, they cannot stand in the way.)
And yet the systemd maintainers and to a lesser extent release team face
conduct that is frankly unacceptable. And in some cases that conduct is
the frustrated reaction to years of interactions complex enough that
we'll never untangle them. No matter how unfortunate the conduct is,
the frustrations, anger and hurt are real. |
It would appear certain behaviour is starting to show signs of not being tolerated for much more... _________________ #define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
Back to top |
|
|
krinn Watchman
Joined: 02 May 2003 Posts: 7470
|
Posted: Mon Sep 23, 2019 1:16 pm Post subject: |
|
|
Naib wrote: | you need to drink moar kool-aid. Zipping/mounting/rsync a /home directory is the simplest action and the real problem comes from different distributions having different versions of different software such that the associated dotfiles are incompatible. |
an issue homed won't fix anyway, but i was refering to user home encryption, that you could lock in to a specific distro |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
Posted: Mon Sep 23, 2019 4:24 pm Post subject: |
|
|
Naib wrote: |
Interesting read and not really about INIT |
No, it's not but that's how he described it (roll eyes not for you, for the author)
Having read the whole thing there's some screwy arguments being brought forward.
No one in their right mind would try running elogind and systemd at the same time, systemd has it's own login manager.
Arguments about polkit, not working well with elogind and something about systemd journaling, again WTF?
Then segue down to well maybe we should just get rid of init scripts and sysvinit altogether.
Quote: | Additional complexity comes in because the elogind in unstable tries to
use apt dependencies/conflicts/replaces/provides to switch out the
elogind library for the systemd libraries. |
Again, what kind of moron wants to try to run elogind alongside systemd.
Edit to add: It's not even allowed on gentoo, you can't install them together
Either the elogind maintainer @debian is a complete moron or maybe he just doesn't understand elogind at all,
but most of us on this thread could demolish the "arguments" they have brought up in a few sentences.
And of course the systemd maintainer, says it's too emotional to even explain the problems along with taking weeks/months to respond.
Basically a bunch of prima-donna's thinking they're gods gift to the distro world, talk about bit-rot.
Just more ammunition for the continuation of devuan as an alternative to debian. _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland
Last edited by Anon-E-moose on Mon Sep 23, 2019 6:23 pm; edited 1 time in total |
|
Back to top |
|
|
GDH-gentoo Veteran
Joined: 20 Jul 2019 Posts: 1791 Location: South America
|
Posted: Mon Sep 23, 2019 7:29 pm Post subject: |
|
|
Anon-E-moose wrote: | Naib wrote: |
Interesting read and not really about INIT |
No, it's not but that's how he described it (roll eyes not for you, for the author) |
In Debianspeak, "init system diversity" means something quite modest for Gentoo standards: supporting anything other that systemd But agreed, the e-mail largely refers to the "elongind in Debian testing" saga (and tangentially on systemd unit files vs the scripts for the horrible Debian rc subsystem).
Anon-E-moose wrote: | Arguments about polkit, not working well with elogind and something about systemd journaling, again WTF? |
That's just the 'joy' of binary-based distributions once more. The easy way to have polkit support elogind, i.e. rebuilding it, meant providing different binary packages built with different compile-time options (which I think is still what Devuan does), which polkit Debian maintainers were not precisely keen on doing.
Debian bug #923244, "'policykit-1: Please support elogind backend"
Instead, the Devuan people managed to convince elogind's maintainer to make libelogind ABI-compatible with libsystemd, so that it could be a drop-in replacement in Debian. And, in turn, sparing Devuan from having to repackage polkit themselves.
Anon-E-moose wrote: | Either the elogind maintainer @debian is a complete moron or maybe he just doesn't understand elogind at all |
Mark Hindley and the "Debian init diversity" group? I don't think so. He and others are just fighting an uphill battle against the Debian bureaucracy. And trying to remain civil, which is quite commendable. If anyone has the time, grab the popcorn:
Debian bug #940034, "libelogind0: replacing a core system library and conflicting against the default init considered harmful"
Debian bug #934132, "unblock: elogind/241.3-1+debian1"
Debian bug #935304, "libpam-systemd: Please relax Depends: systemd-sysv"
Debian bug #930105, "systemd: prerm fail breaks apt and renders system hard to recover"
Anon-E-moose wrote: | Just more ammunition for the continuation of devuan as an alternative to debian. |
I hope they are successful, but from a technical standpoint, they are in a rather complicated position. At the end of the day, they are a derivative distribution, they want to be systemd-free, but their base distribution is simultaneously:
- binary-based
- heavy on policy.
- and has systemd as its main supported software package.
Oh, and Hindley is actually trying to help. |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
Posted: Mon Sep 23, 2019 9:54 pm Post subject: |
|
|
Sounds like debian is slowly trying to be irrelevant.
If everyone out there is a clone of RH with no redeeming value added, is there a need for any other distro?
Edit to add: and I do retract my heavy-handed statement about the elogind maintainer on debian (I was overboard) _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
Posted: Mon Sep 23, 2019 11:35 pm Post subject: |
|
|
In addition to announcing systemd-homed for better user home directories, Lennart Poettering also used this year's All Systems Go conference to drum up support for systemd's boot efforts around SD-Boot and the Boot Loader Specification.
systemd-boot/sd-boot is systemd's UEFI boot manager formerly known as Gummiboot. SD-Boot continues picking up new functionality and at least optional usage by more distributions. The Systemd Boot Loader Specification (also known as the FreeDesktop.org Boot Loader Specification) meanwhile is trying to assist use-cases around dual/multi-boot operating system setups and related use-cases with drop-in file handling, standardized configuration files and the like. _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland |
|
Back to top |
|
|
Zucca Moderator
Joined: 14 Jun 2007 Posts: 3888 Location: Rasi, Finland
|
Posted: Tue Sep 24, 2019 10:51 am Post subject: |
|
|
Instead of using image file for home directories why don't just use eCryptfs, EncFS or similar?
To me, using an image file that resides on CoW fs causes quite a lot of overhead.
Am I missing something here, or is this just plain bad idea? _________________ ..: Zucca :..
My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
Quote: | I am NaN! I am a man! |
|
|
Back to top |
|
|
krinn Watchman
Joined: 02 May 2003 Posts: 7470
|
Posted: Tue Sep 24, 2019 5:07 pm Post subject: |
|
|
Zucca wrote: | Am I missing something here, or is this just plain bad idea? |
The control...
If you patch these programs to tied them with systemd, people might gets totally crazy and may reject your patches.
But if you recreate what exists, you can do whatever shit you want with your work ; and your fellow fanatics will also gets totally happy with something "new and that kick ass" : in case anyone have doubt, you could still create a webpage at freedesktop.org to show them why other programs are buggy, old, sucks... |
|
Back to top |
|
|
Hu Administrator
Joined: 06 Mar 2007 Posts: 23028
|
Posted: Wed Sep 25, 2019 12:45 am Post subject: |
|
|
Zucca wrote: | Instead of using image file for home directories why don't just use eCryptfs, EncFS or similar?
To me, using an image file that resides on CoW fs causes quite a lot of overhead.
Am I missing something here, or is this just plain bad idea? | It provides very easy data portability. If you want to move your home directory to another system, unmount it and copy one file (the looped container) over, then you're done. No need to mess with recursive tar/rsync, or the right options to copy all the right bits of metadata. It's implicitly a quota system of sorts, since your home directory cannot outgrow the current container file. With traditional systems, one user can take over all of /home if not stopped by a filesystem quota.
Having written all that, I still think it's a bad idea. It's not without some merits, but overall, it's a change for the worse. |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
Posted: Wed Sep 25, 2019 9:25 am Post subject: |
|
|
There's only a few things in common between my desktop home and my laptop .bash*, one or two dir's from .config and one or two scripts and that's about it.
For the rest, it's not shared, not compatible, nor do I want it to be.
Everyone is different and that's the problem, it might be a great idea for LP, but not so much for many others. _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland |
|
Back to top |
|
|
Zucca Moderator
Joined: 14 Jun 2007 Posts: 3888 Location: Rasi, Finland
|
Posted: Wed Sep 25, 2019 12:48 pm Post subject: |
|
|
Anon-E-moose wrote: | There's only a few things in common between my desktop home and my laptop .bash*, one or two dir's from .config and one or two scripts and that's about it. | ++
Even some aliases don't make sense on my other platforms. It's even worse when I use different distro on my laptop now. It's practically impossible to have, for example, same version of every software on all of my platforms.
I really hope this $HOME as a disk image isn't forced later on. Just imagine editing a video of several gigabytes on a crypted loopback image that sits on a CoW filesystem. With btrfs I could just disable CoW for the loopback image, but then I lose snapshotting of the whole $HOME. _________________ ..: Zucca :..
My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
Quote: | I am NaN! I am a man! |
|
|
Back to top |
|
|
saellaven l33t
Joined: 23 Jul 2006 Posts: 655
|
Posted: Wed Sep 25, 2019 3:24 pm Post subject: |
|
|
My existing /home traces back to 1994. It has survived numerous hard drives, having originated on a 40MB ST351A/X, moving from IDE to SCSI to SATA and finally sitting on a SSD. It's undergone a dozen CPU/motherboard changes and an architecture change from x86->amd64.
I've never had a problem migrating it from computer to computer or drive to drive.
I have backups of it in various forms, and, like Anon, what is on my portable devices greatly differs from what it on my home desktop vs what is on my business desktop, though I have encrypted access to and from all of those platforms when I need access to something a certain computer doesn't natively contain (my banking info exists in an encrypted file on my home desktop, there's no reason to have it reside on my laptop when I'm traveling, particularly given that my laptop can ssh into my desktop to access it when needed).
I've also never had a problem using a remote desktop through X when I wanted to (but the same FDO gang started out Wayland with a design goal to eliminate that use case too). The same can be said of the way GNOME 2 was crippled compared to GNOME 1 and then GNOME 3 crippled compared to GNOME 2.
The entire point of what they're doing isn't to just cripple everyone into doing things the one way they approve of, it's to force you to become dependent upon them when they break your work flow and/or needs.
I've said it from the beginning and I'll continue to say it again now - the goal of Red Hat, Lennart (a RH employee), and friends is to make RH the one distribution that matters and that Debian, Ubuntu, etc, are destined to become pointless RH clones with a slightly different flavoring on top... at which point, why use anything other than RH if the purpose of using a distro instead of rolling your own is to get support?
In this way, Red Hat is already 90% of the way to completing something that Microsoft and SCO dreamed of but could never manage to do - completely co-opting Linux. RH accomplished it by becoming a dominant player from the start, becoming the first to file an IPO, hiring or outright buying core developers and companies with that IPO money (Alan Cox, Linus' former right hand, was a key kernel dev and one of RH early employees, they bought out Cygnus to gain control over the core tool stack (gcc, gdb, binutils, etc)), threw their weight around with groups like the Linux Foundation, FDO, and others, and they managed to do so because few people initially saw them as a threat from within.
Those of us making a stand for openness and allowing people to do things their own way are called anachronistic, luddites, dinosaurs, etc... but we do it with a purpose - not just for ourselves, but to try to pass on the freedoms that were given to us which allowed us the opportunity to grow. I don't use Linux to do things the one true way - if that's what I wanted, why wouldn't I have been using the dominant desktop OS for the last few decades - Windows? I use Linux because it's flexible enough to let me adapt it to my needs rather than forcing me to adapt to the approved, supported way it wants to make me use it. _________________ Ryzen 3700X, Asus Prime X570-Pro, 64 GB DDR4 3200, GeForce GTX 1660 Super
openrc-0.17, ~vanilla-sources, ~nvidia-drivers, ~gcc |
|
Back to top |
|
|
Tony0945 Watchman
Joined: 25 Jul 2006 Posts: 5127 Location: Illinois, USA
|
Posted: Wed Sep 25, 2019 4:27 pm Post subject: |
|
|
Richard Stallman's forced resignation also means the end of the FSF vision. |
|
Back to top |
|
|
Naib Watchman
Joined: 21 May 2004 Posts: 6069 Location: Removed by Neddy
|
Posted: Wed Sep 25, 2019 7:03 pm Post subject: |
|
|
*Sigh*
https://www.theregister.co.uk/2019/09/25/systemd_inventor_home_directories/
Quote: |
There are some complications, one of which is remote access via SSH.
"If you authenticate via SSH it goes via authorized keys in the home directory. So if you want to authenticate something that is inside of the home directory, so that it can access the home directory, where does the decryption key come from, to access the home directory? It is a chicken-and-egg problem," said Poettering. |
... _________________ #define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0; |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6200 Location: Dallas area
|
Posted: Wed Sep 25, 2019 7:45 pm Post subject: |
|
|
Naib wrote: | *Sigh*
https://www.theregister.co.uk/2019/09/25/systemd_inventor_home_directories/
Quote: |
There are some complications, one of which is remote access via SSH.
"If you authenticate via SSH it goes via authorized keys in the home directory. So if you want to authenticate something that is inside of the home directory, so that it can access the home directory, where does the decryption key come from, to access the home directory? It is a chicken-and-egg problem," said Poettering. |
... |
Well, he can fix it with the new and improved systemd-sshd
Or just replace that clunky old ssh with new and improved dbusd, the enterprise version. _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland |
|
Back to top |
|
|
axl Veteran
Joined: 11 Oct 2002 Posts: 1146 Location: Romania
|
Posted: Wed Sep 25, 2019 9:42 pm Post subject: |
|
|
Anon-E-moose wrote: | improved systemd-sshd |
i thought sshd.socket was the improved systemd-sshd. hmm. |
|
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
|
|