Gentoo Forums :: View topic - systemd-homed ???

systemd-homed ???

View unanswered posts
View posts from last 24 hours

Goto page Previous 1, 2, 3, 4, 5, 6, 7 Next

Gentoo Forums Forum Index

Gentoo Chat

View previous topic :: View next topic

Author

Message

Gatsby
Tux's lil' helper

Joined: 18 Jan 2010
Posts: 121
Location: 127.0.0.1

Posted: Tue Jan 07, 2020 4:20 pm Post subject:

erm67 wrote:

...

Si tacuisses, philosophus mansisses.
_________________
"Its your Gentoo, your way. When it breaks, you can keep all the pieces."
-- NeddySeagoon@forums.gentoo.org

Last edited by Gatsby on Tue Jan 07, 2020 7:49 pm; edited 1 time in total

proteusx
Guru
Guru

Joined: 21 Jan 2008
Posts: 340

Posted: Tue Jan 07, 2020 4:42 pm Post subject:

erm67 wrote:

You mean you'll stick witk glep-0081?

Gentoo works fine without the glep-81 silliness.

dmpogo
Advocate

Joined: 02 Sep 2004
Posts: 3461
Location: Canada

Posted: Tue Jan 07, 2020 7:18 pm Post subject:

tholin wrote:

The threat model he mention is "I if go though customs to a country that I don't trust" but there is no way this approach is secure against that. You simply don't know what sensitive data remains in ram. Don't he realize that? Or if he is realizing it why is he claiming this is secure?

This may lead to an interesting technical discussion but is really a bogus argument in real life. Customs simply demand that you enter the password to decrypt your drive and if you refuse, can just deny your admission (and that would be your best outcome, the denial of admission to US say can easily come with 5 year ban on entering US).
And if you are a returning citizen, and the admission can not be denied, they are entitled (for instance in Canada) to temorarily seize your device and send it to their lab for
analysis. You'll get it back, in few months. All of it is such a hassle that is justified only if you have really somethng serious on your drive. And if so you should not be taking
your /home to the border. Really if one is concerned about the border, one should wipe the disks/phones, and restore it from online backups on the other side, if possible.
Which can also be monitored, of course, but perhaps less personally.

Last edited by dmpogo on Tue Jan 07, 2020 7:21 pm; edited 1 time in total

Naib
Watchman

Joined: 21 May 2004
Posts: 6069
Location: Removed by Neddy

Posted: Tue Jan 07, 2020 7:19 pm Post subject:

Hu wrote:

Now we need someone to propose an outlandish systemd feature that, if ever implemented, would bring about the downfall of systemd. Being outlandish, no one will seriously expect it to actually happen. Then, one day it will, because so far truth is stranger than fiction.

pjp wrote:

Hu wrote:

the downfall of systemd

That seems unlikely at this point unless MS ports AD to Linux, thereby gifting IBM fond memories of their OS/2 days.

raise a feature request pushing for AD capability in SystemD
_________________
#define HelloWorld int
#define Int main()
#define Return printf
#define Print return
#include <stdio>
HelloWorld Int {
Return("Hello, world!\n");
Print 0;

erm67
l33t
l33t

Joined: 01 Nov 2005
Posts: 653
Location: EU

Posted: Tue Jan 07, 2020 8:22 pm Post subject:

Gatsby wrote:

erm67 wrote:

...

Si tacuisses, philosophus mansisses.

Yeah but I got this today after an update:

Code:

Scaricamento di:1 http://deb.debian.org/debian buster-backports/main armel libnss-systemd armel 244-3~bpo10+1 [194 kB]

And at the same time I noticed the messages below:

proteusx wrote:

erm67 wrote:

You mean you'll stick witk glep-0081?

Gentoo works fine without the glep-81 silliness.

How do you stop those funny packages from emerging?

Code:

1578392612: ::: completed emerge (84 of 533) acct-user/messagebus-0 to /
1578392636: ::: completed emerge (85 of 533) acct-user/systemd-journal-remote-0 to /
1578392660: ::: completed emerge (86 of 533) acct-user/systemd-coredump-0 to /
1578392683: ::: completed emerge (87 of 533) acct-user/systemd-network-0 to /
1578392707: ::: completed emerge (88 of 533) acct-user/systemd-resolve-0 to /
1578392731: ::: completed emerge (89 of 533) acct-user/systemd-timesync-0 to /
1578392815: ::: completed emerge (92 of 533) acct-user/man-1 to /
1578392878: ::: completed emerge (94 of 533) acct-user/mysql-0 to /
1578392902: ::: completed emerge (95 of 533) acct-user/avahi-0 to /
1578392926: ::: completed emerge (96 of 533) acct-user/mail-0 to /
1578392950: ::: completed emerge (97 of 533) acct-user/postmaster-0 to /
1578392973: ::: completed emerge (98 of 533) acct-user/redis-0 to /
1578392997: ::: completed emerge (99 of 533) acct-user/fetchmail-0 to /
1578393021: ::: completed emerge (100 of 533) acct-user/ddclient-0 to /
1578393045: ::: completed emerge (101 of 533) acct-user/mpd-0 to /

Apparently systemd-sysusers.d and glep-0081 have overlapping goals how will the conflict be solved?
When upstream wil begin installing droplets with user and goup info into /etc/sysusers.d portage will intercept them and create glep-0081 packages automatically for non systemd users? And is this a lucky coincidence or something else?
_________________
Ok boomer
True ignorance is not the absence of knowledge, but the refusal to acquire it.
Ab esse ad posse valet, a posse ad esse non valet consequentia

My fediverse account: @erm67@erm67.dynu.net

GDH-gentoo
Veteran
Veteran

Joined: 20 Jul 2019
Posts: 1778
Location: South America

Posted: Tue Jan 07, 2020 10:21 pm Post subject:

erm67 wrote:

Apparently systemd-sysusers.d and glep-0081 have overlapping goals how will the conflict be solved?

Short term? I'd say GLEP 81 wins, and /etc/sysusers.d/* files get ignored. GLEP 81 ebuilds integrate with the package manager, and there's currently no systemd-emerge

erm67 wrote:

When upstream wil begin installing droplets with user and goup info into /etc/sysusers.d portage will intercept them and create glep-0081 packages automatically for non systemd users?

Depends, who is sysusers.d's 'audience', upstreams or distributions?

Morality124
Tux's lil' helper

Joined: 20 Feb 2018
Posts: 102

Posted: Tue Jan 07, 2020 11:00 pm Post subject:

erm67 wrote:

How do you stop those funny packages from emerging?

nuke-acct-shit.sh:

#!/bin/sh

set -e
err_handler () { [ $? -eq 0 ] && exit; }
trap err_handler EXIT

portageDir="$(awk '
BEGIN {p = "PORTDIR="} $0 ~ p {
gsub(p "|\"|\047", "")
print
}' /etc/portage/make.conf)"
packageProvided=/etc/portage/profile/package.provided
mkdir -pv $packageProvided
tempDir=$(mktemp -d --suffix=_acct-shit)

for suffix in group user
do
find "$portageDir"/acct-$suffix/* -type d | \
awk -F '/' -v suffix="$suffix" -v packageProvided="$packageProvided" '
   (!/systemd/) {
      print
      print $(NF - 1) "/" $NF "-0" > packageProvided "/acct-" suffix ".provided"
   }'

mv -v "$portageDir"/acct-$suffix "$tempDir"
done

Last edited by Morality124 on Wed Jan 08, 2020 3:58 am; edited 3 times in total

saellaven
l33t
l33t

Joined: 23 Jul 2006
Posts: 655

Posted: Tue Jan 07, 2020 11:06 pm Post subject:

erm67 wrote:

Last month, Gentoo devs were debating again how they should go about implementing GLEP-81 and whether or not they should change their approach. It's another case of "DO then think" about the consequences later, then maybe revise, but meanwhile, constantly screw up production systems in the meantime.

Given the latest systemd news and the history of trying to force systemd's caveats on non-systemd users, I expect GLEP-81 to be revised in a way to force a systemd approach. I say that half tongue in cheek, but I wouldn't be surprised to see someone propose it.
_________________
Ryzen 3700X, Asus Prime X570-Pro, 64 GB DDR4 3200, GeForce GTX 1660 Super
openrc-0.17, ~vanilla-sources, ~nvidia-drivers, ~gcc

Tony0945
Watchman

Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA

Posted: Wed Jan 08, 2020 1:20 am Post subject:

acct-groups i.e. GLEP-81 I have them all masked. I unmask them as needed after checking what they are going to do. So far I haven't found one that changed an existing group assignment. When one does occur, I suppose I can add it to package provided or (more likely) add an ebuil;d to local overlay that assigns the group I already have. Unless I think the devs made the right choice and I made the wrong one. I would have just rather have had a wiki article listing the preffered group assignments.

Anon-E-moose
Watchman

Joined: 23 May 2008
Posts: 6185
Location: Dallas area

Posted: Wed Jan 08, 2020 10:10 am Post subject:

I have a few userid/groupids that would change if I activated the acct-* stuff. But the base system goes back to 2004 and things have changed since then. What those acct* numbers are, are meaningless, other than they are supposed to be unique for each device. I have a script that runs daily that puts all acct-* stuff in package.provided and then ,if needed, I just take care of it myself, kind of the way you used to years ago. _________________ UM780, 6.12 zen kernel, gcc 13, openrc, wayland

asturm
Developer

Joined: 05 Apr 2007
Posts: 9320

Posted: Wed Jan 08, 2020 2:29 pm Post subject:

saellaven wrote:

The policy around GID/UID assignment was debated and changed, yes, GLEP-81 itself no. No change for end users, no need to spread fud.

Morality124
Tux's lil' helper

Joined: 20 Feb 2018
Posts: 102

Posted: Thu Jan 09, 2020 12:55 am Post subject:

asturm wrote:

No change for end users, no need to spread fud.

...

proteusx wrote:

I delete them from the tree to prevent annoyances such as this one:

Code:

~ # equery u pdnsd
!!! Ambiguous package name. Choose from:
acct-user/pdnsd
net-dns/pdnsd
acct-group/pdnsd

Anon-E-moose wrote:

It's taking progressively longer to run a simple check "emerge -pvuDU @world" just to have it report that absolutely nothing needs to be emerged.
If it's not that more packages are added to make it longer to churn through everything, then do you think it's caused by sunspots. :roll:

Note: it's not that acct* stuff is that much but when you add all the useless virtual's along with it and a few other things, it all adds up.

pjp wrote:

That solves the x systems, y packages problem. I'm counting 46 ebuilds in acct-group,user, so that overhead would be eliminated.

Naib wrote:

I would want some architecting, some modeling, some system engineering so the end user doesn't get something like this

Code:

# emerge syncthing
Calculating dependencies /

[ Results for search key : syncthing ]
* acct-group/syncthing
Latest version available: 0
Latest version installed: [ Not Installed ]
Size of files: 0 KiB
Homepage:
Description: Group for the system-wide net-p2p/syncthing server
License:

* acct-user/syncthing
Latest version available: 0
Latest version installed: [ Not Installed ]
Size of files: 0 KiB
Homepage:
Description: User for the system-wide net-p2p/syncthing server
License:

* net-p2p/syncthing
Latest version available: 1.2.1
Latest version installed: [ Not Installed ]
Size of files: 21,993 KiB
Homepage: https://syncthing.net
Description: Open Source Continuous File Synchronization
License: MPL-2.0

[ Applications found : 6 ]

!!! The short ebuild name "syncthing" is ambiguous. Please specify
!!! one of the above fully-qualified ebuild names instead.

... done!

That is poor oversight... Signs of "I know what lets do foo" without actually modeling the problem and working through scenarios

tholin wrote:

The update didn't go that smoothly for me. Acct-user/qemu broke my setup.

I have a gaming VM using evdev device passthrough. The user running the VM needs direct access to the input devices for that to work. Typically you put the unprivileged user into a group with permission to access the needed hardware resource so I have the "qemu" user in the "input" group. Installing acct-user/qemu will change the group membership of "qemu" to a hardcoded list of only qemu,kvm and not input. The VM will then fail to start with Permission denied.

The GLEP says "The proposal also respects direct sysadmin modifications." but obviously not in this case.

Well that's a load of malarkey...

pjp
Administrator

Joined: 16 Apr 2002
Posts: 20545

Posted: Thu Jan 09, 2020 5:18 am Post subject:

Morality124 wrote:

pjp wrote:

That solves the x systems, y packages problem. I'm counting 46 ebuilds in acct-group,user, so that overhead would be eliminated.

Out of curiosity, a quick check appears to indicate 105 users, 119 groups. I only did a cursory check to see if there were any duplicates.
_________________
Quis separabit? Quo animo?

asturm
Developer

Joined: 05 Apr 2007
Posts: 9320

Posted: Thu Jan 09, 2020 3:13 pm Post subject:

Morality124 wrote:

...

All of which is completely irrelevant to my post.

saellaven
l33t
l33t

Joined: 23 Jul 2006
Posts: 655

Posted: Thu Jan 09, 2020 5:32 pm Post subject:

asturm wrote:

saellaven wrote:

The policy around GID/UID assignment was debated and changed, yes, GLEP-81 itself no. No change for end users, no need to spread fud.

Because GLEP-81 wasn't fully thought through... devs are still debating whether or not they can try to force GID/UIDs across all installs to try to simplify containers and such, and what that means for legacy installs where UIDs/GIDs are already in use. There's also the debate over what UIDs/GIDs should be restricted for system use.

This is something you (royal you) are supposed to think through before implementing things... or you're likely to screw things up and cause more headaches. I understand having to go back and tweak an implementation because some corner case arrived, but these are very fundamental questions to the very purpose of the need for GLEP-81 to begin with.
_________________
Ryzen 3700X, Asus Prime X570-Pro, 64 GB DDR4 3200, GeForce GTX 1660 Super
openrc-0.17, ~vanilla-sources, ~nvidia-drivers, ~gcc

Zucca
Moderator

Joined: 14 Jun 2007
Posts: 3807
Location: Rasi, Finland

Posted: Sat Jan 25, 2020 12:58 pm Post subject:

Naib wrote:

I would want some architecting, some modeling, some system engineering so the end user doesn't get something like this

Code:

That is poor oversight... Signs of "I know what lets do foo" without actually modeling the problem and working through scenarios

Indeed. There should be an option (in make.conf by default) to exclude such categories from any action unless specifically appended with the category name to the argument.
_________________
..: Zucca :..

My gentoo installs:

init=/sbin/openrc-init
-systemd -logind -elogind seatd

Quote:

I am NaN! I am a man!

tld
Veteran
Veteran

Joined: 09 Dec 2003
Posts: 1848

Posted: Sat Jan 25, 2020 3:24 pm Post subject:

Zucca wrote:

Indeed. There should be an option (in make.conf by default) to exclude such categories from any action unless specifically appended with the category name to the argument.

Wow...don't get me started as to how that one has been annoying the crap out of me:

Code:

equery list -p mythtv
* Searching for mythtv ...
!!! Ambiguous package name. Choose from:
media-tv/mythtv
acct-group/mythtv
acct-user/mythtv

Given how many applications create users with the same name as the app, this was all but designed to create this scenario.

Tom

duane
Apprentice

Joined: 03 Jun 2002
Posts: 193
Location: Oklahoma City

Posted: Wed Feb 05, 2020 10:07 am Post subject: "At last, the fix no one asked for"

https://www.theregister.co.uk/2020/02/03/linux_home_directories_merged_into_systemd/

Anon-E-moose
Watchman

Joined: 23 May 2008
Posts: 6185
Location: Dallas area

Posted: Wed Feb 05, 2020 11:36 am Post subject: Re: "At last, the fix no one asked for"

duane wrote:

https://www.theregister.co.uk/2020/02/03/linux_home_directories_merged_into_systemd/

From the link

Quote:

One use case is where a user has a PC running Linux in both their home and office, and is able to carry their home directory with them on a portable storage device. The advent of cloud storage has made this less of a problem than would have been the case a few years back, and a common reaction to the new systemd approach is that the problems it fixes are not pressing and may be outweighed by potential incompatibilities.

So I need homed because I can't figure out how to put a few files that I'm working on at the office on a usb stick and take it home to continue to work on it!!!
_________________
UM780, 6.12 zen kernel, gcc 13, openrc, wayland

Tony0945
Watchman

Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA

Posted: Wed Feb 05, 2020 3:32 pm Post subject: Re: "At last, the fix no one asked for"

Anon-E-moose wrote:

So I need homed because I can't figure out how to put a few files that I'm working on at the office on a usb stick and take it home to continue to work on it!!!

LOL!
RedHat, the distro for complete incompetents.

gtwrek
Tux's lil' helper

Joined: 10 Mar 2017
Posts: 110
Location: San Jose, CA

Posted: Wed Feb 05, 2020 4:10 pm Post subject: Put home under revision control

There's many (old!) articles out there regarding putting "home" directories under revision control. I always thought that might be an interesting experiment, but as yet haven't tried it. I have shared home directories over NFS among a few of my servers at home. I haven't decided if that's a good idea or not. Pretty sure I'd not trust systemd to something similar and do it well.

dmpogo
Advocate

Joined: 02 Sep 2004
Posts: 3461
Location: Canada

Posted: Wed Feb 05, 2020 4:44 pm Post subject: Re: Put home under revision control

gtwrek wrote:

There's many (old!) articles out there regarding putting "home" directories under revision control. I always thought that might be an interesting experiment, but as yet haven't tried it.
I have shared home directories over NFS among a few of my servers at home. I haven't decided if that's a good idea or not.

Pretty sure I'd not trust systemd to something similar and do it well.

Sharing home directories over NFS was a standard setup on university networks in the 90-ies. Remember during my first postdoc, all departamental computers were named after canadian wildlife, all disks - after canadian trees/plants, and nobody really paid attention to what computer what disk was physically attached.

gtwrek
Tux's lil' helper

Joined: 10 Mar 2017
Posts: 110
Location: San Jose, CA

Posted: Wed Feb 05, 2020 6:04 pm Post subject: Shared home

Quote:

Sharing home directories over NFS was a standard setup on university networks in the 90-ies.

How is this not standard over any organization? It's situation normal at any largish organization I've worked for during the last 30 years or so.
How could one manage things any differently across any decent sized org, where there's more than just a few servers?

The only drawback for me doing this at home, is making sure the one NFS server (hosting "home" directories) is always up before attempting to do anything useful on any other server. I guess I'm not yet comfortable with me keeping this NFS server up 24/7, as a prerequisite for any other server being useful. The latter servers would boot of course, but any user logging in (with /home unavailable) doesn't pass the Spousal Acceptance Factor.

dmpogo
Advocate

Joined: 02 Sep 2004
Posts: 3461
Location: Canada

Posted: Wed Feb 05, 2020 6:36 pm Post subject: Re: Shared home

gtwrek wrote:

Quote:

Sharing home directories over NFS was a standard setup on university networks in the 90-ies.

I have my home desktop up for almost two years right now (well, just rebooted, since I finally upgraded the kernel), previous home record of uptime was something like 4 years, from 2004 - Gentoo install, to 2007, hard drive failed.

Anyway, my current university physics department (50 profs) has no research network management at all. Funding for system admins was cut years ago (when I came in 2001, there were two, for UNIX and Windows machines), everybody have their own desktops/server/laptop, bought with personal grants and shared only with there own group/students if any. Faculty level has IT people which manage network access to backbone and has somebody looking for secretarial windows machines. But for research we are on our own - just have to go to faculty to negotiate when we need static IP or ports open, or space for server housing.

dmpogo
Advocate

Joined: 02 Sep 2004
Posts: 3461
Location: Canada

Posted: Wed Feb 05, 2020 6:46 pm Post subject: Re: Shared home

[quote="gtwrek"]

Quote:

doesn't pass the Spousal Acceptance Factor.

In my case my spouse refuse "to have anything to do with your network". To the extend that she bought a printer for herself and only connects it to her mac by USB cable,
refusing to make printer visible on wifi

Well, she has her own hammer, screwdriver and a set of nails as well

The only thing I do, is silently backing up her mac laptop (initiating backups from the server)

Display posts from previous:

	Gentoo Forums Forum Index Gentoo Chat	All times are GMT Goto page Previous 1, 2, 3, 4, 5, 6, 7 Next
Page 4 of 7

Jump to:

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

Copyright 2001-2024 Gentoo Foundation, Inc. Designed by Kyle Manna © 2003; Style derived from original subSilver theme. | Hosting by Gossamer Threads Inc. © | Powered by phpBB 2.0.23-gentoo-p11 © 2001, 2002 phpBB Group
Privacy Policy