Scheduled Maintenance: We are aware of an issue with Google, AOL, and Yahoo services as email providers which are blocking new registrations. We are trying to fix the issue and we have several internal and external support tickets in process to resolve the issue. Please see: viewtopic.php?t=158230

 

 

 

[Discussion] Why are severely broken point releases being released for Bookworm

Here you can discuss every aspect of Debian. Note: not for support requests!
Message
Author
CwF
Global Moderator
Global Moderator
Posts: 2719
Joined: 2018-06-20 15:16
Location: Colorado
Has thanked: 41 times
Been thanked: 201 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#21 Post by CwF »

LeeE wrote: 2024-02-15 17:35 CwF wrote: ↑2024-02-14 17:10
To delay the point release because of Nvidia should not be considered an option.

I think this statement needs justification with regard to why the timing of point releases should be considered critical and thus cannot be delayed.
This is a corollary to the above comments on having some mechanism for Nvidia users to have an temporarily alternative path - whatever that is. Nvidia is a second class citizen of Linux.
Nothing is critical, especially Nvidia - and I type this on a Nvidia.

pizza-rat
Posts: 91
Joined: 2023-05-16 21:38
Has thanked: 39 times
Been thanked: 21 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#22 Post by pizza-rat »

I still haven't seen anybody actually elaborate on why exactly point releases have to be rushed out with known bugs that will affect a sizeable amount of users. The "responses" I've read so far have given me the impression that any of the following may by why:
  • Nvidia users are considered second class
  • Wifi users are considered second class
  • Desktop users are considered second class
Who exactly, within the userbase - corporate or desktop - is so desperately wanting these point releases ASAP that they must be pushed even when it's known they will cause breakage for a noteworthy amount of users? To assume the worst case here: is it purely to please investors? Is Debian going the route of Ubuntu? Please correct me if I'm wrong here, as I would much rather assume otherwise.

lindi
Debian Developer
Debian Developer
Posts: 452
Joined: 2022-07-12 14:10
Has thanked: 1 time
Been thanked: 88 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#23 Post by lindi »

pizza-rat wrote: 2024-02-15 19:33 I still haven't seen anybody actually elaborate on why exactly point releases have to be rushed out with known bugs that will affect a sizeable amount of users. The "responses" I've read so far have given me the impression that any of the following may by why:
  • Nvidia users are considered second class
  • Wifi users are considered second class
  • Desktop users are considered second class
Who exactly, within the userbase - corporate or desktop - is so desperately wanting these point releases ASAP that they must be pushed even when it's known they will cause breakage for a noteworthy amount of users? To assume the worst case here: is it purely to please investors? Is Debian going the route of Ubuntu? Please correct me if I'm wrong here, as I would much rather assume otherwise.
Would delaying point releases actually help much here as you would still get the buggy packages from security updates in many cases anyway? (At least the ext4 issue affected people even before the point release was made).

LeeE
Posts: 52
Joined: 2006-04-23 18:31
Location: Harlow, Essex, UK
Has thanked: 9 times
Been thanked: 7 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#24 Post by LeeE »

lindi wrote: 2024-02-15 20:00 Would delaying point releases actually help much[...]
I think it would have resulted in many fewer views and a lot less anxiety
...or something

CwF
Global Moderator
Global Moderator
Posts: 2719
Joined: 2018-06-20 15:16
Location: Colorado
Has thanked: 41 times
Been thanked: 201 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#25 Post by CwF »


pizza-rat
Posts: 91
Joined: 2023-05-16 21:38
Has thanked: 39 times
Been thanked: 21 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#26 Post by pizza-rat »

lindi wrote: 2024-02-15 20:00 Would delaying point releases actually help much here as you would still get the buggy packages from security updates in many cases anyway? (At least the ext4 issue affected people even before the point release was made).
All I can say is I've never been affected by a buggy package via just security updates, but I have been affected by this point release. Are security updates that are known to cause breakage (to the extent that these recent point releases have) also pushed out regardless? It sounds like a more nuanced situation, given that security is involved, than the topic at hand.

pizza-rat
Posts: 91
Joined: 2023-05-16 21:38
Has thanked: 39 times
Been thanked: 21 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#27 Post by pizza-rat »

I don't think this is the excellent riposte you seem to think it is, and the question being posed is still unanswered.

CwF
Global Moderator
Global Moderator
Posts: 2719
Joined: 2018-06-20 15:16
Location: Colorado
Has thanked: 41 times
Been thanked: 201 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#28 Post by CwF »

pizza-rat wrote: 2024-02-15 20:27 I don't think this is the excellent riposte you seem to think it is,
I think it is.
pizza-rat wrote: 2024-02-15 20:27 , and the question being posed is still unanswered.
It has. Over sights happen. Your weighting, is yours. There are similar oversights every release.
...moving on

steve_v
df -h | grep > 20TiB
df -h | grep > 20TiB
Posts: 1418
Joined: 2012-10-06 05:31
Location: /dev/chair
Has thanked: 80 times
Been thanked: 191 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#29 Post by steve_v »

pizza-rat wrote: 2024-02-15 19:33 The "responses" I've read so far have given me the impression that any of the following may by why:
  • Nvidia users are considered second class
  • Wifi users are considered second class
  • Desktop users are considered second class
Nobody said anything about class. It's simply that issues with optional proprietary drivers are not the pants-on-fire "critical components" "completely broken" hold-everything situations the OP claims they are, and delaying a release so close to rollout for the sake of a single out-of-tree driver would arguably be just as problematic.

If a particular kernel version causes problems for your hardware configuration, the obvious answer is as obvious as ever: Keep calm and run another version until it is fixed.
Once is happenstance. Twice is coincidence. Three times is enemy action. Four times is Official GNOME Policy.

pizza-rat
Posts: 91
Joined: 2023-05-16 21:38
Has thanked: 39 times
Been thanked: 21 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#30 Post by pizza-rat »

steve_v wrote: 2024-02-15 23:35 and delaying a release so close to rollout for the sake of a single out-of-tree driver would arguably be just as problematic.
Why? Who is rushing to update to the next point release? What's an extra week or two?

appye
Posts: 33
Joined: 2007-11-21 16:26
Has thanked: 1 time
Been thanked: 6 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#31 Post by appye »

Only thing I can weigh in on here is that anyone who is new to Debian, hits the "download" button on the main site and has an nvidia card will be disappointed with their experience, greeted with kernel panics or a system that is stuck at CLI, and the need to peruse forums to try and figure out why their system is broken.

My experience with this issue as a Debian regular since Stretch, as I remember it at least:

I set up a system for my daughter last Saturday using 12.4. I chose to enable automatic updates during the installation. I downloaded the image from the site, dd'd it to a usb stick, got her up and running, installed the nvidia driver, configured a few things and her system was happily humming along for a day or two. She rebooted her computer to play a game in Windows, and was greeted with a kernel panic when trying to get back into the system. At least this is how I remember it. Perhaps I ran the apt full-upgrade and didn't pay attention to the dkms error, or the automatic updates updated things to this state. I don't remember for sure. This was certainly unexpected, mainly because I have always trusted Debian Stable not to have issues that would cause kernel panics ...

At least it was not hardware failure.

pwzhangzz
Posts: 441
Joined: 2020-11-11 17:42
Has thanked: 10 times
Been thanked: 27 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#32 Post by pwzhangzz »

? ? ?

What is your point exactly? ? ?

appye
Posts: 33
Joined: 2007-11-21 16:26
Has thanked: 1 time
Been thanked: 6 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#33 Post by appye »

My point was in the first paragraph, that many new debian users might be put off by the experience.

User avatar
Chaussettes
Posts: 50
Joined: 2022-09-29 03:45
Has thanked: 7 times
Been thanked: 20 times
Contact:

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#34 Post by Chaussettes »

I think a lot of what the NVIDIA side is trying to get across here is that it's not exactly clear why a point release update had to be pushed if this problem was known about beforehand. Unless you have a lot of play-money to throw around into the new Intel GPU market, AMD and NVIDIA are your only choices. While it would be perfect if everyone used AMD and didnt have to worry about these issues, ignoring the NVIDIA userbase just seems irresponsible.

A lot of us have a hard time understanding why a point release could not have been delayed until the problem was fixed. Is that not a valid thing to ask or expect from a distro that wants to be usable on both server AND desktop? Debian got a big desktop userbase boost in the bookworm release from media coverage like youtube and there is not an insignificant number of people running an NVIDIA card. I caught the bug before updating by looking at the forum and I'm sure most people here on the forum do things like read the release posts and check the forum/subreddit beforehand, but if you want the honest truth, most desktop users don't. While the user is ultimately in charge of what happens on their computer, the developer should have some measure of responsibility if they are trying to cater a usable desktop system.

Finally, I'd like to remind people that the default Debian desktop is GNOME and GNOME Software has automatic system updating turned on by default. This could quite literally be an unavoidable issue for a large number of people.
Image
Creating art on deviantArt - https://www.deviantart.com/chaussettes99

User avatar
alienspy
Posts: 158
Joined: 2023-02-12 15:37
Has thanked: 97 times
Been thanked: 5 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#35 Post by alienspy »

I can't find detailed release notes for 12.5 on the web. All i found is this - https://www.debian.org/News/2024/20240210. And there is nothing about possible nvidia driver problem. I got no mail in debian-announce mailing list about 12.5 also. Yesterday and today i tried to subscribe to other mailing lists... but... i can't. Tried with Firefox and Chrome, https://lists.debian.org just keeps reloading without actually doing anything. Probably information about possible errors should be easy to find.

Borg
Posts: 33
Joined: 2024-02-16 22:49
Has thanked: 3 times
Been thanked: 3 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#36 Post by Borg »

steve_v wrote: 2024-02-13 04:22
Chaussettes wrote: 2024-02-12 22:33a large portion of the desktop/laptop user base
FTFY.
The majority of Debian (and GNU/Linux in general) installations are on servers, which don't care one iota about wireless (especially with closed-source blobs) or nvidia being nvidia and not keeping up with kernel changes.
That is a null argument. It is well known that Linux is used on servers much more often than Windows.
But for the 2-3% of desktop users who use Linux, Debian does have a large install base and that counts, because we're talking about these 2-3%.

Borg
Posts: 33
Joined: 2024-02-16 22:49
Has thanked: 3 times
Been thanked: 3 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#37 Post by Borg »

None1975 wrote: 2024-02-13 16:55 The interesting thing is that Debian version 11 didn't have these problems, which makes me think that Debian's quality is going down...
Debian 11 had the Firefox-ESR security problem in 4. quarter of 2022. It took a while, until this was fixed.

Borg
Posts: 33
Joined: 2024-02-16 22:49
Has thanked: 3 times
Been thanked: 3 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#38 Post by Borg »

My conclusion is:
This survey requires other data collection tools. Tools such as inxi.

Reason:
If you use Linux on servers or office PCs, then you obviously don't install a dedicated NVidia card, but rather use the iGPU of the Intel and AMD CPU.
And on multi-user systems, programs like adduser or locales lead to misleading data because these are usually office computers and when every locale of multiple users counts, then this means that the same computer is counted multiply times. Also logically, the number of iGPU users is very large when the same users use the same system.

If you want to know what percentage of desktop users use Nvidia video cards or especially dedicated video cards at home, then the Stream hardware surveys are far more informative:
https://store.steampowered.com/hwsurvey ... form=linux

There select Linux only, then click on Video Card Description (Linux)

Here it is 17.76 % for NVidia, 10.17 % for Intel and 63.01 % for AMD and 9,06 % unknown.
For dedicated video cards only, you have to filter out the iGPUs.

EDIT:
With iGPUs separated, it is only 18,3 % for dedicated AMD videocards.
Thus only 0,54 % more than dedicated NVidia GPUs. Both are therefore equally represented.

EDIT 2:
BTW from these 63.01 % AMD videocards 42.18 % belong to Valve's Steam Deck Device. Thus from all these cards (100 %) it is only 20.83 % for AMD iGPUs and dedicated GPUs combined on regular PCs.
https://en.wikipedia.org/wiki/Steam_Deck

EDIT 3:
BTW, when these 42.19 % are Steam Deck only, then we can conclude, that these other 57.81 % are desktop PCs. And now take a look at that Wikipedia article above and read the sales section:
There it is stated:
Valve stated in November 2023 that they had sold "multiple millions" of the Steam Deck.[86]
https://en.wikipedia.org/wiki/Steam_Deck#Sales
This means, there are multiple millions of Linux Desktop machines, that use Linux for gaming.
Last edited by Borg on 2024-02-17 03:00, edited 2 times in total.

Borg
Posts: 33
Joined: 2024-02-16 22:49
Has thanked: 3 times
Been thanked: 3 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#39 Post by Borg »

appye wrote: 2024-02-16 05:53 Only thing I can weigh in on here is that anyone who is new to Debian, hits the "download" button on the main site and has an nvidia card will be disappointed with their experience, greeted with kernel panics or a system that is stuck at CLI, and the need to peruse forums to try and figure out why their system is broken.
There should be an optionally disableable package that just contains a script that runs when the package is installed and resets the default kernel version to the previous version based on certain criteria in the grub boot menu.
This package could be called fast_reaction_debian_script.

This would allow the Debian maintainers to intervene quickly and provide a customized script after an update.
The script then checks the computer according to various criteria, in this case whether an NVidia card is present and the nvidia drivers are used. And if that is the case, then it reconfigures grub and makes the old kernel the default kernel.
Then when the problem is fixed in the repository, the Debian maintainers send out a new package of that package with a new script that undoes this change.

This would give Debian maintainers a quick way to respond in a variety of ways and to all types of issues that might arise when updating Debian.
And for those who do not want such interference in the system, it could be made so, that the download of the Fast reaction script can be easily disabled.

CwF
Global Moderator
Global Moderator
Posts: 2719
Joined: 2018-06-20 15:16
Location: Colorado
Has thanked: 41 times
Been thanked: 201 times

Re: [Discussion] Why are severely broken point releases being released for Bookworm

#40 Post by CwF »

Borg wrote: 2024-02-17 02:14 If you use Linux on servers or office PCs, then you obviously don't install a dedicated NVidia card, but rather use the iGPU of the Intel and AMD CPU
We are talking Debian here, not Linux et al.

I would guess most Debian machines have no igpu, not the ones your thinking! Matrox, aspeed, qxl, compound many gpus and of course, headless likely all outnumber single gpu dedicated Nvidia installations. How would you like to count a quad socket server with 8x16 GPU's on a mezzanine bus - and headless!

The Debian market is a little more than internet surfing and games. Steam stats are of little relevance. That really is
"a fraction of a minority within a select group of a few"

Locked