I’m a solid Manjaro fan that recommends the system to family and friends and anyone who will listen, but I’ve just had my first serious fight with it when a routine upgrade suddenly rendered my system unable to boot. I had to spend many unhappy hours figuring out that it’s just a matter of course with the rolling release. For more context, see Lightdm fails to start after update
I marked the solution there because I discovered the cause, but there is currently no real solution to the rolling release with dropped hardware support problem. I see this as a serious liability for the long term health of Manjaro and Arch-based distros.
Many Linux users want to keep using old reliable hardware until it fails. For various reasons (ethics, resource conservation, efficiency, stability, …), they don’t want to be forced to buy a newer GPU just to install or upgrade a basic calculator app. They should be able to do so without having to fight unnecessary changes that suddenly break the system.
What makes Manjaro great is that it puts simple control in the hands of the user, and this should extend to critical drivers as well as the kernel. For application logic dependencies, rolling seems fine, but hardware should be more stable. We need a simple hold back mechanism to help users keep their systems in a working state because this came as a surprise and took way too long to figure out.
“Enjoy the simplicity.” Yes, let’s keep the system working first, and then maybe try an upgrade for all the fancy new GPU features. Let us go back and stay there if something really breaks, just as we can do with the kernel.
There are thousands of different variations of hardware, its impossible for the manjaro team to test for them all, rolling release is what arch is and is the distro manjaro derived from, steps have been taken to thoroughly test for system breakage by having an extra testing stage compared to upstream but things can get overlooked, what worked for 99% of people, there is bound to be that 1% where the upgrade breaks for them, its not fair to reverse an upgrade cause of that 1%. Solutions have been provided, its what this forum is for, if super stability is what you are looking for, ubuntu based distros would be your best bet.
There is a simple hold back mechanism, but it is control of the manjaro team. To mitigate the effect of broken software, you should use backups. Example: Timeshift.
Sadly saying: At some point that is not possible on a rolling release. A point release would be a better option.
Yeah, that’s exactly what point releases do. They don’t add new features, but add bug fixes and security fixes.
So at the end, you want Manjaro to be a point release?
I’m running Manjaro on a PC I built >8 years ago and have had no issues to date. I think a big part of that is I have an AMD GPU (actually APU). All these issues relating to NVIDIA drivers are not the fault of the Manjaro developers. Planned obsolescence by NVIDIA is to blame.
And to be fair there is no dropped support, as Manjaro kept the driver supporting this so called “dropped support hardware”, and the information for the people affected by latest release of Nvidia driver is in the update announcement post, just have two commands to type to have no issue with Nvidia driver so… what can Manjaro do more?
If you have to ask, you weren’t on the forum the last…err…about…well I can’t say more than a year, because that’s how long I’ve been here, so at least a year.
In that time it has become glaringly obvious to me that the least Manjaro could do is magic. With that all problems would be a thing, or things, of the past.
Just adding my thoughts on this:
Whilst reading this, I remember that there was also a problem with nvidia GPU’s I was affected.
And if I understood it correctly: Nvidia drops support for gen X and some users are unhappy.
Since I do not believe that every user can read the updateannouncements (I’ve a long todolist which screams to me, that I shall finish these tasks). I believe, that there could be an easy solution like the following: Check during update if the hardware will not be compatible, if that is the case, alert the user and make a “one click suggestion”. (Alertbox: “Hey your hardware maybe incompatible, the maintainer suggests the following Solution: Switch to open source driver? Yes/(No, I risk breaking my system)”)
I, on the other hand, say it is your responsibility to read before you do the update. I know I’ll get a lot of hate for this, I always do for saying what I believe is the right thing.
Just like what’s happening now, there are those that’ll select Yes to continue and the forum will be flooded when stuff breaks. Not if, but when.
They cant do an other way. Most of new Linux users are trained by Microsoft for years clicking “Yes” and nothing critical is happens.
I understand the wish just if the user is only a enduser and not interested in IT at all. And I think there is a need for the dev-team which enduser and usability they aim with manjaro.
For me it is ok like it is at now.
I only casual read the update announcements … if something breaks while update I see there a chance to learn something new.
Thanks everyone! Manjaro rocks, in no small part thanks to you the community! I should have watched the wiki more carefully and especially appreciate those who pointed things out with patience.
I do want to keep recommending Manjaro as a primary system for family, friends, even kids who are just starting to learn Linux. If their boxes fall over due to proprietary planned obsolescence, well, it’s not entirely their fault. I don’t expect magic from Manjaro devs, I’m a dev myself and fully appreciate what goes into the project. Just trying to help Manjaro succeed by putting fences around the pitfalls of user experience.
The hard work of compatibility consideration was already done; the warnings were written in release notes1. But plenty of users were predictably broken by the upgrade, and it could have been avoided simply by encoding the compatibility knowledge into the upgrade process. Something like Melc’s suggestion: if (gpu.is_kepler() || ...) { warn_and_explain_with_link_to_wiki(); offer_driver_selection_change(); }.
Many tenacious types (self included) will figure things out and fix their systems. But some users will simply leave and never come back. The troubleshoot post offering clear descriptions of issues & fixes is great, but most users aren’t going to read everything, so coding a few safeguard lines would be tremendously helpful for them. Even a hard stop in the user-friendly interface would be better than steam rolling boot capability: if (gpu_driver_is_mainline_proprietary() && gpu.is_known_unsupported()) { refuse_to_upgrade_with_link_to_explanation(); }
Guard rails are better than signs saying, “One more step and you’ll fall to your doom!”
Sure, some of us will drop into a terminal to run commands with forcing options, but then we’re the type that expects breakage. Not so for the kids; they just need the machine to come up so they can learn how to rice their desktops with Linux. Gotta start somewhere and I’d like it to be Manjaro!
Thanks again for everyone who took the time to consider and read all this. Cheers!
now.
