Canonical quality control

[gm10]

What's going on over at Canonical? Just recently the 4.15.0-24 kernel debacle releasing a patch that was already known to be bugged upstream, a proposed fix that just rolls the problematic change back rather than actually fixing it - with a known fix, grub2-signed install issues, and today they actually managed another broken kernel release in bionic-proposed, for 4.15.0-28 they literally forgot to add the actual amd64 kernel image to the repo:

https://launchpad.net/ubuntu/bionic/amd ... 28-generic

The Canonical Kernel Team's PPA has the file, it clearly just got lost copying it from one repo to the other.

No checks in place. None.

Say it with the Don: Sad!

[catweazel]

https://launchpad.net/ubuntu/bionic/amd ... 28-generic

Published 12 seconds ago

I've seen this before. I presume they set up the pages before they push updates.

[mwbworld]

Between the mesa thing recently, the kernel and the grub install issue - that is a lot in a short time.

I'm chalking it up to the pressure of the new 18.04 release and trying to prep the .1. Plus the whole "we're just trying to get bought out" by the company can't be helping morale.

I've noticed a certain amount of crankiness over on the Ubuntu forums dealing with these things. Of course, one of the reasons I prefer mint is even when these things happen - the forums here are still a way more pleasant a place to seek answers than the Ubuntu ones. (Of course, they are way better than the general Arch ones - admittedly not that high a standard!)

[gm10]

Published 12 seconds ago

I've seen this before. I presume they set up the pages before they push updates.

Look at them, stealing my thunder. They had pushed that release hours ago actually, minus that one file, I only noticed because it somewhat broke the Update Manager. At least they caught it faster this time. Still shouldn't happen.

Between the mesa thing recently, the kernel and the grub install issue - that is a lot in s short time.

The Mesa issue upon upgrading? That was ultimately purely a Mint issue, no wrong-doing on the Ubuntu side.

[catweazel]

Published 12 seconds ago

I've seen this before. I presume they set up the pages before they push updates.

Look at them, stealing my thunder.

Still shouldn't happen.

You're quite right. It shouldn't.

[catweazel]

Plus the whole "we're just trying to get bought out" by the company can't be helping morale.

Actually, I dare say that's the primary reason. People make mistakes when they're worried about their jobs.

the forums here are still a way more pleasant a place to seek answers than the Ubuntu ones.

We get booted if we attack people. The mods do a good job.

[mwbworld]

Between the mesa thing recently, the kernel and the grub install issue - that is a lot in s short time.

The Mesa issue upon upgrading? That was ultimately purely a Mint issue, no wrong-doing on the Ubuntu side.

Yeah, I see your point. It didn't break their stuff but was a backporting that caused problems with mint that was ultimately changed.

[KBD47]

I bet I'm not the only one looking forward to LMDE release. Debian gets complained about for being slow and outdated, but I don't think I've ever seen so many errors in such a short amount of time coming from Ubuntu, and are not tiny errors either.

[gm10]

I bet I'm not the only one looking forward to LMDE release. Debian gets complained about for being slow and outdated, but I don't think I've ever seen so many errors in such a short amount of time coming from Ubuntu, and are not tiny errors either.

You are not. Starting to feel like Debian's sid over here sometimes. All the untested content you could wish for right in your Update Manager.

[KBD47]

You are not. Starting to feel like Debian's sid over here sometimes. All the untested content you could wish for right in your Update Manager.

I've never seen so many problems with a new version of Mint. 99% is directly from Ubuntu, especially that .24 kernel. Yet I understand Debian Sid dealt with the .24 kernel problem months ago, so Sid might actually be more stable at this point

[gm10]

You are not. Starting to feel like Debian's sid over here sometimes. All the untested content you could wish for right in your Update Manager.

I've never seen so many problems with a new version of Mint. 99% is directly from Ubuntu, especially that .24 kernel. Yet I understand Debian Sid dealt with the .24 kernel problem months ago, so Sid might actually be more stable at this point

Debian has already fixed it, yes. And properly, unlike Ubuntu so far.

[michael louwe]

What's going on over at Canonical? Just recently the 4.15.0-24 kernel debacle releasing a patch that was already known to be bugged upstream, a proposed fix that just rolls the problematic change back rather than actually fixing it - with a known fix, grub2-signed install issues, and today they actually managed another broken kernel release in bionic-proposed, for 4.15.0-28 they literally forgot to add the actual amd64 kernel image to the repo: ...

.
Likely, because of the recent Meltdown & Spectre patching fiasco and the need to maintain all the various Ubuntu release cycles as per schedule, the Ubuntu team have been over-worked = suffering fatigue. Maybe Canonical Inc/Mark Shuttleworth should temporarily postpone or cancel the next releases, eg Ubuntu 18.10 and 18.04.1.

[gm10]

eg postpone Ubuntu 18.10 and 18.04.1.

Ubuntu 18.11? Heresy.

[Pjotr]

This hasn't got anything to do with quality control, but everything with the goal of "proposed". This is what I've written in another thread:

Things like this happen regularly with "proposed".... They just dump packages in it when they're ready for testing, sometimes even irrespective of dependencies.

The "proposed" repo is not for the faint of heart.

Strictly speaking it's no blunder of Canonical.... After all, "proposed" is only meant for testing packages, not for maintaining a stable system. Perhaps it's simply efficient: the sooner a package (irrespective of its dependencies) becomes available for testing and analysis, the better.

The Canonical Kernel Team PPA on the other hand, does aim at being useful for (more or less) stable systems. So if you want a newer kernel for a (somewhat) stable system, try that PPA and forget "proposed".

[gm10]

Strictly speaking it's no blunder of Canonical.... After all, "proposed" is only meant for testing packages, not for maintaining a stable system. Perhaps it's simply efficient: the sooner a package (irrespective of its dependencies) becomes available for testing and analysis, the better.

The Canonical Kernel Team PPA on the other hand, does aim at being useful for (more or less) stable systems. So if you want a newer kernel for a (somewhat) stable system, try that PPA and forget "proposed".

* had first posted this in the other thread and suggested to move here, but you have already found it so deleted it there *

Since we're strictly speaking, that's not actually how it works. The PPA comes first, if it builds correctly it can then get moved to -proposed. The workflow is described here: https://wiki.ubuntu.com/Kernel/KernelPPA

In this case they simply blundered by forgetting to copy one of the files into -proposed.

[Pjotr]

Strictly speaking it's no blunder of Canonical.... After all, "proposed" is only meant for testing packages, not for maintaining a stable system. Perhaps it's simply efficient: the sooner a package (irrespective of its dependencies) becomes available for testing and analysis, the better.

The Canonical Kernel Team PPA on the other hand, does aim at being useful for (more or less) stable systems. So if you want a newer kernel for a (somewhat) stable system, try that PPA and forget "proposed".

* had first posted this in the other thread and suggested to move here, but you have already found it so deleted it there *

Since we're strictly speaking, that's not actually how it works. The PPA comes first, if it builds correctly it can then get moved to -proposed. The workflow is described here: https://wiki.ubuntu.com/Kernel/KernelPPA

In this case they simply blundered by forgetting to copy one of the files into -proposed.

I stand corrected: that's indeed the order of things.

Nevertheless, the point I'm trying to make is this: "proposed" is just a package dump, intended to be enabled exclusively in dedicated test rigs and VM's. Its goal is only to enable you to help with testing all sorts of packages, for every part of your system; breakage is frequent, expected and normal. Sometimes its breakage can only be repaired by restoring a snapshot (Timeshift, VirtualBox).

No previous quality control worth that name: the testers *are* the real quality control.

The Canonical Kernel Team PPA on the other hand, is only for testing kernels and nothing more. As such, the damage it can do is much more limited and easily reversible. Which makes it usable for (more or less) stable systems, as emergency measure applied by advanced users.

[gm10]

Nevertheless, the point I'm trying to make is this: "proposed" is just a package dump, intended to be enabled exclusively in dedicated test rigs and VM's. Its goal is only to enable you to help with testing all sorts of packages, for every part of your system; breakage is frequent, expected and normal. Sometimes its breakage can only be repaired by restoring a snapshot (Timeshift, VirtualBox).

Well, it's not quite that random. Proposed is the verification repo for bug fixes before they can be moved into the main repos. That's all that goes into it. If you report an Ubuntu bug on launchpad you'll be asked to test the fix in -proposed as part of the verification process. This is for normal Ubuntu users, not just dedicated testing rigs, but you should for sure not just blindly install packages from there.

This overview page allows you to correlate the packages in -proposed to the respective bug reports so you can follow what's going on:
http://people.canonical.com/~ubuntu-arc ... g-sru.html

Also interesting, progress of special verification process for proposed kernels:
http://kernel.ubuntu.com/sru/kernel-sru-workflow.html

[Pjotr]

Well, I think we mostly agree, and the differences between our opinions are more differences in nuances.

In my glorious n00b days, in the first year of my switch to desktop Linux in general and Ubuntu in particular, after a while I used to enable "proposed" as a matter of course. Why? Because it seemed like a good idea at the time....

Unsurprisingly, my Ubuntu often became unstable and broken. And some updates were strangely "held back" for a long time, due to missing dependencies. All in all, I had to re-install Ubuntu quite a lot.

Only after some months did I realize that there might be a connection with "proposed".

Although it's over a decade ago, I've never forgotten this valuable lesson about "proposed"....

[gm10]

No, I completely hear you on that. I'm even careful with what I update from the main repos on a production system. Unless there's a compelling reason, "never change a running system" is still the best advice IMHO.

I often find compelling reasons in kernels though.

[JeremyB]

So gm10 was the 4.15.0-28 kernel a dependency of package linux-generic or linux-image-generic at the time with bionic-proposed. Or was it in linux-image-generic-hwe-16.04 or linux-image-generic-hwe-16.04-edge in xenial-proposed? Ubuntu does kernel updates differently than Linux Mint, most Mint users may just have linux-kernel-generic installed for a kernel meta package.