Internet unstable after recent update

[vazicebon]

What if you disable the IPv4 connection method in the settings (Method–>Disabled) ?

In case it prioritize IPv6.

edit: confirmation IPv6 is the default for me

[Archie1]

Link to the kernel creation guide, please.

@albersc2

Personally, I’ve only ever used GRUB, as that’s what I’m familiar with. However, make sure that the following files are in the correct location: (The first two are essential)

vmlinuz-linux-6.16.4-arch1-1-fix
initramfs-linux-6.16.4-arch1-1-fix.img
System.map-linux-6.16.4-arch1-1-fix

The linux kernel (i.e. the file vmlinuz-linux), initramfs (i.e. the file initramfs-linux.img) and Systemap (i.e. the file System.map-linux) all of them have the linux kernel version appended. And you use GRUB.
How did you achieve that with GRUB?

[Archie1]

@UncleSpellbinder

An awful lot of this today…

You use Waterfox? Is it some aspects better than LibreWolf and Firefox+ublockOrigin ? Can you please share your views on the browser?

[Archie1]

I am unsing Linux Kernel 6.12.42-1-lts with IPV4 for the past few weeks or so and have not had this issue. Will hold off on the upgrade for few weeks.

But my MTU value is the standard/default one, dont know if that is relevant.

@Moonbase59 and @albersc2, this forum is pretty addictive. I have learnt a lot by just lurking and occassionally . And yes EOS is the only thing which keeps me on Arch. This forum is one of the best for linux apart from MX Linux Forum, Debain Forum and Mint Forums.
Hats off to all those who take part and spend their time over here. You guys are the best.

[albersc2]

@Archie1

The linux kernel (i.e. the file vmlinuz-linux), initramfs (i.e. the file initramfs-linux.img) and Systemap (i.e. the file System.map-linux) all of them have the linux kernel version appended. And you use GRUB.
How did you achieve that with GRUB?

Hey Archie1, nice to meet you!

The step-by-step process I use to build a custom, patched kernel is explained here:

Internet unstable after recent update

Objective of This Guide Step-by-step creation of the 6.16.4 kernel patched with Oscar Maes’ fix, ensuring fully functional internet connectivity. This is by no means a “guide” in the strict sense, as there are many ways to prepare, compile, and install a kernel. Rather, it’s a step-by-step walkthrough of the method that works for me—one I’ve learned and refined over time. It has been tested and works well on my system, which is a 10th-generation i5 with a dedicated NVIDIA 4060 GPU. On hybrid …

[UncleSpellbinder]

@Archie1, I’d rather not hijack this thread with my views on Waterfox. Having said that, I just like using it. I have several browsers installed (Vivaldi, Librewolf, Waterfox, Firefox). Waterfox has been my primary lately.

[UncleSpellbinder]

Just updated this morning. Among the packages to be updated was core/linux-api-headers 6.16-1 -> 6.16-2. I’m wondering, is this anything that would affect internet connectivity?

[firefly]

@Archie1

I am unsing Linux Kernel 6.12.42-1-lts with IPV4 for the past few weeks or so and have not had this issue. Will hold off on the upgrade for few weeks.

From what I noticed (my machine is running on the LTS kernel by default), the bug was introduced in version 6.12.43 and carried over into the current 6.12.44.

My personal fix was to start using a VPN with full IPv6 support, because otherwise the system becomes literally unusable. Honestly, it somewhat boggles my mind why this issue (especially given that a fix already exists) wasn’t prioritized by the Arch package maintainers..

[thefrog]

@firefly

Honestly, it somewhat boggles my mind why this issue (especially given that a fix already exists) wasn’t prioritized by the Arch package maintainers..

Its probably not that its not Prioritized they are dealing with DDOS attacks and I’m sure that is taking up a lot of the resources that would be fixing issues with the Kernel and not having to waste time stopping these attacks.

[albersc2]

@thefrog

Its probably not that its not Prioritized they are dealing with DDOS attacks and I’m sure that is taking up a lot of the resources that would be fixing issues with the Kernel and not having to waste time stopping these attacks.

Agree. Imo, the kernel developers bear the main responsibility for this bug ending up in all branches without being noticed, and for it lingering for so many days even after it was discovered. This isn’t Arch’s fault—they already have plenty to deal with.

[albersc2]

@UncleSpellbinder

Just updated this morning. Among the packages to be updated was core/linux-api-headers 6.16-1 -> 6.16-2. I’m wondering, is this anything that would affect internet connectivity?

Don’t worry, you can safely update these linux-api-headers. They don’t affect the running kernel at all.

What does concern me is this other thing:

@firefly

From what I noticed (my machine is running on the LTS kernel by default), the bug was introduced in version 6.12.43 and carried over into the current 6.12.44.

That’s exactly what I thought, because afaik the regression hasn’t been officially fixed in any branch yet, although your tests the other day made me doubt it. If I didn’t have so much work these days, I would install 6.12.44 to test it on my system.

[s0ww0s]

Thanks to everyone, kernel 6.16.1 is working perfectly for me. I didn’t notice the package loss at first, but when I kept losing connection to the Hellldivers 2 servers, I took a closer look.

[lorebett]

I think due to this bug it’s nearly impossible to install or update a Neovim distribution with plugins, due to continuous failures to connect to github.

CachyOS 6.16.4-2-cachyos doesn’t seem to have the problem…

[albersc2]

Cachy and Nobara fixed the problem a few days ago. However, they used the worst of all available fixes, Brett’s initial one. I reported this situation on Nobara’s Discord, and they informed me that they already have a better fix scheduled for the next patch. I imagine it will be the same for Cachy, because both distros apply very similar patches.

You can read a short story about the mess with the fixes here:

Internet unstable after recent update

In my opinion, the folks at netdev bear a significant part of the responsibility. I joined their mailing list a week ago, and the truth is that they are juggling a lot of work. But let’s go over the history a bit—it’s quite interesting. Here is the commit that introduced the bug, dated July 15. They removed the line fi = NULL in __mkroute_output(). More than a month passed without anyone noticing the issue, until August 21–22, when kernel 6.16.2 was released. On that same day, Brett Sheffiel…

EDIT: More information: The CachyOS kernel that had the insecure fix was 6.16.3-2. It is almost sure that a better one will have been included in 6.16.4-2.

[albersc2]

This seems to confirm that everything is going well, and both Nobara and Cachy are currently resolving the issue with Brett’s second fix:

baseos/kernel/6.16/0001-cachyos-base-all.patch

[lorebett]

What drives me crazy is that I have 3 installation of EndeavourOS on the same computer, all up-to-date and all with the same network configuration.

On the one with Sway and on the one with Hyprland, when installing/updating Neovim plugins I sometimes get errors about github.com cannot be resolved. By specifying “1.1.1.1” as additional DNS server, everything works fine.

On the one with KDE, I always get errors of this shape, no matter what:

fatal: unable to access 'https://github.com/monaqa/dial.nvim.git/': Failed to connect to github.com port 443 after 117 ms: Could not connect to server

which completely blocks the installation…

[Moonbase59]

Personally, I don’t like CloudFlare, but you might want to add their IPv6 DNS servers, too, in case your machine/network/router/ISP can use IPv6:

• 2606:4700:4700::1111
• 2606:4700:4700::1001

Their other IPv4 DNS is 1.0.0.1.

Source: https://developers.cloudflare.com/1.1.1.1/ip-addresses/

[lorebett]

I’ve just tried with a different DNS server.
However, the main problem is still there: why on 2 installations changing DNS works while in the other one it just doesn’t have any effect… that’s what I don’t understand…

[Moonbase59]

That can depend on a multitude of factors:

• OS usage of IPv4 and/or IPv6
• resolv.conf settings
• Kernel version (and bugs)
• router and ISP settings (and use of IPv4/IPv6)
• Browser (some, if not most, browsers use their own certificates and possibly also DNS settings)

For instance, in my network, I let the router advertise the DNS server to be used (itself) via DHCP, and have set up my preferred DNS servers for the router’s WAN side within the router (for both IPv4 and IPv6).

The theory behind that is that all devices (OS-independent) use my router for DNS resolving, thus being able to access local devices via DNS names, while the router can cache, resolve and overall provide a better & faster interface to the devices. It can also handle DNS rebind situations, if configured for that. Another advantage is that I can set my preferred DNS resolvers in the router, and they are immediately available for all devices, without changes on the device.

This also makes running your own DNS server much easier (i.e., PiHole, Technitium, AdGuard, your own BIND, etc.).

And still I’ve seen browsers trying to override that… it can be a mess…

[noobr]

I click n this thread for a different reason, but, now that you mentioned it yes—it’s happened a couple times a day since last update. I thought it as my dns acting up as I also had issues with quad9 and then switched to cloud flare to resolve-it didn’t resolve the issue.

Same symptoms as everyone else0I open a browser and home page doesn’t load, I click refresh the home page loads. I do a search and click on of the links, page cannot be displayed, I refresh page loads fine.

It is not happening so much I consider it even an annoyance-it’s just a noticeable event in my day.

I use Firefox and Librewolf and it happens in both browsers.

Wierd