CI build seems to have stalled?

[jcowgar]

The README states builds are made at least every week, but the latest is 8/29.

The build worked for me and allowed me to install Guix, this report is just more of a FYI.

[Erudition]

Seeing the same, check the Actions tab to see all the failing builds.... Here's the crash: https://github.com/SystemCrafters/guix-installer/actions/runs/7054056706/job/19202262401#step:4:2800

[daviwil]

This is a problem with the action that installs Guix, it apparently tries to download the latest nightly build of Guix from the CI server and fails because it can't be found. My guess is that they're cleaning up CI artifacts at a time before this scheduled job triggers.

See also: https://github.com/PromyLOPh/guix-install-action/issues/22

[moesasji]

The upstream bug is closed with a new release (v1.5) to fix this issue.

However it appears that the CI-action still fails, potentially because it is using v1 instead of the new v1.5 release. Is this worth another look?

[ghisvail]

It's a different error now.

It used to fail at the Install Guix stage with:

Run wget -nv https://ci.guix.gnu.org/search/latest/archive?query=spec:tarball+status:success+system:x86_64-linux+guix-binary.tar.xz -O guix-binary-nightly.x86_64-linux.tar.xz
https://ci.guix.gnu.org/download/1153:
2023-11-08 00:31:25 ERROR 500: Internal Server Error.
Error: Process completed with exit code 8.

Now it fails at the Build ISO stage with:

building /gnu/store/6mjig6012yf31cv4fgkracncjcx3hynl-linux-6.6.14.drv...
building /gnu/store/f6rp3sji60p3l35gwqn2wdvax8h6vypk-linux-firmware-20240115.tar.xz.drv...
..building /gnu/store/h9vggz1cjirjr32hrw71w14pr6dhlb5x-linux-firmware-20240115.drv...
substitute: 
substitute: updating substitutes from 'https://ci.guix.gnu.org/'...   0.0%guix substitute: error: TLS error in procedure 'write_to_session_record_port': Error in the push function.
guix system: error: `/gnu/store/c3ygnp9k0h5pnx9kibd8hkmmlq0z6kcy-guix-1.4.0-16.aeb4943/bin/guix substitute' died unexpectedly
Error: Process completed with exit code 1.

[ghisvail]

See https://github.com/PromyLOPh/guix-install-action/issues/23, same kind of error.

[ghisvail]

However it appears that the CI-action still fails, potentially because it is using v1 instead of the new v1.5 release. Is this worth another look?

See #28

[moesasji]

That unfortunately wasn't the solution; The challenge in resolving this is that the error message itself points to a network issue. For example guix bug 67914 has the exact same error message, where it is resolved by switching off VPN. Other bugs in debbugs with the same error message have a similar pattern, which unfortunately makes this one hard to debug as it might be the way the Ubuntu image is configured....

[moesasji]

The above guix bug that gets resolved by switching off a VPN (and other hard to reproduce network issues) makes me wonder whether this has something to do with MTU settings not being standard. MTU issues appears to be a "known" problem in github actions, see for example this bug report.

[port19x]

I tried with ubuntu-20.04 on my fork to no success. Was worth a try

[port19x]

I want to point out that as a stopgap solution, nonguix generates isos with the vanilla kernel and firmware as well. I did not know about that before just now, so I'll leave the link.

[ghisvail]

nonguix generates isos with the vanilla kernel and firmware as well

That's nice to hear.

I want to point out that as a stopgap solution

From your link:

** package updates including:
firefox 108.0.1, firefox-esr 102.6.0esr, linux 6.0.12, nvidia-driver 515.76
steam 1.0.0.75

That's quite outdated. Hence, the motivation to have an installer image updated and released regularly.

[port19x]

CI appears to work well again. Did you forget to close the issue?

[kjkent]

@port19x you jinxed it ;) it looks like it's currently failing due to a warning about an impending nodejs version bump