search

SUSE-Enceladus / img-proof

img-proof provides a command line utility to test images in the Public Cloud
https://img-proof.readthedocs.io/en/latest/
GNU General Public License v3.0
14 stars 7 forks source link

The tty for aarch64 in 15-SP5 is no longer tty1 #359

Closed ricardobranco777 closed 1 year ago

ricardobranco777 commented 1 year ago

Related ticket: https://progress.opensuse.org/issues/126002 Failing test: https://openqa.suse.de/tests/10835083

Solves the issue with: E AssertionError: assert 'console=tty1' in ['BOOT_IMAGE=/Image-5.14.21-150500.46-default', 'root=UUID=41a60a97-0d45-480b-bd09-63aba9d14659', 'console=ttyS0', 'net.ifnames=0', 'dis_ucode_ldr', 'earlyprintk=ttyS0', ...]

rjschwei commented 1 year ago

Should have read the ticket first. The test failure is a false positive. Basic images are not published and not setup for aarch64. It is not possible in the build service to skip a specific architecture build for a specific flaivor.

ricardobranco777 commented 1 year ago

Should have read the ticket first. The test failure is a false positive. Basic images are not published and not setup for aarch64. It is not possible in the build service to skip a specific architecture build for a specific flaivor.

You should have read the PR description to notice that it didn't "disappear". It changed to console=ttyS0.

It was very rude of you to just close this PR without further conversation, without even proposing a fix. We in QE-C need to solve this test issue. I will propose to use a fork.

rjschwei commented 1 year ago

The ticket states

openQA test in scenario sle-15-SP5-Azure-Basic-ARM-x86_64-publiccloud_img_proof@64bit fails in

Now the architecture is probably x86_64 and ARM in this case stands for Azure Resource Manager. The PR is for changes in the aarch64 implementation of the test. The image definition for 15 SP5 for aarch64 contains

kernelcmdline="console=tty1....

So again, Azure-Basic are not setup for aarch64 because we do not publish those and that should not be tested. But given that the test is, based on the name most likely being run on x86_64 a change in the aarch64 section of the test would not actually make a change.