Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Containers: fix schedule for JeOS #13672

Merged
merged 1 commit into from
Nov 22, 2021
Merged

Containers: fix schedule for JeOS #13672

merged 1 commit into from
Nov 22, 2021

Conversation

jlausuch
Copy link
Contributor

This fixes:
https://progress.opensuse.org/issues/102272
https://progress.opensuse.org/issues/102278

This basically enables running the default jeos tests as it was done before, and then run the specific container tests.

In a follow-up PR, we could come up with a more smart default jeos test for container jobs only, instead of running all the default ones, but this is a different topic than this PR tries to fix.

VRs:

Some things need to be done in job configuration:

  • Add NUMDISKS=2 for validate_btrfs test
  • Remove BOOT_HDD_IMAGE -> JeOS tests don't need that variable.
  • Remove CONTAINERS_UNTESTED_IMAGES -> this is a HOST test based to test container tools and packages on JeOS. We already have a job to validate the "untested" container image: https://openqa.opensuse.org/tests/2029146. If we want to enable also container image validation on JeOS, we should create a separate job.

@jlausuch
Copy link
Contributor Author

Please be aware of this proposal: #13668
But I would really prefer to re-use code and not duplicating loadtest calls.

@b10n1k
Copy link
Contributor

b10n1k commented Nov 12, 2021

Please be aware of this proposal: #13668 But I would really prefer to re-use code and not duplicating loadtest calls.

i am not sure any more. i have a few observations. they are minor atm because i think it is more important to solve the issue. So in theory we can just merge this and discuss further later. However still cant decide as this PR seems to require additional commit for the variables you mentioned. ping me whenever you want

@jlausuch
Copy link
Contributor Author

Please be aware of this proposal: #13668 But I would really prefer to re-use code and not duplicating loadtest calls.

i am not sure any more. i have a few observations. they are minor atm because i think it is more important to solve the issue. So in theory we can just merge this and discuss further later. However still cant decide as this PR seems to require additional commit for the variables you mentioned. ping me whenever you want

To fix the scheduling issue, no new PRs are needed. The variables will be set in the UI.

Can you have a look at validate_btrfs test? Any ideas why it fails?

@jlausuch
Copy link
Contributor Author

VR to make sure default JeOS schedule is not modified:
https://openqa.opensuse.org/tests/2031338

@pdostal
Copy link
Member

pdostal commented Nov 12, 2021

Related to: #13645

@jlausuch
Copy link
Contributor Author

Ok, wait. This change breaks these tests:
https://openqa.opensuse.org/tests/1995349#

But I would say that the job setup for this test is completely wrong. It's booting a CentOS HDD using JeOS flavor... I guess we need to change that, create a new flavor or re-use existing one...

@jlausuch
Copy link
Contributor Author

IMO, we should use this test to run the container tests after hdd creation and maybe place them in Leap 15.3 Updates job instead of Leap 15.3 images.

@b10n1k
Copy link
Contributor

b10n1k commented Nov 12, 2021

Please be aware of this proposal: #13668 But I would really prefer to re-use code and not duplicating loadtest calls.

i am not sure any more. i have a few observations. they are minor atm because i think it is more important to solve the issue. So in theory we can just merge this and discuss further later. However still cant decide as this PR seems to require additional commit for the variables you mentioned. ping me whenever you want

To fix the scheduling issue, no new PRs are needed. The variables will be set in the UI.

Can you have a look at validate_btrfs test? Any ideas why it fails?

validate_btrfs was not part of the scheduling there. the reason is likely the unmatched size which this module is quite sensitive.

@@ -80,7 +80,7 @@ sub load_host_tests_docker {
loadtest 'containers/registry' if is_x86_64;
loadtest 'containers/docker_compose';
}
loadtest 'containers/validate_btrfs' if is_x86_64;
loadtest 'containers/validate_btrfs' if (is_x86_64 && !is_jeos);
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This module is not executed in SLE JeOS either, so I created this ticket to enable it later on.
https://progress.opensuse.org/issues/102377

@jlausuch
Copy link
Contributor Author

My idea is to reorder the O3 test like this:

Screenshot_20211114_190246

http://fromm.arch.suse.de/tests/overview?distri=opensuse&version=15.3&build=9.279&groupid=50
http://fromm.arch.suse.de/tests/overview?distri=opensuse&version=15.2&build=31.593&groupid=51

This would also need like this:
os-autoinst/openqa-trigger-from-obs#145

@jlausuch
Copy link
Contributor Author

@Vogtinator @DimStar77 Would you be ok with my proposal to group container image tests so everything looks more tidy?

@Vogtinator
Copy link
Member

@Vogtinator @DimStar77 Would you be ok with my proposal to group container image tests so everything looks more tidy?

Yes! That was pretty much the goal from the beginning, but implementing that in openqa-trigger-from-obs isn't trivial (See discussion on os-autoinst/openqa-trigger-from-obs#145), particularly because of BUILD number differences.

It would be possible to decouple container images into a separate project, but that is arguably overkill and makes it harder (if not practically impossible) to use latest JeOS as host, which is simpler and quicker than using a GM image + installing updates.

@jlausuch
Copy link
Contributor Author

The following ticket has been resolved: https://progress.opensuse.org/issues/97958
So, we have now an ISOS POST with the flavor Container-Image to separate the container tests that boot some HDD (CentOS, Ubuntu, etc) from JeOS flavor, so we don't have that dependency any more.
So, now, JeOS flavor is ONLY used for JeOS tests, and Container-Image flavor for image validation.

This is ready to be merged.

@mloviska mloviska merged commit ddcce2f into os-autoinst:master Nov 22, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
5 participants