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

document and narrow usage of criu tests #5658

Open
judovana opened this issue Sep 28, 2024 · 3 comments
Open

document and narrow usage of criu tests #5658

judovana opened this issue Sep 28, 2024 · 3 comments

Comments

@judovana
Copy link
Contributor

I'm unable to run https://github.com/adoptium/aqa-tests/tree/master/external/*criu* tests. Is there some additional documentation? If no, can it be written?

In addition, it seems that those test are bound to only very specific subset of jvms, and images. That is causing quite intensive if-elif-elif-elif....fi.fi.fi in the external scripts.

  • Instead of this, can they simply exit if incorrect combination is set?
  • There seems to be mechanism, which allows to move to much vendor-tight tests out of main repo. Is this applicable here?
    • The main runners would lost osme weight

Thoughts please? Maye I'm just missing some part of history or documentation, but some pointers would be helpfull

@smlambert
Copy link
Contributor

smlambert commented Oct 8, 2024

CRIU is a feature of OpenJ9, currently unsupported in hotspot implementations (though there is some effort to support a similar functionality of checkpoint and restore via CRaC). So at this time, it is unlikely that you would need to run these tests unless you have been tasked with testing OpenJ9 or IBM's Semeru JDK.

You can see that all of the test targets are restricted to run against an OpenJ9 implementation

<impls>
	<impl>openj9</impl>
</impls>

in the associated playlists, example here. As such, would not be run by you, who are testing a hotspot implementation.

However, because you have suggested a major overhaul of the external test scripts in #5553, I asked Longyu to check that those changes did not break the testing for the OpenJ9 implementation, as a courtesy to our colleagues from IBM who add many features and help support the overall AQAvit approach to testing. In the relatively near future, we will also be able to pull images and run these tests at ci.adoptium.net (thus removing the bottleneck on PR testing, but at present the UBI images are not in a conveniently accessible location).

In this particular case, I do not want these tests pulled out of the aqa-tests repo, because I believe we will eventually need to support a checkpoint and restore feature in hotspot (once that work progresses upstream).

@smlambert
Copy link
Contributor

Please feel free to capture the above information, or information about the CRIU feature, https://blog.openj9.org/tag/criu/, and convert it into documentation to close off this issue you have raised.

@judovana
Copy link
Contributor Author

judovana commented Oct 8, 2024

I'm quite familiar with everything what you wrote, including that blog post. And I'm really glad you asked IBM pelpe to try that ou. I'm really grateful. But before we (anybody) ask them, I would like to test them as close as they are expected to be run as possible. I was unable to trigger locally many code flow paths, withuut actually . By other word I would like to test locally openj9 images. I'm probably missing something basic, but I have really failed in local reproducibility of openj9 targeted targets (on openj9 of course).

By but at present the UBI images are not in a conveniently accessible location I think It just confirmed what I thought. That no one can test those code flow paths.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Status: Todo
Development

No branches or pull requests

2 participants