enable ENTRYPOINT to be more than one test, ensure shutdown when on ARM
ClosedPublic

Authored by garretraziel on Aug 15 2016, 11:23 AM.

Details

Summary

This adds possiblity to specify more than one test in ENTRYPOINT
variable - test names should be separated with spaces and loading is done
as with POSTINSTALL.

Add _console_shutdown test to ARM image deployment.

Test Plan

run install_arm_image_deployment_upload, followed by base_services_start_arm
and ensure that disk was created in the first test and second test run without problem.

Diff Detail

Repository
rOPENQATESTS os-autoinst-distri-fedora
Lint
Automatic diff as part of commit; lint not applicable.
Unit
Automatic diff as part of commit; unit tests not applicable.
garretraziel retitled this revision from to enable ENTRYPOINT to be more than one test, ensure shutdown when on ARM.Aug 15 2016, 11:23 AM
garretraziel updated this object.
garretraziel edited the test plan for this revision. (Show Details)
garretraziel added a reviewer: adamwill.

So I've chosen to enable ENTRYPOINT to have more than one test specified to solve this issue, rather than reorganizing main. Reasoning is that our ARM tests really aren't "installation" tests per se, so it wouldn't make sense to let them go through "installation" and "postinstallation" phases. We could separate "ensure shut down" phase to be run outside "postinstallation", but there are tests for which we don't need to ensure shutting down so I've simply added _console_shutdown test to be run after install_arm_image_deployment. I've tested it and both ARM tests should run (but they've passed on my computer even before).

adamwill accepted this revision.Aug 15 2016, 11:59 PM

"I've tested it and both ARM tests should run (but they've passed on my computer even before)."

Yeah, this has changed a bit with the recent big changes to VM shutdown upstream, I think. I think that before those changes, even if we didn't shut down the VM as part of the test, it got auto-killed somehow before the disk upload stage. Now it doesn't, and that check is hit.

I recall that we didn't shut down the system in the upload tests at first, and they didn't hit this error at that time. So I'm pretty sure it's down to the recent upstream changes. If you're running an older os-autoinst locally, it's definitely plausible that it wouldn't hit the error.

The diff looks good, thanks. If we start getting lots more ARM tests it would maybe make sense to do this with main.pm logic - maybe we just rebrand postinstall to something a bit more generic - but for now this is fine.

This revision is now accepted and ready to land.Aug 15 2016, 11:59 PM
This revision was automatically updated to reflect the committed changes.