QA » openQA Project » openQA Tests » qe-yam

I also came to the same assumption as Joaquin after my investigations. I've started 20 VRs, 10 of them were failed due to the same reason: https://openqa.suse.de/tests/overview?distri=sle&version=15-SP3&build=poo88524_oorlov

The reason seems to be as Joaquin described. Seems like Libyui selects Format device radio button while all the elements on the screen are not shown yet. So that radiobutton become selected, but event to enable Filesystem combobox is not triggered.

Sure, this looks like a product issue, but it is reproduced with artificial conditions and will not affect users, so most likely will not be fixed.

Manually the issue is not reproduced by me. Although, I've reproduced it by running the test module and stopping the execution once it is failed. If click the radio button again, then the event is fired.

So, I see two possible solutions here:

1. Hacky, dirty, but quick and possible to implement: select Format device radiobutton until Filesystem combobox is enabled;
2. Most proper, but I'm not sure yet if it is possible to implement: update libyui-rest-api to allow checking if an element is loaded and visible on screen. That will allow to not manipulate with the items once they loaded but not shown on UI.

It should work just ensuring that radiobutton 'do not format device' is selected so those events are chained in the right order.

sub select_format {
    my ($self) = @_;
    # wait for 'rb_no_format_device' selected
    return $self->{rb_format_device}->select();
}

Could you try that? I think that would not be hacky, only notice that if user select 'role: operating-system' and 'should_format: 1' we should give a hint/warning (no error) about wrong test data / no optimal selection, so basically it would mean to click on the radiobutton even is already selected, which I have seen in some places (but not sure if here). Other than that that check should be always true and similar with 'select_not_format'.

It should work just ensuring that radiobutton 'do not format device' is selected so those events are chained in the right order.

I'm afraid that this solution may be confusing when someone will use it in future. Just because if the one wants to select "Format Device" radio button, it is not known that it first checks if "Do not format device" is checked. In this case we would rely on such behavior that "Do not format device" is selected by default (but I'm not sure that this is true for all the cases) and if "Format Device" will be already selected then, the test may fail.

Anyways, I'm open for discussions, we can do that online if you want.

I've chosen the solution to select "Format Device" until "Filesystem" combobox is enabled, because they are linked together and when calling select_format_device method you do not care about anything. It just does what it should. I do not know if it is possible to fix the issue in libyui-rest-api itself right now, so this solution may be at least temporary.

The tests passed 10/10: https://openqa.suse.de/tests/overview?version=15-SP3&distri=sle&build=148.1_oorlov&groupid=96

Yes, it is better solution, thanks. I will paste it just for reference:
https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/11977