action #112577
Updated by JERiveraMoya over 2 years ago
### Motivation
The [mau-extratests-yast2ui](https://openqa.suse.de/tests/9119936#) testsuite is sometimes affected by [bsc#1191112](https://bugzilla.suse.com/show_bug.cgi?id=1191112).
For now we should apply the same workaround that we used in [poo#105046](https://progress.opensuse.org/issues/105046).
### Scope
Affected modules are [yast2_bootloader](https://openqa.suse.de/tests/9119936#step/yast2_bootloader/17) and [yast2_instserver](https://openqa.suse.de/tests/9119936#step/yast2_instserver/14), currently scheduled in the `mau-extratests-yast2ui_dev` testsuite that runs in our [development group](https://openqa.suse.de/group_overview/446)
### Acceptance criteria
**AC1**: Add workaround make the next screen visible in the affected areas
**AC2**: Reference [bsc#1191112](https://bugzilla.suse.com/show_bug.cgi?id=1191112) as soft failure when applying the workaround
### Suggestion
Send f10 to maximize and unmaximize the window n times, as done for example [here](https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/14342/files#diff-e0260d845efecccf481402143f2fac57540bcc4f5ff550981e513d5b11a7c7ccR96). This forces the YaST screen to, at some point, refresh the proper content, and `send_key_until_needlematch` will match the needle then and succeed.
Care must be taken with `send_key_until_needlematch`, as the number of keys sent is one more than specified in the command arguments, so I would suggest you stick to odd numbers for the repetitions (eg `9` instead of `10` in the case of `send_key_until_needlematch('yast2_control-center_wake-on-lan_overview', 'alt-f10', 9, 2);`).
Otherwise the screen will not match because the window will be left maximized.
### Further info
Fix the same problem if appear multiple times, but if it is a different issue, file a different ticket for it.
Since we have noticed this issue only for MU sp4, it might be worth it, at some point, to convert the `mau-extratests-yast2ui` testsuite to use libyui rest api. This would overstep the current refresh issue as the rest api would provide the correct screen information.
In that case this workaround would need to be reverted.