this more or less covers desktop_error_checks and desktop_
update_notification, though it can't really distinguish
between them easily. All we know is that if both the live and
postinstall versions of this test pass, both of those tests
pass. Any fails will have to be investigated manually.
Details
Run the tests for both KDE and Workstation, see
what happens. Workstation will fail for F25 and Rawhide at
present, due to SELinux/abrt notifications.
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.
I'm getting
+++ worker notes +++ end time: 2016-09-20 08:45:13 result: setup failure: /var/lib/openqa/share/factory/hdd/disk_Workstation-live-iso_64bit.qcow2 does not exist! uploading autoinst-log.txt
for desktop_notifications_postinstall test, but I guess it's problem somewhere in my openQA instance. Have you tried running it on staging?
yeah...well, at least, I did for KDE - e.g. https://openqa.stg.fedoraproject.org/tests/42974 - not totally sure if I did on Workstation. But it uses the same settings as desktop_terminal so I can't see how it would go wrong. It could just be that you tested with an image where the default_upload test fails? I don't recall precisely how the children's logs look in that case, and it may differ between openQA versions...
it looks like GNOME 'getting started' is working again in Rawhide today, so I'll do a run there just to check.
I landed this, but there's an issue with the KDE needles that I'll try and fix tomorrow.
OK, should be good now. Turns out the systray icons on KDE move all over the place so the 'no notifications' needle for KDE was not safe, but we can get a reliable 'no notifications' check by opening the 'extended' systray and clicking on Notifications. In fact I think we don't strictly speaking need to click on Notifications - simply its presence there implies that there are no fresh notifications, otherwise it's not there - but this seems safest in case upstream changes the interface or anything.