On-prem the distro and architecture are set after the wizard has been
opened. This triggers a reload of the image types and makes the tests
very flaky.
This tmt[0] test runs the playwright tests as gating tests. Having the
gating tests upstream avoids duplication across fedora and centos
dist-git repositories, and running them upstream should keep them in
working order.
Only add x86_64 for now, the aarch runners seem to be a bit too slow.
[0]: https://tmt.readthedocs.io/en/stable/index.html
There seems to be a change in behaviour from Fedora 41 to Fedora 42.
When a script is executed remotely a non-interactive session is created.
However, the user session with noTTY in Fedora F42 is picked up by the
`who`, `w` & `users` commands.
Since we run a remote script that checks for logged in users with the
`who` command, the `update_github_status.sh` script blocks and creates
an endless loop. This happens because the session only closes once the
remote script has finished executing.
The fix is a bit ugly, but we can filter this session out by checking
for user sessions that don't have a terminal device `?`.
Makes the test's base url, user & password configurable through the
environment. Abstracts differences between hosted and cockpit where
possible to lib, and adds conditionals otherwise.
Snyk is now being used for code analysis in favor of Sonarqube. This
commit drops Sonarqube. Schutzbot was only neccessary for running
SonarQube, so it has also been dropped.
This adds the ability to use our Schutzbot Gitlab CI and run Sonarqube
scan there. We have pretty much the exact same thing in weldr-client
repo and use it only for Sonarqube. This could also be used in the
future if there is any need to use our own CI.
The added scan is just informative and is by no means supposed to be
used to gate PRs, there will be just one more link to
check the results in case anyone is interested.
