-
Notifications
You must be signed in to change notification settings - Fork 188
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Unbreak various tests in CI #3371
Conversation
I've updated the PR:
|
Ok, I'm confused the test results show as red here but visiting: https://ci.centos.org/job/bodhi-pipeline/job/PR-3371/3/ I see is it still running? |
Right now it still shows 6 as "expected". While the tests are in flight, I always look into the Pipeline Steps and, of the ones marked red, check their log (terminal symbol beside the red bubble). Basically, we have one issue in the pip-integration/-build tests with an unsigned package, and the sphinx problem in rawhide-docs. |
ae09f02
to
d935451
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nice work tracking this down! :)
This is an obscure and hard to debug (therefore unfixed) problem in DNF which caused installing Fedora Infrastructure RPM packages to fail in CI containters. It's documented at: https://bugzilla.redhat.com/show_bug.cgi?id=1699396 Thanks to Brian Stinson for the suggested workaround! Signed-off-by: Nils Philippsen <nils@redhat.com>
@nphilipp Will |
Yes, we checked that, it was added in 2007 (iirc) |
Prior to pytest 5.0, ExceptionInfo objects had a __str__() method of their own which rendered the wrapped exception as a string. It was removed in version 5.0 which made calling str() on the object use __repr__() as a fallback which gives totally different results. Here we use str() on the .value member (i.e. the wrapped exception) and a variable name that makes it (more) explicit that the context manager isn't an exception. Signed-off-by: Michal Konečný <mkonecny@redhat.com> Signed-off-by: Nils Philippsen <nils@redhat.com>
As discussed, let's broaden the scope of this PR and include:
|
Oh, and as discussed I've replaced |
RUN rpm --import "$(awk -F= '{if ($0 ~ "gpgkey *=") {print $2;}}' < /etc/yum.repos.d/infra-tags.repo)" | ||
# deal with https://fedoraproject.org/wiki/Changes/Set_skip_if_unavailable_default_to_false | ||
RUN dnf install -y dnf-plugins-core | ||
RUN dnf config-manager infrastructure-tags --setopt skip_if_unavailable=True --save |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Every command adds a layer in the container, do you know if Randy and Aurélien preferred the one command / line approach? Otherwise, it is often recommended, to have multiple command per "RUN" using \
to delimit them.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't know their preference. In the Dockerfiles there are instances of one RUN line containing multiple commands (though not many), but also multiple consecutive RUN lines each containing a single command.
For reference, here's an example of |
In previous versions, skip_if_unavailable defaulted to True so the missing infrastructure-tags repository for the Rawhide release version didn't make tests fail. Restore this setting for the infrastructure-tags repository. Signed-off-by: Nils Philippsen <nils@redhat.com>
So only thing failing now will be rawhide docs test? Or this is a new issue? |
No, it's that |
I see that there are plenty of integration tests failing, but this will be probably related to #3370 |
This is because pydocstyle 4.0.0 can't cope with arguments described in the docstrings of nested functions, as in bodhi/server/renderers.py: PyCQA/pydocstyle#370 Signed-off-by: Nils Philippsen <nils@redhat.com>
As discussed: considering that we verified the changes in this PR to fix what they're supposed to do and that they aren't causing the remaining and new CI pipeline failures, I'll attempt to merge them manually, which should make fixing the outstanding issues easier as you don't have to dig through the errors this one is fixing. |
This should fix the failing CI tests
Signed-off-by: Michal Konečný mkonecny@redhat.com