Re: Running a test from dist-git
On Wed, Oct 10, 2018 at 11:26:01PM +0200, Miroslav Vadkerti wrote:
Hi,
sorry for the late response :(
On Tue, Oct 2, 2018 at 9:42 AM Richard W.M. Jones <rjones(a)redhat.com> wrote:
> On Mon, Oct 01, 2018 at 11:26:59AM +0200, Miroslav Vadkerti wrote:
> > For this time I rescheduled the run:
> >
> >
> >
>
https://jenkins-continuous-infra.apps.ci.centos.org/blue/organizations/je...
>
> It appears to have failed, but I've got no clue why. How do I see
> the output of the command?
>
Ugh, logs are now gone. Which package were you testing? We can take a look
what was the problem.
https://src.fedoraproject.org/rpms/libguestfs
The test is to run a single command (libguestfs-test-tool). This
command is already installed by the package, so there's no need for
any test harness or script. If the command fails (ie. exit code != 0)
then it's vital we see the complete output in order to diagnose the
problem.
https://src.fedoraproject.org/rpms/libguestfs/blob/master/f/tests/tests.yml
Ideally we'd want to run the test as root and non-root, but for now I
just had it run the test as root.
> I think it would be easier if we could just drop a shell script
into
> tests/ and have it run that and display the output. The current
> system seems elaborately overengineered and I don't understand why.
>
Well, it was decided that we will use Ansible to wrap around our tests. Not
my idea. I agree the results could be presented better though, even if the
runner is Ansbile. We are preparing a dashboard to view the result of
testing, to give a better user experience in the future.
It wasn't very obvious to start off with, but it seems as if the
"standard test roles" are designed for the simple case of "just run a
script". However the use of YAML combined with the fact that you
can't easily run a test locally makes the whole thing very awkward.
(Also for some reason I didn't investigate yet, my copy of emacs
appears to hate YAML.)
If the pipeline goes through (i.e. tests were run), the pipeline ends
up
with Pass (test passed) or Unstable state (tests failed). If the test
failed, the pipeline did not execute the testing. In the former case you
should find logs in the Artifacts tab in Jenkins (test results are prefixed
with PASS_ or FAIL_). If you encounter an error, we are here to help an
investigate.
Do the tests run if we submit scratch builds?
Anyway feel free to submit a new build of libguestfs if you want to
see if you can make the test run.
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html