Closed sigalsax closed 1 year ago
Flakiness was previously 1 in every 2 runs was a failure and now it is 1 in every ~4
Feedback from @ismarc : I dug into this some yesterday, and there are still docker pull secrets in use. The reason they're still in use is to pull images across projects for the tests. I believe that what needs to happen is the image-puller role needs to be assigned for the test project against the project the images were uploaded to rather than using the docker pull secret for access (edited)
The reason being that there is a max age on the docker pull secret after which it stops working, so performance of steps prior to it being finished being used can impact whether it expires or not
https://docs.openshift.com/container-platform/4.5/openshift_images/managing_images/using-image-pull-secrets.html#images-allow-pods-to-reference-images-across-projects_using-image-pull-secrets covers how to add a project as an image-puller for another project and I believe will fully resolve this issue
Summary
In our secrets provider pipeline we are experiencing flaky tests on helm tests. The failure occurs on
Running './TEST_ID_17_helm_job_deploys_successfully.sh'
and the error resembles the following:The test failures only occur on OCP current 4.6
Steps to Reproduce
Steps to reproduce the behavior:
Reproducible
Version/Tag number
OCP 4.6
Environment setup
Pipeline build
Additional Information
-