Explicitely check Kubernetes pull failure are traced as script_failure
The goal of this issue is o achieve here is explicitness and not implicitness. It wasn't clear that with the changes introduced in !2995 (merged), the behaviour of the Kubernetes executor was also changed.
With this issue, the intent is to update the Kubernetes executor tests in other to make sure that a ScriptFailure
is returned for pull image error.
More details in the thread linked below.
The following discussion from !2995 (merged) should be addressed:
-
@ratchade started a discussion: (+4 comments) @steveazz back to you
🏓 I was available to make more test and the results show that the change don't break existing runner system failure.
Existing integration tests,
testKubernetesMissingImage
,testKubernetesMissingTag
,testKubernetesFailingToPullImageTwiceFeatureFlag
,testKubernetesFailingToPullServiceImageTwiceFeatureFlag
,testKubernetesFailingToPullHelperTwiceFeatureFlag
, which were failing with a SystemFailureERROR: Job failed (system failure):
(in the main branch) now fail with a ScriptFailureERROR: Job failed:
.In the meantime, the
TestDeletedPodSystemFailureDuringExecution
is still failing as arunner_system_failure
I also rebased to make sure the changes pass the
integration_k8s tests
and it does.I believe it is safe to say that the changes turn the k8s pulling image error (missing image, missing tag, missing service image) into
script_failure
without breaking the existing the existingrunner_system_failure
Please, let me know if there is anything you believe I have missed
🙇🏿