pkg/apis/core/validation TestValidatePersistentVolumeClaimUpdate 0.01s

[deads2k]

https://openshift-gce-devel.appspot.com/build/origin-ci-test/pr-logs/pull/17749/test_pull_request_origin_unit/6887/

validation_test.go:1428: Unexpected failure for scenario: valid-size-update-resize-enabled - [spec: Forbidden: is immutable after creation except resources.requests for bound claims]

github.com/openshift/origin/vendor/k8s.io/kubernetes/pkg/apis/core/validation TestValidatePersistentVolumeClaimUpdate 0.01s

validation_test.go:1428: Unexpected failure for scenario: valid-size-update-resize-enabled - [spec: Forbidden: is immutable after creation except resources.requests for bound claims]

@openshift/sig-storage

[0xmichalis]

https://openshift-gce-devel.appspot.com/build/origin-ci-test/pr-logs/pull/batch/test_pull_request_origin_unit/6932/

[deads2k]

Well, this flakes because it is racing another test that is also twiddling global feature flags.

[gnufied]

/assign

[deads2k]

opened kubernetes/kubernetes#57199 to fix it upstream

[deads2k]

@liggitt want kubernetes/kubernetes#57199 or want a giant lock for a pick?

[gnufied]

@deads2k are you sure this is because of feature flag? it sounds like, the feature flag was turned on - https://github.com/kubernetes/kubernetes/blob/master/pkg/apis/core/validation/validation.go#L1797 but the PVC had some other change other than storage size.

[gnufied]

This is related to kubernetes/kubernetes#56395

[deads2k]

This is related to kubernetes/kubernetes#56395

In a serially run for loop, how would that mutation flake?

[deads2k]

This is related to kubernetes/kubernetes#56395

As an alternative theory, the global feature flag was switched before you entered the if and instead of tolerating some changes, you ran in the very tight "only allow X" block.

[liggitt]

In a serially run for loop, how would that mutation flake?

testcases are iterating over a map, so order is random

[gnufied]

As an alternative theory, the global feature flag was switched before you entered the if and instead of tolerating some changes, you ran in the very tight "only allow X" block.

I am not sure how will that happen. My theory is -ValidatePersistentVolumeClaimUpdate mutates the oldPVC at least in 2 places and each test case gets same copy of pvc (at least for old version). It is likely that one test case is using side-effect of another test case.