test/e2e: Wrap catalog template image adjustment test in eventually blocks #2397
Conversation
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: timflannagan The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
…andlers
Update the "adding catalog template adjusts image used" test block and
wrap flake-sensitive methods in eventually handlers. This also attempts
to fix a flake that occasionally pops up in e2e runs:
```log
• Failure [60.555 seconds]
Catalog represents a store of bundles which OLM can use to install Operators
/home/tflannag/Documents/openshift/operator-framework-olm/staging/operator-lifecycle-manager/test/e2e/catalog_e2e_test.go:35
adding catalog template adjusts image used [It]
/home/tflannag/Documents/openshift/operator-framework-olm/staging/operator-lifecycle-manager/test/e2e/catalog_e2e_test.go:1036
Timed out after 60.001s.
Expected success, but got an error:
<*errors.errorString | 0xc0009e47b0>: {
s: "resource name may not be empty",
}
resource name may not be empty
```
Update the `sourceName` -> `source.GetName()` when firing off that
problematic GET call.
Signed-off-by: timflannagan <[email protected]>
timflannagan
force-pushed
the
update-adding-catalog-template-image-e2e-test
branch
from
80a586a to
e046611
Compare
openshift-ci
bot
added
the
needs-rebase
|
@timflannagan: PR needs rebase. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
|
/lgtm |
| source, err = crc.OperatorsV1alpha1().CatalogSources(source.GetNamespace()).Get(context.TODO(), source.GetName(), metav1.GetOptions{}) | ||
| if err != nil { | ||
| return err | ||
| } | ||
| // create an annotation using the kube templates | ||
| source.SetAnnotations(map[string]string{catalogsource.CatalogImageTemplateAnnotation: fmt.Sprintf("quay.io/olmtest/catsrc-update-test:%s.%s.%s", catalogsource.TemplKubeMajorV, catalogsource.TemplKubeMinorV, catalogsource.TemplKubePatchV)}) | ||
|
|
There was a problem hiding this comment.
Revisiting this as it's been sitting for a while: poking at this locally, it looks like the real issue is we're overwriting this shared source variable in the GET and UPDATE calls. Speculatively, it looks like anytime the update call fails, the source variable will be essentially nil:
catalogsource info: catalog-9t8pz/openshift-operators
catalogsource info: &v1alpha1.CatalogSource{TypeMeta:v1.TypeMeta{Kind:"", APIVersion:""}, ObjectMeta:v1.ObjectMeta{Name:"catalog-9t8pz", GenerateName:"", Namespace:"openshift-operators", SelfLink:"", UID:"1151883a-39f2-44a0-9d42-8c2e28505b47", ResourceVersion:"32421", Generation:1, CreationTimestamp:v1.Time{Time:time.Time{wall:0x0, ext:63772780625, loc:(*time.Location)(0x46b9aa0)}}, DeletionTimestamp:(*v1.Time)(nil), DeletionGracePeriodSeconds:(*int64)(nil), Labels:map[string]string{"olm.catalogSource":"catalog-9t8pz"}, Annotations:map[string]string(nil), OwnerReferences:[]v1.OwnerReference(nil), Finalizers:[]string(nil), ClusterName:"", ManagedFields:[]v1.ManagedFieldsEntry{v1.ManagedFieldsEntry{Manager:"e2e.test", Operation:"Update", APIVersion:"operators.coreos.com/v1alpha1", Time:(*v1.Time)(0xc0001be498), FieldsType:"FieldsV1", FieldsV1:(*v1.FieldsV1)(0xc0001be480), Subresource:""}}}, Spec:v1alpha1.CatalogSourceSpec{SourceType:"grpc", Priority:0, ConfigMap:"", Address:"", Image:"quay.io/olmtest/catsrc-update-test:old", UpdateStrategy:(*v1alpha1.UpdateStrategy)(nil), Secrets:[]string(nil), DisplayName:"", Description:"", Publisher:"", Icon:v1alpha1.Icon{Data:"", MediaType:""}}, Status:v1alpha1.CatalogSourceStatus{Message:"", Reason:"", LatestImageRegistryPoll:(*v1.Time)(nil), ConfigMapResource:(*v1alpha1.ConfigMapResourceReference)(nil), RegistryServiceStatus:(*v1alpha1.RegistryServiceStatus)(nil), GRPCConnectionState:(*v1alpha1.GRPCConnectionState)(nil), Conditions:[]v1.Condition(nil)}}
catalogsource info: /
catalogsource info: /
...
One solution is to just avoid overwriting this source resource by suppressing the variable returned from the update call. I tested out this variation locally and I wasn't able to reproduce this issue anymore.
timflannagan
deleted the
update-adding-catalog-template-image-e2e-test
branch
Update the "adding catalog template adjusts image used" test block and
wrap flake-sensitive methods in eventually handlers. This also attempts
to fix a flake that occasionally pops up in e2e runs:
Update the
sourceName->source.GetName()when firing off thatproblematic GET call.