diff --git a/pkg/monitortestlibrary/pathologicaleventlibrary/duplicated_event_patterns.go b/pkg/monitortestlibrary/pathologicaleventlibrary/duplicated_event_patterns.go index 97bbd12865b4..00753a327e3c 100644 --- a/pkg/monitortestlibrary/pathologicaleventlibrary/duplicated_event_patterns.go +++ b/pkg/monitortestlibrary/pathologicaleventlibrary/duplicated_event_patterns.go @@ -445,6 +445,22 @@ func NewUniversalPathologicalEventMatchers(kubeConfig *rest.Config, finalInterva neverAllow: true, }) + // Ignore repeated TopologyAwareHintsDisabled events from the endpoint + // slice controller's topology cache. The controller emits these events + // if a service (such as the cluster DNS service) enables topology-aware + // hints but certain "safeguards" (such as having sufficient service + // endpoints and sufficient nodes with allocatable CPU in each zone) are + // not met, in which case the default non-topology-aware behavior is + // used. This means that the service remains reachable even if these + // events are emitted. Note also that ovn-kubernetes does not implement + // topology-aware hints; the feature is enabled only for the benefit of + // third-party CNI plugins, such as Calico. + registry.AddPathologicalEventMatcherOrDie(&SimplePathologicalEventMatcher{ + name: "TopologyAwareHintsDisabled", + messageReasonRegex: regexp.MustCompile(`^TopologyAwareHintsDisabled$`), + jira: "https://issues.redhat.com/browse/OCPBUGS-69400", + }) + registry.AddPathologicalEventMatcherOrDie(AllowBackOffRestartingFailedContainer) registry.AddPathologicalEventMatcherOrDie(AllowOVNReadiness)