Kubernetes configuration. We currently only support the use case when infra validator is run by orchestration.KubeflowDagRunner.

Model server will be launched in the same namespace KFP is running on, as well as same service account will be used (unless specified). Model server will have ownerReferences to the infra validator, which delegates the strict cleanup guarantee to the kubernetes cluster.

active_deadline_seconds int32 active_deadline_seconds
service_account_name string service_account_name
serving_pod_overrides PodOverrides serving_pod_overrides