The node-agent-config load affinity https://velero.io/docs/main/data-movement-backup-node-selection/ is only implemented for Backup, not Restore.If OADP DPA spec.configuration.nodeAgent.nodeSelector is used and the datamover is scheduled to a node without a running node-agent. Even the cancellation hangs because there is no node-agent to cleanup the Pod.The Restore will hang until the nodeSelector is removed and forcing the Restore into PartiallyFailed.  Since this is a scheduling issue, it is inconsistent depending on datamover Pod placement. A node-selector specifying a single node only with large node counts makes this easily reproducable.node-agent nodeSelectors is not available in Velero, so this OADP only issue and not in any current release, only dev. Log bundles are included. Code tree used is here, upstream velero from 20241020 with a custom Dockerfile.ubi. https://github.com/msfrucht/openshift-velero/tree/velero_in_openshiftLog bundle is available here. https://ibm.box.com/s/0mj85vo155umgs7zv63voib190gw11pl (edi