Shortly after migrating several Spark batch pipelines from on-premises infrastructure to Azure Kubernetes Service (AKS), we began seeing repeated executor OOM failures in one of our larger jobs. The ...