Currently, when deploying {product-short} using the Helm Chart, two replicas cannot run on different cluster nodes. This might also affect the upgrade from 1.3 to 1.4.0 if the new pod is scheduled on a different node. A possible workaround for the upgrade is to manually scale down the number of replicas to 0 before upgrading your Helm release. Or manually remove the old {product-short} pod after upgrading the Helm release. However, this would imply some application downtime. You can also leverage a Pod Affinity rule to force the cluster scheduler to run your {product-short} pods on the same node.