As part of sharded $lookup, if the origin collection is sharded we will allow the $lookup stage to be moved to the shards part of the splitPipeline. We believe that executing the $lookup in parallel will be beneficial in most cases, however there are times when this is the wrong execution strategy. To avoid regressing existing workloads where the 'from' collection is unsharded, investigate the feasibility of "backing out" of this execution strategy in the DistributedPlanLogic of the DocumentSourceLookUp. The check for whether the 'from' collection is unsharded must be concurrency safe.