It would be handy if we had a reference tool for tailing and recording the oplog to file(s), e.g., for roll-forward recovery from backups/snapshots, for feeding into external indexing systems, etc. Ideally, such a tool would work with single replica sets or sharded clusters.