These are the known problems as of DSS 4.1.2
- Window recipe: in DSS engine, lag computation is wrong when overriding the name of the lag column.
- Scoring using a deployed ensemble fails until said ensemble has been retrained
- Ensembling ensembles is not supported when multi-user-security is enabled
- The evaluation recipe outputs invalid metric values (F1, Recall, Precision) for binary classification
- Writing datasets from R notebooks fails when DSS is in multi-user-security mode
- Multi-user-security mode: Hive recipe validation fails if UNIX user is not the same as Hadoop user. Please contact Dataiku support for workarounds