consider a case in netflix conductor (or orkes) where 2 subworkflows are forked in paralel. The default behaviour of FORK + SUB_WORKFLOW + JOIN tasks is that if the first subworkflow fails, the second one is aborted midway through execution and CANCELLED. The workaround suggested to run both to completion is to mark both SUBWORKFLOW tasks as optional. However the consequence here is that the parent workflow now ends up always COMPLETED only showing the failed SUB_WORKFLOW as "COMPLETED_WITH_ERRORS".
Is there a simple way to have both subworkflows run to completion but have the parent fail if one or more of the subworkflows fail? It feels like that should be the responsibility of the JOIN task but that one is really not configurable.