Even after adding enable_join_spill clause in a query as a hint, the join_spiil is still happenning.
Please see the attached snapshot from MC.
A join spill will occur if the hash table fills up the memory or hits the memory limit of the resource pool. When this occurs, the resources required to have attempted this join will have been wasted and the query is attempted again with the join hint /+add_vertica_options(EE,ENABLE_JOIN_SPILL)/. The hash table is then built out on disk