Description
infer_bucket_sort_convert_join.q and mapjoin_hook.q failed after HIVE-9078 due to same reason.
With the following settings in 2 qfiles:
set hive.auto.convert.join = true; set hive.mapjoin.localtask.max.memory.usage = 0.0001; set hive.mapjoin.check.memory.rows = 2; set hive.auto.convert.join.noconditionaltask = false;
With MR, MapRedTask of join query would failed at first and then driver execute its backup task which would succeed, while we do not have backup task in Spark mode.