It looks like the path to the parquet file that you've used to infer the schema of the table has multiple directories that correspond to partitions (e.g., date=2015-08-17/hour=21). Create the table with the schema and partitioning specified, then use alter partition to add the per-partition files (https://www.cloudera.com/documentation/enterprise/latest/topics/impala_parquet.html) The -1 shown is probably from not running compute stats.
... View more
