Search code examples
clouderaparquetimpala

cloudera impala PARQUET_FALLBACK_SCHEMA_RESOLUTION

It is possible to configure Cloudera Impala (5.12) to default to name instead of position for PARQUET_FALLBACK_SCHEMA_RESOLUTION?

My Parquet files don't always have the same set of columns so we need Impala to look them up by name rather than position, and its a bit of a pain to do this in Hue for every session:

set PARQUET_FALLBACK_SCHEMA_RESOLUTION=name;
Solution

  • Thanks for the info Amos,

    I posted the same question on the Cloudera forums and they pointed me to a way to configure this thru the Cloudera Manager.

    http://community.cloudera.com/t5/Interactive-Short-cycle-SQL/PARQUET-FALLBACK-SCHEMA-RESOLUTION/m-p/62318#M3883