11.118. Release 0.123

General Changes

  • Remove node-scheduler.location-aware-scheduling-enabled config.
  • Fixed query failures that occur when the optimizer.optimize-hash-generationconfig is disabled.
  • Fix exception when using the ResultSet returned from theDatabaseMetaData.getColumns method in the JDBC driver.
  • Increase default value of failure-detector.threshold config.
  • Fix race in queueing system which could cause queries to fail with“Entering secondary queue failed”.
  • Fix issue with histogram() that can cause failures or incorrect resultswhen there are more than ten buckets.
  • Optimize execution of cross join.
  • Run Presto server as presto user in RPM init scripts.

Table Properties

When creating tables with CREATE TABLE or CREATE TABLE AS,you can now add connector specific properties to the new table. For example, whencreating a Hive table you can specify the file format. To list all available table,properties, run the following query:

  1. SELECT * FROM system.metadata.table_properties

Hive Changes

We have implemented INSERT and DELETE for Hive. Both INSERT and CREATEstatements support partitioned tables. For example, to create a partitioned tableexecute the following:

  1. CREATE TABLE orders (
  2.    order_date VARCHAR,
  3.    order_region VARCHAR,
  4.    order_id BIGINT,
  5.    order_info VARCHAR
  6. ) WITH (partitioned_by = ARRAY['order_date', 'order_region'])

To DELETE from a Hive table, you must specify a WHERE clause that matchesentire partitions. For example, to delete from the above table, execute the following:

  1. DELETE FROM orders
  2. WHERE order_date = '2015-10-15' AND order_region = 'APAC'

Note

Currently, Hive deletion is only supported for partitioned tables.Additionally, partition keys must be of type VARCHAR.