Release 0.232
There is a bug in this release that will cause queries with the predicate on bucketed columns to produce incorrect results.
Fix an issue where
DATE_TRUNC
may produce incorrect results at certain timestamp inAmerica/Sao_Paulo
.Improve built-in function resolution performance by caching function resolution results.
Add a MySQL-based function namespace manager implementation that supports creating, altering, and dropping SQL functions (Function Namespace Managers).
Add configuration property
use-legacy-scheduler
and session propertyuse_legacy_scheduler
to use a version of the query scheduler before refactorings to enable full stage retries. The default value is false. This is a temporary property to provide an easy way to roll back in case of bugs in the new scheduler. This property will be removed in a couple releases once we have confidence in the stability of the new scheduler.Add
query_max_total_memory_per_node
andquery_max_memory_per_node
session properties.Add support to show whether functions are built-in in
SHOW FUNCTIONS
.Add
use_exact_partitioning
session property that forces repartitioning if repartitioning is possible.Add support for .
Add configuration property
resource-groups.reload-refresh-interval
to control the frequency of reloading resource group information from the database. The default value is 10 seconds.Add support for using the Thrift protocol to shuffle data. This can be configured using the configuration property
internal-communication.task-communication-protocol
. Possible values are HTTP or Thrift.Add support for using the Thrift protocol to announce node state. This can be configured using the configuration property
internal-communication.server-info-communication-protocol
. Possible values are HTTP or Thrift.Add session property
list_built_in_functions_only
to support hiding user-defined SQL functions inSHOW FUNCTIONS
.
Hive Changes
Fix an issue where queries could fail with buffer overflow when writing ORC files.
Add support for handling statistics to the Alluxio metastore.
Add Alluxio metastore which connects to the Alluxio catalog service <https://docs.alluxio.io/os/user/2.1/en/core-services/Catalog.html>.
Add session property
shuffle_partitioned_columns_for_table_write
to make Presto shuffle data on the partition columns before writing to partitioned unbucketed Hive tables. This increases the maximum number of partitions that can be written in a single query by a factor of the total number of writing workers. The property isfalse
by default. ().Expose Hive table properties via system table$properties table.
Change error code from
HIVE_METASTORE_ERROR
toHIVE_TABLE_DROPPED_DURING_QUERY
when aDROP TABLE
query fails due to another query dropping the table before this query has finished.Upgrade Alluxio version from 2.1.1 to 2.1.2.
- Add
Kerberos
authentication.
Kafka Changes
- Update
Kafka
connector to 2.3.1, which improves implementation and performance ().
- Replace config
pinot.prefer-broker-queries
with the inverse configpinot.forbid-broker-queries
.
Verifier Changes
- Replace
IsHidden
attribute on AggregationFunction and ScalarFunction withvisibility
which can be of the following valuesPUBLIC
, ,HIDDEN
.