Release 350-e LTS (3 Feb 2021)#

Starburst Enterprise platform (SEP) 350-e is the follow up release to the 348-e version. It includes all improvements from the following Presto project releases:

This release is a long term support (LTS) release.

Starburst Enterprise 350-e is the followup release to the 348-e STS release and the 345-e LTS release:

The 350-e release includes all improvements from the following Presto project releases:

This release is a long term support (LTS) release and contains all improvements from the Starburst Enterprise releases since 345-e LTS:

Highlights since 345-e#

  • Support for EXCEPT ALL and INTERSECT ALL

  • Support for RANGE and GROUPS in WINDOW frames

  • Cost-based optimizations for IN <subquery>

  • Improve performance of queries with uncorrelated IN clauses

  • Significant improvements of the Redshift connector

    • Aggregate pushdown

    • Dynamic filtering

    • Table and column statistics

    • Type mapping improvements

    • Support for Amazon’s Redshift JDBC driver 2.0

    • Security enhancements

  • Expanded feature set of the SAP HANA connector

    • Aggregate pushdown

    • Time column predicate pushdown

    • Dynamic filtering

    • Table and column statistics

    • Fixes for calculation views

  • Support for dynamic filtering in the generic JDBC connector

  • Delta Lake support for access control with Ranger policies

  • Add support for OAuth2 authentication for Web UI

  • Improved support for picosecond precision in time and timestamp data types

Breaking changes since 345-e#

  • Require a minimum Java version of 11.0.7 to mitigate Java issue JDK-8206955.

  • Redshift connector changed to use the Redshift JDBC driver, users need to update the connection-url in all catalog files

  • Use consistent NaN behavior for least(), greatest(), array_min(), array_max(), min(), max(), min_by(), and max_by(). NaN is only returned when it is the only value, except for null which are ignored for aggregation functions.

  • Remove configuration properties arrayagg.implementation, multimapagg.implementation, and histogram.implementation.

  • Removed legacy Kubernetes operator, replaced by the Helm-based Kubernetes support since 338-e

350-e initial changes#

The following changes from 350-e.0 and 350-e.1 are all part of the first public release.

General#

  • Add Mission Control web application for cluster and data source management

  • Add Insights web application for cluster and query reporting

  • Add cache service for table scan redirections

Performance#

  • Improve performance of queries using the rank window function

  • Improve performance of certain queries using IS NOT DISTINCT FROM as a join condition, when dynamic filtering is enabled”

Security#

  • Support GRANT OPTION use with Sentry privileges

Hive connector#

  • Support reading nanosecond precision timestamp values in Hive map, array, and struct (row)

Kafka connector#

  • Add encoder and decoder for protobuf messages

  • Add support for OAuth 2.0 authentication using client credentials

  • Support TLS/SSL secured access to the schema registry

Redshift connector#

  • Add support for table statistics

  • Add user impersonation support

  • Update to use Redshift JDBC driver

SAP HANA connector#

  • Extend support for views, including SAP HANA calculation views

Starburst Stargate#

Synapse connector#

350-e.2 changes (18 Feb 2021)#

  • Fix possible S3 connection leak on query failure

  • Disable Hive metastore caching on workers

  • Support non-lowercase partition columns in the Delta Lake connector

  • Fix for Map fields cannot be repeated from Protobuf reader in the Kafka connector

350-e.3 changes (26 Feb 2021)#

  • Add Synapse connector to installation binaries

  • Fix serialization for timestamps with precision higher than three in the MySQL, SQL Server, and SingleStore (MemSQL) connector

  • Fix problems with reading data from Kafka schema registry table

350-e.5 changes (12 Mar 2021)#

  • Fix possible S3 connection leaks when using AVRO table format

  • Fix potential compiler failure when constructing an array with more than 128 elements

  • Fix CREATE TABLE AS with NULL partition keys in Delta Lake

  • Allow configuring process environment variables, including secrets, through /etc/presto/env.sh with RPM-based installation

  • Fix removal of expired cached tables

  • Fix possible S3 connection leaks when using AVRO table format.

350-e.6 changes (19 Mar 2021)#

  • Hive connector: Respect partition filter for DELETE of ACID tables. Previously, the partition filter was ignored, resulting in the deletion of too many rows.

  • Add -XX:PerMethodRecompilationCutoff and -XX:PerBytecodeRecompilationCutoff as JVM safety configuration in RPM

  • Fix failure when subqueries contain EXCEPT or INTERSECT

  • Fix incorrect results when multiple similar subqueries contain UNION clauses that differ only in the ALL vs DISTINCT qualifier

350-e.7 changes (8 Apr 2021)#

  • Fix privilege checks in Ranger global access control for queries that do not read any columns, such as SELECT count(*) FROM <table>.

  • Fix incorrect results when querying the $file_modified_time hidden column in Hive connector.

  • Fix incorrect “SHOW SCHEMA” results in Hive access control with the Privacera platform.

350-e.8 changes (14 May 2021)#

  • Mask value of teradata-parallel-direct.https.keystore.password in log file writing.

  • Fix inconsistent behavior for to_unixtime() with values of type timestamp(p).

  • Change return type of from_unixtime() and from_unixtime_nanos() to timestamp(p) with time zone.

  • Add support for epoch milliseconds exposed as String in the Elasticsearch connector.

  • Support Databricks replace table with schema change in Delta Lake connector.

  • Validate ORC ACID version base on data file metadata as a fallback.

350-e.9 was skipped.

350-e.10 changes (14 Jun 2021)#

  • Correct how delete markers are interpreted for ORC ACID tables in Hive connector, so SEP is not filtering out extraneous rows for certain tables.

  • Fix query failure when reading from a non-ORC insert-only transactional table in Hive connector.

  • Fix incorrect results when accessing BigQuery records with wrong index in BigQuery connector.

  • Fix query failures when accessing the information_schema table in Delta Lake connector.

350-e.11 changes (23 Jun 2021)#

  • Delta Lake connector: Fix query failure after table location changed in the metastore.

350-e.12 changes (18 Aug 2021)#

  • Fix LIKE clause usage with char values.

  • Reject CTAS with skip_header_line_count > 0 in Hive connector.

  • Fix data corruption when performing UPDATE or INSERT on ORC ACID transactional table.

  • Fix column ordering bug in Delta Lake CTAS.

  • Fix DROP SCHEMA and DROP SCHEMA RESTRICT not to drop then schema if it is not empty in MySQL connector.

  • Fix incorrect pushdown of range predicates on char and varchar columns for JDBC connectors where the remote database is case-insensitive or follows a different sort ordering for letters from Trino (MySQL, SingleStore, SQL Server, PostgreSQL, Greenplum).

  • Properly handle number values with negative scale in Teradata.

350-e.13 was skipped.

350-e.14 changes (7 Sep 2021)#

  • Fix invalid result when two decimals are added together. This happened in certain queries where decimals had different precision.

  • Fix incorrect result when using the BigQuery TIME type.

  • Fix incorrect results for queries with a comparison between a VARCHAR column and a CHAR constant.

  • Fix query failure when inserting data into a Hive ACID table that is not explicitly bucketed.

350-e.15 changes (22 Sep 2021)#

  • Fix incorrect results of BigQuery parameterized NUMERIC type.

  • Fix security vulnerability in Okta authentication.

350-e.16 changes (13 Oct 2021)#

  • Fix QueryResult returns empty results for retries.

350-e.17 changes (10 Nov 2021)#

  • Fix incorrect results for queries with nested joins and IS NOT DISTINCT join clauses.

  • Re-enable aggregation pushdown for Redshift tables with diststyle AUTO(ALL)/ALL..

  • Fix incorrect DST result from Redshift TIMESTAMP type.

350-e.18 changes (17 Dec 2021)#

  • Fix deletion without predicate from a transactional table that has original uncompacted files.

  • Fix failure when a schema is dropped during listing tables from BigQuery.

  • Fix connection leak for INSERT, UPDATE, and DELETE operations in the Delta Lake connector.

  • Fix predicate pushdown with expressions using REAL types for MySQL, SingleStore and ClickHouse connectors.

  • Fix treatment of date values before 1582-10-14 in the Greenplum connector.

  • Remove unused log4j dependency from SEP to avoid false alarms about

    Log4Shell.

  • Upgrade log4j in Ranger Admin to version 2.16.0 to mitigate Log4Shell.

  • Disable support for ALTER VIEW ... SET AUTHORIZATION statement.