Release 381 (16 May 2022)#

General#

  • Add support for fault-tolerant execution with exchange spooling on Azure Blob Storage. (#12211)

  • Add experimental support for Table functions. (#1839)

  • Increase the default number of stages allowed for a query from 100 to 150, specified with query.max-stage-count. (#12292)

  • Allow configuring the number of partitions for distributed joins and aggregations when task-based fault-tolerant execution is enabled. This can be set with the fault-tolerant-execution-partition-count configuration property or the fault_tolerant_execution_partition_count session property. (#12263)

  • Introduce the least-waste low memory task killer policy. This policy avoids killing tasks that are already executing for a long time, so the amount of wasted work is minimized. It can be enabled with the task.low-memory-killer.policy configuration property. (#12393)

  • Fix potential planning failure of queries with multiple subqueries. (#12199)

Security#

  • Add support for automatic discovery of OpenID Connect metadata with OAuth 2.0 authentication. (#9788)

  • Re-introduce ldap.ssl-trust-certificate as legacy configuration to avoid failures when updating Trino version. (#12187)

  • Fix potential query failure when a table has multiple column masks defined. (#12262)

  • Fix incorrect masking of columns when multiple rules in file-based system and connector access controls match. (#12203)

  • Fix authentication failure when using the LDAP password authenticator with ActiveDirectory. (#12321)

Web UI#

  • Ensure consistent sort order in the list of workers. (#12290)

Docker image#

  • Improve Advanced Encryption Standard (AES) processing performance on ARM64 processors. This is used for operations such as accessing object storage systems via TLS/SSL. (#12251)

CLI#

  • Add automatic suggestions from command history. This can be disabled with the --disable-auto-suggestion option. (#11671)

BigQuery connector#

  • Support reading materialized views. (#12352)

  • Allow skipping view materialization via bigquery.skip-view-materialization configuration property. (#12210)

  • Support reading snapshot tables. (#12380)

ClickHouse connector#

Druid connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

Elasticsearch connector#

  • Improve query performance by simplifying filters sent to Elasticsearch. (#10717)

  • Fix failure when reading nested timestamp values that are not ISO 8601 formatted. (#12250)

Hive connector#

  • Fix query failure when the table and partition bucket counts do not match. (#11885)

Iceberg connector#

MariaDB connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

MySQL connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

Oracle connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

PostgreSQL connector#

  • Prevent data loss when non-transactional insert fails. (#12225)

Redis connector#

  • Allow specifying the refresh interval for fetching the table description with the redis.table-description-cache-ttl configuration property. (#12240)

  • Support setting username for the connection with the redis.user configuration property. (#12279)

Redshift connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

SingleStore (MemSQL) connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

SQL Server connector#

  • Prevent data loss when non-transactional insert fails. (#12229)

SPI#

  • Remove deprecated ConnectorMetadata methods without the retry mode parameter. (#12342)