ContextQMD
Back to Presto

Release 0.252

presto-docs/src/main/sphinx/release/release-0.252.rst

0.2972.7 KB
Original Source

============= Release 0.252

Highlights

  • Add support for returning partial results.
  • Add automatic query retry functionality for transient failures.
  • Add support for running base32 encoded ElasticSearch queries.

Details

General Changes

  • Fix the join operator to prevent hanging when spilling is enabled and the probe side finishes before the hash builder starts.
  • Add new spilling strategy to spill when a combination of user, system and recovable memory on a node exceeds max_total_memory_per_node. This can be enabled by setting the system configuration experimental.spiller.task-spilling-strategy to PER_QUERY_MEMORY_LIMIT.
  • Add support for logging formatted prepared query. This can be enabled by setting session property log_formatted_query_enabled to true.
  • Add support for returning partial results for the queries by setting partial_results_enabled session property. Additionally partial_results_max_execution_time_multiplier, partial_results_completion_ratio_threshold session properties can be set to configure the max execution time multiplier and minimum completion ratio threshold for the queries.
  • Add automatic query retry functionality for transient failures. This can be enabled by setting per-query-retry-limit to a non-zero integer to indicate the per query retry count.
  • Add support to coordinator endpoint /v1/info/state to return ACTIVE when the coordinator is not shutting down and the cluster has the minimum required workers.
  • Add functions :func:!chisquared_cdf and :func:!inverse_chisquared_cdf.

Security Changes

  • Add file based password authenticator plugin. See :doc:/security/password-file.

Hive Changes

  • Update Alluxio cache config property cache.alluxio.timeout-enabled to be true by default.

JDBC Changes

  • Add support to serialize PrestoArray to a string.

Elasticsearch Connector

  • Add support for running base32 encoded ElasticSearch queries. See :doc:/connector/elasticsearch.

Presto On Spark Changes

  • Improve bucketed table write parallelism.
  • Fix a bug when Presto on Spark doesn't start because the temporary storage is not initialized.

Contributors

Ahmad Ghazal, Andrii Rosa, Arunachalam Thirupathi, Bhavani Hari, Bin Fan, Chen Li, Chi Tsai, Costin V Cozianu, Dongliang Chen, James Petty, James Sun, Junyi Huang, Ke Wang, Mayank Garg, Naveen Kumar Mahadevuni, Nikhil Collooru, Rebecca Schlussel, Rohit Jain, Rongrong Zhong, Shixuan Fan, Sreeni Viswanadha, Tal Galili, Tim Meehan, Vic Zhang, Zhenxiao Luo, imjalpreet, sophiashang, tanjialiang, v-jizhang