10.21.2021 Release Notes

Here are the release notes for October 21, 2021.

🚀 FEATURES🚀

  • Add a separate component state of waiting “Cluster Capacity” to indicate when component is waiting to process due to insufficient cluster capacity, instead of keeping components in “Ready” state
  • Site Admins can view the Spark UI for the Spark cluster that powers the ‘Query’ panel in the Ascend interface

ENHANCEMENTS

  • [Snowflake Write Connector] Improve speed in cases where the cluster has capacity to process additional Ascend partitions in parallel
  • [Snowflake Write Connector] If the desired schema doesn’t exist in the database and the ‘alter’ option is chosen, automatically create the Snowflake schema
  • Improve performance of schema inference for Redshift and Snowflake write connections for large Dataframes
  • Unblock writing large strings into Redshift by coercing string type in Redshift to varchar(65535) instead of the Spark default varchar(256)
  • Add automatic schema management options for the new Redshift Write Connection
  • Improve clarity of running partition count while a component is running

🔧 BUGFIXES🔧

  • Ensure that Write Connector tasks are always failed if any of the write subtasks fail
  • Fix edge case in checking for schema matches in Snowflake and Redshift connections
  • Fix MS SQL Test Connection bug where Access check on Alter Table fails incorrectly
  • Fix edge case where saving legacy Write Connectors would fail
  • Fix edge case where details fields don’t load while editing a connector
  • Handle edge cases in S3 Test Read Connection where an exception arises during object read, hiding the actual failure during “Test Connection”
  • Fix “Text” data source for Spark jobs running in clustered mode
  • Fix MS SQL statistics collection for tables larger than 2B records by using COUNT_BIG instead of COUNT
  • Fixed case where not all components showed up as available as inputs in Additional Input Config for transforms
  • Fix errors with Object Aggregation where there’s an empty directory at the root of the aggregation
  • Reduce cases of Failed to process task: exceeded task submission deadline by increasing timeout
© Ascension Labs Inc. | All Rights Reserved