Skip to content
This repository was archived by the owner on Jan 2, 2023. It is now read-only.

v0.2.0

Choose a tag to compare

@iconara iconara released this 14 Apr 06:51
· 44 commits to master since this release

This release includes many improvements and updates to the documentation, including:

  • You can now set the maximum bad records to allow when loading the BigQuery table: --max-bad-records
  • Specifying the BigQuery table ID is now optional, it will default to the same name as the Redshift table
  • The transfer uses Redshift's unload manifest to determine which files to transfer, instead of relying on globs
  • The unloaded and transferred files will be removed from S3 and Cloud Storage after the BigQuery table has been loaded
  • The data will be compressed to minimize transfer costs
  • The S3 and Cloud Storage locations specified are now required to be empty or the transfer will fail
  • Logging has been improved, for example the number of files and their sizes are logged before the transfer starts