Change8

v1.32.4

📦 weaviateView on GitHub →
4 features🐛 7 fixes🔧 6 symbols

Summary

This patch release focuses on bug fixes and minor enhancements, notably updating the default model for text2vec-google and improving stability around database operations and compaction.

✨ New Features

  • Backported BlockMax And functionality.
  • Added support for setting dimensions in the text2vec-google module.
  • Added support for backup aliases as part of collection backup operations.
  • Added support for the debug parameter in grouped generative search.

🐛 Bug Fixes

  • Default model for text2vec-google changed to gemini-embedding-001.
  • Database delete tenant now also deletes inactive tenants.
  • Fixed parsing of generative grouped responses.
  • Avoided force loading lazy loaded shards on GetStatus() calls.
  • Delayed compactions until after WAL recovery.
  • Fixed loading of broken compaction segments with metadata in filenames.
  • Moved WAL recovery back to the bucket.

🔧 Affected Symbols

text2vec-googledb delete tenantLazyShard:GetStatuscompactionwal recoverygenerative grouped search