I think WORKING_DIR/vdb_chunks.json should also be removed while starting a new run in the same directory. I had issues when querying because a specific chunk ID was not present in kv_chunks_text_units.json, but was there in vdb_chunks.json. Maybe new additions are appended, and old ones are getting referred?
I think
WORKING_DIR/vdb_chunks.jsonshould also be removed while starting a new run in the same directory. I had issues when querying because a specific chunk ID was not present in kv_chunks_text_units.json, but was there in vdb_chunks.json. Maybe new additions are appended, and old ones are getting referred?