
When first entering Query Editor v2, you might mistakenly click “Load data” to import data into Redshift native tables.

Load data is for COPY data from S3 into Redshift internal tables, suitable when you want to store data in Redshift.
But our goal is to query external data from S3/Glue without copying.
Initially, you might create external schema pointing to craw_data_catalog (raw data) instead of processed database.
SELECT schemaname, tablename FROM svv_external_tables WHERE schemaname = 'taxi_raw';

Glue table has year/month appearing in both normal columns and partition keys.
External table “taxi_processed.processed_yellow_taxi_trip_data” has column “year” duplicated
ETL job wrote year/month into Parquet data, while S3 path used Hive partitions year=…/month=…
Glue Crawler inferred duplicates.
Edit Glue table schema: remove year/month from normal columns, keep in partition keys.
Then refresh Redshift metadata.
If query fails with access denied, check if IAM role has permissions to read S3 buckets and Glue catalog.
Role needs: AmazonS3ReadOnlyAccess, AWSGlueServiceRole policies.
Ensure all resources are in the same region us-east-2.