Type something to search...
← All notes

Glue Crawler mixes schemas when CSV layouts share a folder

AWS Glue

What happened

I pointed a Glue Crawler at an S3 folder holding two CSV layouts. It registered one table with a mix of both column sets.

What I learned

The Crawler treats one folder as one table. Files with different structures must live in different folders, otherwise the Data Catalog inherits the mess.

The rule I keep

One folder, one schema, one layout. Tidy on the way in, or expensive on the way out.