Ian Hansen
05/19/2026, 2:30 PMkapa.ai
05/19/2026, 2:30 PMInput Glue table is not an iceberg table: <your table name>.
Fix: Drop the existing Glue table. Note that dropping tables from the console may not immediately delete them — you may need to use the AWS API to drop all table versions and wait for background processing to complete. [S3 Data Lake Setup]
2. Missing Glue Permissions If the IAM policy attached to your Airbyte user/role doesn't include the required Glue permissions, the connector may fall back to writing raw Parquet files without properly registering the Iceberg table metadata in Glue. Make sure your S3 policy includes all the required Glue actions, such as glue:CreateTable, glue:GetTable, glue:UpdateTable, etc. [S3 Data Lake Setup]
3. AWS Lake Formation (if applicable) If you're using AWS Lake Formation, you must also grant:
• Data location_access on the S3 path
• Create database on the catalog (if you want Airbyte to create databases)
• Create table, Describe on specific databases (if managing databases manually) [S3 Data Lake Setup]
Recommended Steps 1. Drop the existing Glue table (and verify it's fully deleted via the AWS API).
2. Verify your IAM policy includes all required Glue permissions.
3. Set the warehouse location to s3://<bucket name>/path/within/bucket.
4. Re-run the sync — Airbyte should now create a proper Iceberg table in Glue. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.