Skip to content

partition condition not taken into account when partition is a date #202

Open
@cjaber

Description

@cjaber

Description

When querying a Delta table stored in S3 with a WHERE clause on a partition column, DuckDB appears to access files from other partitions that do not match the filter condition.

Query used

SELECT count(*)
FROM delta_scan('s3://<your-bucket-name>/tables/silver/usage_record/')
WHERE partition_ref_date = '2025-05-24';

Expected Behavior
Only files in the partition:
s3:///tables/silver/usage_record/partition_ref_date=2025-05-24/
should be accessed.

Observed Behavior
DuckDB attempts to read files from other partitions, such as:

s3:///tables/silver/usage_record/partition_ref_date=2024-12-22/part-00030-...parquet
Which results in an error:

IO Error:
Could not establish connection error for HTTP HEAD to 'https://<your-bucket-name>.s3.amazonaws.com/tables/silver/usage_record/partition_ref_date%3D2024-12-22/part-00030-...parquet'

System Info
DuckDB version: 1.3.0

OS: macOS version 15.5

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions