Skip to content

[Parquet] pq.write_table fails with OSError: Lz4 compression failure. #50613

Description

@skaae

Describe the bug, including details regarding any error messages, version, and platform.

I encountered a table with where pq.write_table throws compression error:

pq.write_table(table, "foo", compression='lz4')
*** OSError: Lz4 compression failure.

using compression='gzip' works.

The minimal table I was able to find contains 42 rows with a single columns of strings. Each string is 30-60 mb.
The strings in the table are xml and can be compressed to ~100mb in total.

I can share the table, but would like to not do it publicly.

Python 3.12.3
pyarrow 23.0.1
lz4 4.4.5
ubuntu 24.04

Component(s)

Parquet

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions