Apache Iceberg version

1.9.2 (latest release)

Query engine

Spark

Please describe the bug 🐞

🧭 Problem Summary

My ENV is :

• iceberg: 1.7.x ~1.9.x
• spark: 3.5.3
• jdk: openjdk version "17.0.2" 2022-01-18
• linux: 5.12.5-1.el7.elrepo.x86_64

When querying a partitioned Iceberg table (by year, month) with Parquet bloom filter enabled on a STRING column (resource_id), the query returns 0 rows on Linux (Iceberg 1.9.x + Spark 3.5.3) but returns correctly on Windows or when downgrading to Iceberg 1.7.x.

This discrepancy leads to incorrect query results and is platform-dependent.

📦 Table DDL

CREATE TABLE IF NOT EXISTS iceberg_catalog.test.xxh (
date_time TIMESTAMP,
operate_type INT,
resource_id STRING,
year INT,
month INT,
day INT
)
USING iceberg
PARTITIONED BY (year, month)
TBLPROPERTIES (
'write.distribution-mode' = 'hash',
'write.metadata.delete-after-commit.enabled' = 'true',
'write.metadata.previous-versions-max' = '2',
'write.parquet.bloom-filter-enabled.column.resource_id' = 'true',
'write.parquet.compression-codec' = 'zstd',
'write.target-file-size-bytes' = '4294967296'
);

on linux iceberg v1.7.2 return correct result, v1.9.2 can not return correct result

on windows iceberg v1.9.2 can return correct result

spark version > 3.5.3 with iceberg 1.7.1 will get another error

when i use this code it worked well. But i know how to set vectorization-enabled in sql

this is my data file

data.zip

Willingness to contribute

• I can contribute a fix for this bug independently
• I would be willing to contribute a fix for this bug with guidance from the Iceberg community
• I cannot contribute a fix for this bug at this time