shubham19may opened a new pull request, #14498: URL: https://github.com/apache/iceberg/pull/14498
Fixes: #14430 & #14046 #### Description This PR fixed two critical issues preventing vectorized Arrow Reads in iceberg when: 1. Reading Parquet files with `TIMESTAMP_MILLIS` encoding. 2. Using shaded Arrow classes in runtime JARs (e.g., `iceberg-spark-runtime`) Error: ```bash java.lang.ClassCastException: class org.apache.iceberg.shaded.org.apache.arrow.vector.TimeStampMicroTZVector cannot be cast to class org.apache.iceberg.shaded.org.apache.arrow.vector.BigIntVector (org.apache.iceberg.shaded.org.apache.arrow.vector.TimeStampMicroTZVector and org.apache.iceberg.shaded.org.apache.arrow.vector.BigIntVector are in unnamed module of loader 'app') ``` Cause: when reading parquet files with `TIMESTAMP_MILLIS` logical type annotation, the **VectorizedArrowReader** incorrectly allocated a *TimeStampMicroTZVector* based on iceberg schema (which expects microsecond precision), but the actual reader (*TimeStampMicroTZVector*) writes raw long values that require a BigIntVector. Fix: * Explicitly create a **`Field`** with **`ArrowType.Int(Long.SIZE, true)`** type * Allocate a **`BigIntVector`** that matches what *TimestampMillisReader* expects and return **`ReadType.TIMESTAMP_MILLIS`** to trigger millisecond-to-microsecond conversion Also, setting **`arrow.memory.allocation.manager.type=Netty`**, if already not set. ### Testing 1. Creating a Parquet file with **`spark.sql.parquet.outputTimestampType=TIMESTAMP_MILLIS`** 2. Sending the parquet file path to the Iceberg Table to consider it as a data file (an `AddFile()` API) 3. Querying with **`spark.sql.iceberg.vectorization.enabled=true`** 4. Confirming successful vectorized reads with correct timestamp values -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
