Smallfu666 commented on PR #5160: URL: https://github.com/apache/datafusion-comet/pull/5160#issuecomment-5150076027
The one CI failure here — `PR Build (macOS) / macos-14/Spark 4.0, JDK 17, Scala 2.13 [scans]` — is #5023, not a failure of this PR. It's a JVM `SIGSEGV at pc=0x0` (no test assertion failed anywhere in the log), and symbolizing the `hs_err` log from that job points at `hdfsThreadDestructor`: it calls `(*env)->GetJavaVM` on a stale `JNIEnv`, reads a null function entry, and branches to address zero. That's the long-standing [HDFS-16021](https://issues.apache.org/jira/browse/HDFS-16021), still open upstream and still present in Hadoop trunk. It only reproduces on macOS because `hdrs` is enabled with `vendored` under `[target.'cfg(target_os = "macos")'.dependencies]`, and `ParquetReadFromFakeHadoopFsSuite` deliberately routes `fake://` through libhdfs (`COMET_LIBHDFS_SCHEMES = "fake,hdfs"`). That path is disjoint from this PR: the changes are three files under `native/spark-expr/` (`sum_decimal.rs`, `avg_decimal.rs`, `utils.rs`), all decimal aggregation, with no scan/parquet/hdfs code touched. The Linux jobs and both Spark SQL test matrices are green. Could a committer re-run that job? Thanks! -- 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]
