Native tests failing randomly in CI

[mbutrovich]

Describe the bug

I've noticed this in different CI runs and have not been able to reproduce this locally:

https://github.com/apache/datafusion-comet/actions/runs/18099577786/job/51498760915?pr=2487

Steps to reproduce

Unclear. It shows up in CI occasionally, running locally does not reproduce.

Expected behavior

It shouldn't panic.

Additional context

No response

[andygrove]

Another one:

CometTPCHQuerySuite:
25/09/29 16:53:54 INFO core/src/lib.rs: Comet native library version 0.11.0 initialized
- q1 (8 seconds, 114 milliseconds)
- q2 (6 seconds, 495 milliseconds)
- q3 (4 seconds, 540 milliseconds)
- q4 (2 seconds, 979 milliseconds)
- q5 (7 seconds, 79 milliseconds)
- q6 (1 second, 549 milliseconds)
- q7 (8 seconds, 820 milliseconds)
- q8 (7 seconds, 803 milliseconds)
- q9 (9 seconds, 347 milliseconds)
- q10 (4 seconds, 287 milliseconds)
- q11 (4 seconds, 183 milliseconds)
- q12 (2 seconds, 920 milliseconds)
#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGILL (0x4) at pc=0x00007f55dd822db4, pid=1203, tid=1456
#
# JRE version: OpenJDK Runtime Environment Zulu11.82+19-CA (11.0.28+6) (build 11.0.28+6-LTS)
# Java VM: OpenJDK 64-Bit Server VM Zulu11.82+19-CA (11.0.28+6-LTS, mixed mode, tiered, compressed oops, g1 gc, linux-amd64)
# Problematic frame:
# C  [libcomet-14465776396313664081.so+0x4022db4]  regex_automata::meta::regex::Builder::configure::hfaf7dcc163427968+0x114
#2482 

[andygrove]

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGILL (0x4) at pc=0x00007fd69ca1d208, pid=1208, tid=1536
#
# JRE version: OpenJDK Runtime Environment Zulu11.82+19-CA (11.0.28+6) (build 11.0.28+6-LTS)
# Java VM: OpenJDK 64-Bit Server VM Zulu11.82+19-CA (11.0.28+6-LTS, mixed mode, tiered, compressed oops, g1 gc, linux-amd64)
# Problematic frame:
# C  [libcomet-6400017282685449905.so+0x361d208]  parquet::util::bit_pack::unpack16::h027291c077fce3b2+0x418

[andygrove]

The issue first happened on the main branch with the highlighted PR. The issue is intermittent so the issue was not necessary introduced in the highlighted PR but perhaps an earlier one.

edit: It is also possible that something changed in the environment

[andygrove]

Another variation:

[INFO] -------------------------------------------------------
[INFO]  T E S T S
[INFO] -------------------------------------------------------
[INFO] Running org.apache.comet.parquet.TestColumnReader
Aborted (core dumped)

[wForget]

Spark SQL flaky test: FileStreamSinkSuite."cleanup complete but invalid output for aborted job", created a ticket https://issues.apache.org/jira/browse/SPARK-53756

[info] - cleanup complete but invalid output for aborted job *** FAILED *** (430 milliseconds)
[info]   java.io.UncheckedIOException: java.nio.file.NoSuchFileException: /__w/datafusion-comet/datafusion-comet/apache-spark/target/tmp/spark-2cf998bb-fd3c-4621-88d8-05e9decad882/output @#output/.part-00009-cb134e84-5d4a-42d2-a342-b5edc52776ce-c000.snappy.parquet.crc
[info]   at java.base/java.nio.file.FileTreeIterator.fetchNextIfNeeded(FileTreeIterator.java:87)