[Bug] SparkE2eTest fails on trailing Spark shutdown ERROR logs other than CoarseGrainedScheduler

[thswlsqls]

Search before asking

• I searched in the issues and found nothing similar.

Paimon version
master @ 1d368b4 (2.2-SNAPSHOT)

Compute Engine
Spark (paimon-e2e-tests)

Minimal reproduce step
Run the e2e-flink1 CI lane; SparkE2eTest.testFlinkWriteAndSparkRead fails intermittently in E2eReaderTestBase.checkQueryResult because the spark-sql shutdown races with RPC dispatch and logs an ERROR to stdout after the correct result rows. SparkE2eTest.stripCoarseGrainedSchedulerShutdownError() (#8907) only cuts the ERROR Utils: Uncaught exception in thread dispatcher-CoarseGrainedScheduler message, but the race also logs ERROR TransportRequestHandler: Error while invoking RpcHandler#receive() for one-way message. with a Could not find AppClient stack trace. The exit code is 0, so the tail passes the guard and breaks the isEqualTo comparison.

Observed runs:

• https://github.com/apache/paimon/actions/runs/31138366506 (PR [core] Fix stale file index cleanup #9085)
• https://github.com/apache/paimon/actions/runs/33298854847 (master push)
• https://github.com/apache/paimon/actions/runs/34942967900 (PR [core] Expose Avro header and physical block locations #9835)

What doesn't meet your expectations?
The comparison should ignore any trailing Spark shutdown log, not only the one message #8907 special-cased.

Anything else?
Spark's default log4j2 pattern (%d{yy/MM/dd HH:mm:ss} %p %c{1}: %m%n%ex) gives every ERROR line the same timestamp prefix, so the test can cut at the first such line.

Are you willing to submit a PR?

• I'm willing to submit a PR!