[SPARK-27032] Flaky test: org.apache.spark.sql.execution.streaming.HDFSMetadataLogSuite.HDFSMetadataLog: metadata directory collision

• Type: Test
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 3.0.0
• Fix Version/s: 3.0.0
• Component/s: Spark Core, Tests
• Target Version/s: 3.0.0
• Assignee: Sean R. Owen
• Reporter: Sean R. Owen
• Created: 02/Mar/19 14:49
• Updated: 12/Dec/22 18:10
• Resolved: 04/Mar/19 04:38

Description

Locally and on Jenkins, I've frequently seen this test fail:

Error Message
The await method on Waiter timed out.
Stacktrace
      org.scalatest.exceptions.TestFailedException: The await method on Waiter timed out.
      at org.scalatest.concurrent.Waiters$Waiter.awaitImpl(Waiters.scala:406)
      at org.scalatest.concurrent.Waiters$Waiter.await(Waiters.scala:540)
      at org.apache.spark.sql.execution.streaming.HDFSMetadataLogSuite.$anonfun$new$19(HDFSMetadataLogSuite.scala:158)
      at org.apache.spark.sql.execution.streaming.HDFSMetadataLogSuite.$anonfun$new$19$adapted(HDFSMetadataLogSuite.scala:133)
...

See for example https://amplab.cs.berkeley.edu/jenkins/view/Spark%20QA%20Test%20(Dashboard)/job/spark-master-test-maven-hadoop-2.7/6057/testReport/

There aren't obvious errors or problems with the test. Because it passes sometimes, my guess is that the timeout is simply too short or the test too long. I'd like to try reducing the number of threads/batches in the test.

Issue Links

• links to: GitHub Pull Request #23937