Description
This is a chicken-and-egg problem between thread leak detection and failure detection rule. I have an idea how to fix this. It'd prevent cascading failures resulting from previous zombie thread errors, like this one:
http://builds.flonkings.com/job/Lucene-trunk-Linux-Java7-64-test-only/81259/consoleText
[junit4] Tests with failures: [junit4] - org.apache.lucene.codecs.lucene45.TestLucene45DocValuesFormat.testSortedSetVariableLengthVsUninvertedField [junit4] - org.apache.lucene.codecs.lucene45.TestLucene45DocValuesFormat (suite) [junit4] - org.apache.lucene.util.junitcompat.TestFailOnFieldCacheInsanity.testFailOnFieldCacheInsanity [junit4] - org.apache.lucene.util.junitcompat.TestSystemPropertiesInvariantRule.testRuleInvariantBeforeClass [junit4] - org.apache.lucene.util.junitcompat.TestSystemPropertiesInvariantRule.testRuleInvariantAfterClass [junit4] - org.apache.lucene.util.junitcompat.TestSystemPropertiesInvariantRule.testRuleInvariantInTestMethod [junit4] - org.apache.lucene.util.junitcompat.TestSystemPropertiesInvariantRule.testNonStringProperties [junit4] - org.apache.lucene.util.junitcompat.TestLeaveFilesIfTestFails.testLeaveFilesIfTestFails