[ZOOKEEPER-3987] Build failures when running surefire tests concurrently due to bind address already in use

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.7.0
• Fix Version/s: 3.7.0, 3.8.0
• Component/s: tests
• Labels: pull-request-available
• Assignee: Christopher Tubbs
• Reporter: Christopher Tubbs
• Created: 01/Nov/20 19:00
• Updated: 28/Mar/21 08:54
• Resolved: 17/Feb/21 17:15

Description

It seems that many of the surefire/JUnit tests reuse the same bind address. This causes many tests to fail due to a bind address being already in use.

A recent run (using -Dsurefire-forkcount=2 shows several tests that failed for this reason. These same tests do not fail when forkcount is 1 (and they don't always fail when the forkcount is higher... just sometimes, depending on timing):

[ERROR] Errors: 
[ERROR]   SaslAuthTest.testZKOperationsAfterClientSaslAuthFailure:172 » Timeout Failed t...
[ERROR]   SnapshotDigestTest.testSnapshotDigest:86->reloadSnapshotAndCheckDigest:190 » Runtime
[ERROR]   CloseSessionTxnTest.testCloseSessionTxnCompatile:44->testCloseSessionWithDifferentConfig:87->QuorumPeerTestBase.waitForOne:536 » Runtime
[ERROR]   QuorumPeerMainTest.testFailedTxnAsPartOfQuorumLoss:784->QuorumPeerTestBase.waitForOne:536 » Runtime
[ERROR]   ChrootAsyncTest.setUp:38->AsyncOpsTest.setUp:47->ClientBase.setUp:478->ClientBase.setUpWithServerId:503->ClientBase.startServer:519->ClientBase.createNewServerInstance:429 » Bind
[ERROR]   GetChildren2Test.setUp:42->ClientBase.setUp:478->ClientBase.setUpWithServerId:503->ClientBase.startServer:519->ClientBase.createNewServerInstance:429 » Bind
[ERROR]   SaslKerberosAuthOverSSLTest.testAuth » Bind Address already in use
[ERROR]   UnsupportedAddWatcherTest.setUp:105->ClientBase.setUp:478->ClientBase.setUpWithServerId:503->ClientBase.startServer:519->ClientBase.createNewServerInstance:429 » Bind
[INFO] 
[ERROR] Tests run: 2846, Failures: 2, Errors: 8, Skipped: 4

Issue Links

• links to: GitHub Pull Request #1606