Robbie Gemmell created ARTEMIS-3652:
---------------------------------------
Summary: ElasticQueueTest can leave behind task loops creating
producers and consumers
Key: ARTEMIS-3652
URL: https://issues.apache.org/jira/browse/ARTEMIS-3652
Project: ActiveMQ Artemis
Issue Type: Test
Components: Tests
Reporter: Robbie Gemmell
ElasticQueueTest can leave behind tasks creating connections in a loop and
sending+receiving messages on them, and catching exceptions such as those
caused when trying to stop the task running. The small snippet from a CI log
below shows it happening. The producer task logs its expected-failure itself
and can be seen repeating, but it is also visible from the number of connection
failure logs that there is another loop, for the consumer task which does not
log its failure itself.
{noformat}
expected send failure: javax.jms.JMSException: finishConnect(..) failed:
Connection refused: localhost/127.0.0.1:61616 PID: 10268637, uri:
amqp://localhost:61617
[FailoverProvider: async work thread] 05:39:30,229 ERROR
[org.apache.qpid.jms.provider.failover.FailoverProvider] Failed to connect
after: 1 attempt(s)
[FailoverProvider: async work thread] 05:39:30,229 WARN
[org.apache.qpid.jms.JmsConnection] Connection
ID:b454c4b7-c41d-46d7-8e9e-84183ed9b698:1 has failed due to: finishConnect(..)
failed: Connection refused: localhost/127.0.0.1:61617
[FailoverProvider: async work thread] 05:39:30,230 ERROR
[org.apache.qpid.jms.provider.failover.FailoverProvider] Failed to connect
after: 1 attempt(s)
[FailoverProvider: async work thread] 05:39:30,230 WARN
[org.apache.qpid.jms.JmsConnection] Connection
ID:8604bcff-98b0-4905-80f2-53a0f06516ee:1 has failed due to: finishConnect(..)
failed: Connection refused: localhost/127.0.0.1:61616
expected send failure: javax.jms.JMSException: finishConnect(..) failed:
Connection refused: localhost/127.0.0.1:61616 PID: 10268637, uri:
amqp://localhost:61617
[FailoverProvider: async work thread] 05:39:30,231 ERROR
[org.apache.qpid.jms.provider.failover.FailoverProvider] Failed to connect
after: 1 attempt(s)
[FailoverProvider: async work thread] 05:39:30,231 WARN
[org.apache.qpid.jms.JmsConnection] Connection
ID:67bb156e-3611-4670-96fd-c0739bfba4e3:1 has failed due to: finishConnect(..)
failed: Connection refused: localhost/127.0.0.1:61617
[FailoverProvider: async work thread] 05:39:30,231 ERROR
[org.apache.qpid.jms.provider.failover.FailoverProvider] Failed to connect
after: 1 attempt(s)
[FailoverProvider: async work thread] 05:39:30,231 WARN
[org.apache.qpid.jms.JmsConnection] Connection
ID:9ea1925a-3a4a-473f-889b-302d8c867bfe:1 has failed due to: finishConnect(..)
failed: Connection refused: localhost/127.0.0.1:61617
[FailoverProvider: async work thread] 05:39:30,237 ERROR
[org.apache.qpid.jms.provider.failover.FailoverProvider] Failed to connect
after: 1 attempt(s)
[FailoverProvider: async work thread] 05:39:30,237 WARN
[org.apache.qpid.jms.JmsConnection] Connection
ID:4d529e4f-c2de-4efb-a5f1-a90c80736c98:1 has failed due to: finishConnect(..)
failed: Connection refused: localhost/127.0.0.1:61616
{noformat}
Looking at the test code, this would clearly be caused due to the 'done'
AtomicBoolean not being tripped and the loop just continuing. This would happen
as the tests own assertions caused it to fail before it could do so, as the
summary failures show suggests is what happened:
{noformat}
[ERROR] Failures:
[ERROR] ElasticQueueTest>Assert.fail:89 Thread leaked
[ERROR]
ElasticQueueTest.testScale0_1_CombinedProducerConsumerConnectionWithProducerRole:516->Assert.assertTrue:53->Assert.assertTrue:42->Assert.fail:87
[ERROR]
ElasticQueueTest.testScale0_1_CombinedRoleConnection:615->Assert.assertTrue:53->Assert.assertTrue:42->Assert.fail:87
{noformat}
That failure leaving these tasks behind then probably caused the resulting
connections/production/consumption to also break a number of tests that
followed. A small snippet of those continuing directly from above:
{noformat}
[ERROR]
RedirectTest.testLeastConnectionsRedirect:149->testEvenlyRedirect:215->Assert.assertEquals:647->Assert.failNotEquals:835->Assert.fail:89
Messages of node 2 expected:<1> but was:<0>
[ERROR]
TargetKeyTest.testClientIDKey:121->Assert.assertEquals:633->Assert.assertEquals:647->Assert.failNotEquals:835->Assert.fail:89
expected:<1> but was:<4>
[ERROR]
TargetKeyTest.testClientIDKey:121->Assert.assertEquals:633->Assert.assertEquals:647->Assert.failNotEquals:835->Assert.fail:89
expected:<1> but was:<2>
{noformat}
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
