Itermittent test timeout on PersistentFailoverE2ETest.testSimpleConsumerEventsWithoutPartition #1614
Comments
|
discussed with @merlimat - it is a bit unknown how the flaky test can happen after spending a few days on it. so we are going to defer this to 2.1. |
sijie
pushed a commit
that referenced
this issue
May 21, 2018
### Motivation Fixes #1614 #1617 The flaky test was actually because of a race condition in the dispatcher code. If there is a pending read when the consumer is switched, the actual switch happens after the read completes, but the dispatcher wouldn't trigger the notification at that point. ### Modifications Ensure the active/inactive notification is sent after the read completes.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Expected behavior
The test is timing out several times. eg:
https://builds.apache.org/job/pulsar_precommit_java8/724/testReport/junit/org.apache.pulsar.broker.service/PersistentFailoverE2ETest/testSimpleConsumerEventsWithoutPartition/
Test thread is blocked on:
https://github.com/apache/incubator-pulsar/blob/8fbb926089effcadf59e199fe69ec65b4bbe4eae/pulsar-broker/src/test/java/org/apache/pulsar/broker/service/PersistentFailoverE2ETest.java#L217
In particular it's waiting on the
listener.activeQueue.take().intValue())https://github.com/apache/incubator-pulsar/blob/8fbb926089effcadf59e199fe69ec65b4bbe4eae/pulsar-broker/src/test/java/org/apache/pulsar/broker/service/PersistentFailoverE2ETest.java#L102-L105
The text was updated successfully, but these errors were encountered: