Gary Yao created FLINK-14680:
--------------------------------
Summary: Enable KafkaConsumerTestBase#runFailOnNoBrokerTest to pass with new DefaultScheduler
Key: FLINK-14680
URL:
https://issues.apache.org/jira/browse/FLINK-14680 Project: Flink
Issue Type: Sub-task
Components: Runtime / Coordination, Tests
Affects Versions: 1.10.0
Reporter: Gary Yao
Assignee: Gary Yao
Fix For: 1.10.0
{{KafkaConsumerTestBase#runFailOnNoBrokerTest}} has assumptions on the causal chain of the {{JobExecutionException}}. In particular, it assumes that the exception caused by user code is the direct cause of {{JobExecutionException}}. However, this is no longer true when using the {{DefaultScheduler}}, which wraps the exception in an {{JobException}}, which additionally specifies the reason of the job recovery suppression.
The code in question is listed below:
{code:java}
} catch (JobExecutionException jee) {
if (kafkaServer.getVersion().equals("0.9") ||
kafkaServer.getVersion().equals("0.10") ||
kafkaServer.getVersion().equals("0.11") ||
kafkaServer.getVersion().equals("2.0")) {
assertTrue(jee.getCause() instanceof TimeoutException);
TimeoutException te = (TimeoutException) jee.getCause();
assertEquals("Timeout expired while fetching topic metadata", te.getMessage());
} else {
assertTrue(jee.getCause() instanceof RuntimeException);
RuntimeException re = (RuntimeException) jee.getCause();
assertTrue(re.getMessage().contains("Unable to retrieve any partitions"));
}
}
{code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
Locked
