Github user HeartSaVioR commented on the issue:
https://github.com/apache/spark/pull/22138
From the stack trace of test failure, it doesn't look like relevant to the
code change I guess. Jenkins shows REGRESSION but the test is added at Aug 25,
2018 so aged less than 7 days, which is not enough to consider it as regression
or not I think. The test itself could be flaky. The test succeeds 5 times in a
row in local dev. and I'll try to run more.
https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/95474/testReport/org.apache.spark.sql.kafka010/KafkaSourceStressForDontFailOnDataLossSuite/stress_test_for_failOnDataLoss_false/
@zsxwing Do you have any idea on this? I'm curious how checking query
exception after query.stop() is relevant to `failOnDataLoss = false`, given
that we check query exception all the time in loop.
https://github.com/apache/spark/blob/bb3e6ed9216f98f3a3b96c8c52f20042d65e2181/external/kafka-0-10-sql/src/test/scala/org/apache/spark/sql/kafka010/KafkaDontFailOnDataLossSuite.scala#L270-L274
Will retrigger build for now.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
