[
https://issues.apache.org/jira/browse/HIVE-19673?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16511375#comment-16511375
]
Zoltan Haindrich commented on HIVE-19673:
-----------------------------------------
there seems to be some inconsistency caused by the fact that the next q file
"starts" with the previous qfile's session; this thing keeps hitting
back....QTestUtil has a very sophisticated operation...it even prevented
setting proper per case timeouts...
> qtest: parquet_ctas.q breaks sample6.q
> --------------------------------------
>
> Key: HIVE-19673
> URL: https://issues.apache.org/jira/browse/HIVE-19673
> Project: Hive
> Issue Type: Bug
> Components: Tests
> Reporter: Zoltan Haindrich
> Assignee: Zoltan Haindrich
> Priority: Major
>
> there is a strange diff happening currently from time to time...sample6 is
> failed
> running it on its own is fine...however when parquet_ctas.q runs before it;
> the *result set* is different
> {code}
> time mvn install -pl itests/qtest -DskipSparkTests -Pitests
> -Dtest=TestCliDriver -Dqfile=parquet_ctas.q,sample6.q -Dtest.output.overwrite
> {code}
> note: sample6.q is run also via the spark driver; and the intresting is that
> the "fluctuating" new resultset matches with the spark driver's output
> {code}
> diff -Naur ./ql/src/test/results/clientpositive/sample6.q.out
> ./ql/src/test/results/clientpositive/spark/sample6.q.out | grep val_
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
