[
https://issues.apache.org/jira/browse/FLINK-20237?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17237983#comment-17237983
]
Rui Li commented on FLINK-20237:
--------------------------------
Hi [~dwysakowicz], I'd prefer not to add hive jars to HADOOP_CLASSPATH. Because
HADOOP_CLASSPATH is an env variable recognized by hadoop itself, e.g. in the
[RunJar
util|https://github.com/apache/hadoop/blob/rel/release-2.7.5/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/RunJar.java#L68].
So we'd better not to tamper with it.
The recommended way to add hive dependency is to use a
"flink-sql-connector-hive" uber jar. User just needs to add this one jar, as
well as setting HADOOP_CLASSPATH, and then is ready to integrate to hive.
> Do not recommend putting hive-exec in flink/lib
> -----------------------------------------------
>
> Key: FLINK-20237
> URL: https://issues.apache.org/jira/browse/FLINK-20237
> Project: Flink
> Issue Type: Improvement
> Components: Connectors / Hive, Documentation
> Affects Versions: 1.12.0
> Reporter: Dawid Wysakowicz
> Priority: Critical
> Fix For: 1.12.0
>
>
> The Hive setup page links to Flink's Hadoop setup page which recommends
> passing hadoop dependencies via:
> {code}
> export HADOOP_CLASSPATH=`hadoop classpath`
> {code}
> This should in 99% of cases put the hive-exec dependency on the classpath
> already. I'd recommend removing the recommendation to put {{hive-exec}} into
> \lib separately.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
