[
https://issues.apache.org/jira/browse/FLINK-23449?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17392680#comment-17392680
]
Kai Chen edited comment on FLINK-23449 at 8/4/21, 3:10 AM:
-----------------------------------------------------------
[~gaborgsomogyi] [~trohrmann] I'm not stick to code change. It was just a
little uncompatible and confusing when upgrading from flink-1.10 to upper
versions when I followed official guide and used flink hive connector.
Flink-1.10 still uses flink-shaded-hadoop-2-uber.jar which includes
hadoop-mapreduce-client-core.jar. And starting from Flink 1.11, using
flink-shaded-hadoop-2-uber releases is not officially supported by the Flink
project and thus hadoop-mapreduce-client-core.jar is not included in JM/TM
classpath.
In the other hand, users follow [the
documentation|https://ci.apache.org/projects/flink/flink-docs-release-1.13/docs/deployment/resource-providers/yarn]
, set up HADOOP_CLASSPATH and usually take it for granted that
HADOOP_CLASSPATH is added to JM/TM classpath. That's why I changed the code and
added MAPREDUCE_APPLICATION_CLASSPATH.
As I said, I'm not stick to code change. An updating/extending documentation
looks good to me, too.
was (Author: yuchuanchen):
[~gaborgsomogyi] [~trohrmann] I'm not stick to code change. It was just a
little uncompatible and confusing when upgrading from flink-1.10 to upper
versions when I followed official guide and used flink hive connector.
Flink-1.10 still uses flink-shaded-hadoop-2-uber.jar which includes
hadoop-mapreduce-client-core.jar. And starting from Flink 1.11, using
flink-shaded-hadoop-2-uber releases is not officially supported by the Flink
project and thus hadoop-mapreduce-client-core.jar is not included in JM/TM
classpath.
In the other hand, users follow [the
documentation|https://ci.apache.org/projects/flink/flink-docs-release-1.13/docs/deployment/resource-providers/yarn]
, set up HADOOP_CLASSPATH and usually take it for granted that
HADOOP_CLASSPATH is added to JM/TM classpath. That's why I changed the code and
added MAPREDUCE_APPLICATION_CLASSPATH.
As I said, I'm not stick to code change. A better documentation looks good to
me, too.
> YarnTaskExecutorRunner does not contains MapReduce classes
> -----------------------------------------------------------
>
> Key: FLINK-23449
> URL: https://issues.apache.org/jira/browse/FLINK-23449
> Project: Flink
> Issue Type: Bug
> Components: Connectors / Hive, Deployment / YARN
> Affects Versions: 1.11.3
> Environment: flink-1.11
> flink on yarn cluster
> jdk1.8
> hive1.2.1
> hadoop2.7
> hadoop classes is provided with {{export HADOOP_CLASSPATH=`hadoop classpath`
> when submitting test APP. （described in
> [https://ci.apache.org/projects/flink/flink-docs-release-1.11/ops/deployment/hadoop.html]
> ）}}
> {{}}
> Reporter: Kai Chen
> Priority: Major
> Labels: pull-request-available
>
> I followed instructions described in
> [https://ci.apache.org/projects/flink/flink-docs-release-1.11/dev/table/hive]
> and tested hive streaming sink, met this exception
> Caused by: java.lang.ClassNotFoundException:
> org.apache.hadoop.mapred.JobConf
> [http://apache-flink.147419.n8.nabble.com/Flink-td7866.html] met the same
> problem.
>
> I checked TM jvm envs and the code and found that flink only set up
> YARN_APPLICATION_CLASSPATH, but without MAPREDUCE_APPLICATION_CLASSPATH.
> See:
> [https://github.com/apache/flink/blob/ed39fb2efc790af038c1babd4a48847b7b39f91e/flink-yarn/src/main/java/org/apache/flink/yarn/Utils.java#L119]
>
> I think we should add MAPREDUCE_APPLICATION_CLASSPATH as well, as the same as
> spark does.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
