[ https://issues.apache.org/jira/browse/FLINK-17819?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17112721#comment-17112721 ]
Yang Wang commented on FLINK-17819: ----------------------------------- Just a side input, we could not just verify the HADOOP_CLASSPATH environment. The {{flink-shaded-hadoop}} are widely used in many companies, which just because the Flink job deployer is not always running on a machine with hadoop installed. > Yarn error unhelpful when forgetting HADOOP_CLASSPATH > ----------------------------------------------------- > > Key: FLINK-17819 > URL: https://issues.apache.org/jira/browse/FLINK-17819 > Project: Flink > Issue Type: Improvement > Components: Deployment / YARN > Reporter: Arvid Heise > Assignee: Kostas Kloudas > Priority: Critical > Labels: usability > Fix For: 1.11.0 > > > When running > {code:bash} > flink run -m yarn-cluster -yjm 1768 -ytm 50072 -ys 32 ... > {code} > without some export HADOOP_CLASSPATH, we get the unhelpful message > {noformat} > Could not build the program from JAR file: JAR file does not exist: -yjm > {noformat} > I'd expect something like > {noformat} > yarn-cluster can only be used with exported HADOOP_CLASSPATH, see <url> for > more information{noformat} > > I suggest to load a stub for YarnCluster deployment if the actual > implementation fails to load, which prints this error when used. -- This message was sent by Atlassian Jira (v8.3.4#803005)