Here is the full log: https://pastebin.com/A9KQimSL
The command used is the following:
> ./bin/flink run -m yarn-cluster -yn 4 -yjm 1024 -ytm 4096 ./examples/batch/WordCount.jar
It is the demo command described here: https://ci.apache.org/projects/flink/flink-docs-release-1.5/ops/deployment/yarn_setup.html#run-a-flink-job-on-yarn
So, I found the solution before I finish writing the answer. So here it is:
With 1.4.2, when I start a job, I see the line:
> Using the result of 'hadoop classpath' to augment the Hadoop classpath: (And then a lot of paths...)
I don't see this "augmentation" with 1.5.2, so I did it myself by running:
> export HADOOP_CONF_DIR=`hadoop classpath`
> export YARN_CONF_DIR=`hadoop classpath`
Not sure if that is the right way to do it, but now it works !
De : vino yang <yanghua1127@xxxxxxxxx>
Envoyé : mardi 7 août 2018 10:59
À : Florian Simond
Cc : Gary Yao; user@xxxxxxxxxxxxxxxx
Objet : Re: Could not build the program from JAR file.
The log you gave before prints the phrase "Use the help option (-h or --help) to get help on the command." and the log just shows "-yn" as a jar file that looks like yours. There is a problem with the parameter specification.
Can you share the full client log and your commit command?
2018-08-07 15:22 GMT+08:00 Florian Simond <florian.simond@xxxxxxxxxx>: