Spark run on yarn,
The JDK: 1.7.0 _71
Hadoop: 2.6
The spark: 1.6

Spark configuration: configuration only a spark - env. Sh file, the other is no configuration, configure the following
Export SPARK_HOME=/yarn/spark - 1.6.0 - bin - hadoop2.6
Export HADOOP_HOME=/yarn/hadoop - server
Export YARN_CONF_DIR=/yarn/hadoop - server/etc/hadoop
Export HBASE_CONF_DIR=/yarn/hbase - 0.98.14 - hadoop2/conf

Problem a:
Run: a simple wordcount program
Spark - submit - class com. Cyc. Spark. JavaWordCount - master yarn - deploy - mode of client/yarn/file/sparktest jar
Always print:
16/02/23 21:11:30 INFO Client: Application report for application_1456228788292_0009 (state: ACCEPTED)
16/02/23 21:11:31 INFO Client: Application report for application_1456228788292_0009 (state: ACCEPTED)
16/02/23 21:11:32 INFO Client: Application report for application_1456228788292_0009 (state: ACCEPTED)
16/02/23 21:11:33 INFO Client: Application report for application_1456228788292_0009 (state: ACCEPTED)
16/02/23 21:11:34 INFO Client: Application report for application_1456228788292_0009 (state: ACCEPTED)

In the regulation of yarn page, showing

To solve, why the program cannot run?


Problem two:
Modify the spark - env. Sh file to export YARN_CONF_DIR=/yarn/hadoop - server/etc/hadoop literally change a path, such as the export YARN_CONF_DIR=/yarn/hadoop
Program has also been Application report for Application in the regulation of the yarn page also has showed the corresponding Application,
To solve, why wrong path can be yarn received

CodePudding user response:

The building Lord, a problem of how to solve?

CodePudding user response:

The building Lord, a problem how to solve

CodePudding user response:

Look at the log, so don't see what