YARN client authentication fails with SIMPLE authentication is not enabled. Available:[TOKEN] - spring

Question

I've setup a simple local PHD 3.0 Hadoop cluster and followed the steps described in the Spring Yarn Basic Getting Started guide
Running the app against my Hadoop cluster gives
org.apache.hadoop.security.AccessControlException: SIMPLE authentication is not enabled. Available:[TOKEN]
and the following stack trace in the YARN ResourceManager:
org.apache.hadoop.security.AccessControlException: SIMPLE authentication is not enabled. Available:[TOKEN]
at org.apache.hadoop.ipc.Server$Connection.initializeAuthContext(Server.java:1554)
at org.apache.hadoop.ipc.Server$Connection.readAndProcess(Server.java:1510)
at org.apache.hadoop.ipc.Server$Listener.doRead(Server.java:762)
at org.apache.hadoop.ipc.Server$Listener$Reader.doRunLoop(Server.java:636)
at org.apache.hadoop.ipc.Server$Listener$Reader.run(Server.java:607)
This is probably a very basic question.
I'd like simply to run a YARN app test without setting up any authentication.
As I understand, YARN does not allow SIMPLE client authentication:
https://issues.apache.org/jira/browse/YARN-2156
According to this question
How can I pass a Kerberos ticket to Spring Yarn application
I might end up setting up a Kerberos authentication.
Is there a way to run Spring YARN example without elaborate authentication setup?

Answer 1

My mistake was simple.
I had to add
spring:
hadoop:
resourceManagerAddress: myyarnhost:8050
resourceManagerSchedulerAddress: myyarnhost:8030
to the application.yml too, but mixed up the port numbers (8030 for manager and 8050 for ManagerScheduler).
And that typo has caused such effect.
Maybe adding these two configuration properties to the getting started guide could save some time to the next readers.
Also, to run the example against a freshly installed PHD3.0 I had to modify the HDFS client user name by exporting the default HADOOP_USER_NAME:
export HADOOP_USER_NAME=hdfs

Answer 2

I just tried that with 5 node phd30 cluster and everything was ok:
In build.gradle I used phd30 packages instead of vanilla(which depends on hadoop 2.6.0). Versions in this case should not matter afaik.
compile("org.springframework.data:spring-yarn-boot:2.2.0.RELEASE-phd30")
testCompile("org.springframework.data:spring-yarn-boot-test:2.2.0.RELEASE-phd30")
In src/main/resources/application.yml I changed hdfs, rm and scheduler addresses to match cluster settings:
spring:
hadoop:
fsUri: hdfs://ambari-2.localdomain:8020
resourceManagerAddress: ambari-3.localdomain:8050
resourceManagerSchedulerAddress: ambari-3.localdomain:8030
Then I just ran it externally from my own computer:
$ java -jar target/gs-yarn-basic-single-0.1.0.jar
There's one appmaster and one container executed and app should succeed.
If it still doesn't work then there's something else. I didn't deploy hawk if that makes a difference.