Spark trying to use wrong java

[David Rosenstrauch]

Trying to get shark up and running and working through various bugs and
issues. Just hit a new one: apparently spark is defaulting to use the
oracle jvm:

14/06/23 22:01:34 INFO cluster.SparkDeploySchedulerBackend: Executor
app-20140623220134-0001/9 removed: class java.io.IOException: Cannot run
program "/usr/lib/jvm/java-7-oracle/bin/java" (in directory
"/var/run/spark/work/app-20140623220134-0001/9"): error=2, No such file
or directory

I don't have the oracle jvm installed, however, and so need to instruct
it to use openjdk.

I tried setting JAVA_HOME to /usr/lib/jvm/default-java (which points to
openjdk on my system) in the spark-env.sh file, but it doesn't look like
the spark worker daemon is picking that up.

Anyone have an idea what's the proper way to configure this?

Thanks,

DR

[Honey Joshi]

If you are running it from the shark-shell, try running the command to
export the Java Home before starting the shark shell,
export JAVA_HOME = path
I was having the same issue with java 1.6 and java 1.7.