Hi, You can specify it in the zeppelin-env.sh, or in the Dockerfile. Zeppelin will look for that variable first in the interpreter settings, and if it does not find it, it will look for it on zeppelin environment variables; so you can specify it in both sides, but as it does not change frenquently it is better on zeppelin environment variable.
El sáb., 20 oct. 2018 a las 0:25, Alex Dzhagriev (<dzh...@gmail.com>) escribió: > Thanks for the quick reply. Should I specify it to the Zeppelin process or > the Spark interpreter? > > Thanks, Alex. > > On Fri, Oct 19, 2018 at 4:53 PM Jeff Zhang <zjf...@gmail.com> wrote: > >> You need to specify SPARK_HOME which is where spark installed. >> >> >> Alex Dzhagriev <dzh...@gmail.com>于2018年10月20日周六 上午3:12写道： >> >>> Hello, >>> >>> I have a remote Spark cluster and I'm trying to use it by setting the >>> spark interpreter property: >>> >>> master spark://spark-cluster-master:7077, however I'm getting the >>> following error: >>> >>> java.lang.RuntimeException: SPARK_HOME is not specified in >>> interpreter-setting for non-local mode, if you specify it in >>> zeppelin-env.sh, please move that into interpreter setting >>> >>> version: Docker Image 0.8.0 >>> >>> Thanks, Alex. >>> >>
