Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
Description
Our users may be new to Hadoop. And many of our developers are not willing to set up hadoop and maintain it. I propose a docker image dedicated for try/develop submarine in one container.
The image contains vanilla Hadoop, spark and submarine. When running, below commands can give a sense of how it's like:
build_submarine_dev_env.sh docker run -it -h submarine-dev --net=bridge --privileged local/hadoop-docker:submarine /bin/bash /tmp/hadoop-config/bootstrap.sh su yarn && cd && cd submarine && ./run_submarine_minist_tony.sh