-
Notifications
You must be signed in to change notification settings - Fork 7
Developers
Clone the project first. Once cloned, navigate to the project directory and follow these instructions. You should be up and running in less then a min.
This project uses Gradle as its build and dependency management (see http://www.gradle.org/). Gradle is self-provisioning build framework which means you don't have to have Gradle installed to follow the rest of the procedure.
BUILD for development:
Depending on the IDE you are using execute the following gradle script. ECLIPSE:
./gradlew clean eclipse
IntelliJ IDEA
./gradlew clean idea
The above will generate all necessary IDE-specific artifacts to successfully import the project. Import the project into your IDE. For example in Eclipse follow this procedure:
File -> Import -> General -> Existing Project Into Workspace -> browse to the root of the project and click Finish
Once project is successfully imported, navigate to src/main/java/
The initial YARN API used in these demos is in the oz.hadoop.yarn.api package. Please refer to javadocs for more information on its usage. more information will be coming here as well.
One of the main goals of this project is to introduce rapid development with YARN, so it attempts to ensure that all the moving pieces are in place to get up and running in less then 5 min. One of the most essential thing it has is the Mini Yarn Cluster which is essentially a copy of Hadoop's test cluster with few minor modifications. See its javadoc for more details. You can start it by simply executing StartMiniCluster.java. Once started you can run one of the demos right away.
The current state of the project is already setup to run SampleYarnAppLocal . All you need is to start Mini Cluster (see above) and run this demo. The logs will be in
<project-home>/target/LOCAL_YARN_CLUSTER folder. See the javadocs and also go through Introduction section of this wiki to get more information on how to customize.
SampleYarnAppRemote is an identical copy of the previous demo. However it is intended to work with remote cluster. So it contains instructions on what you need to do which mainly point to the fact that you need to have valid yarn-site.xml and core-site.xml on your classpath. Otherwise its identical on all levels. Logs would obviously be in the hadoop cluster's <hadoop-home>/logs/userlogs folder. See the javadocs and also go through Introduction section of this wiki to get more information on how to customize.
More information is coming
Please send question and updates via pull requests and/or raising issues on this project.
Cheers!