Apache Spark enhanced with native Kubernetes scheduler back-end: NOTE this repository is being ARCHIVED as all new development for the kubernetes scheduler back-end is now on https://github.com/apache/spark/
PR #453 implements a new server token-refresh-server that renews HDFS tokens used by long-running Spark jobs. The PR made the server an optional component that gets activated under a separate maven profile Pkubernetes-hdfs-extra.
We should include the profile in the Jenkins CI setup after #453 is merged in, so that subsequent PRs would build the project and run the unit tests. (The integration test will be addressed later)
For the main build, the command would look like the following (the bold part is new):
We should add this after #453 is merged in because maven will crash at the missing project resource-managers/kubernetes/token-refresh-server in the -pl parameter of the unit test build command line. (Missing profiles, on the other hand, are simply ignored)
Please feel free to ask questions or make suggestions. (This was discussed in the SIG meeting this morning)
@ifilonenko @liyinan926 @foxish @erikerlandson
PR #453 implements a new server
token-refresh-serverthat renews HDFS tokens used by long-running Spark jobs. The PR made the server an optional component that gets activated under a separate maven profilePkubernetes-hdfs-extra.We should include the profile in the Jenkins CI setup after #453 is merged in, so that subsequent PRs would build the project and run the unit tests. (The integration test will be addressed later)
For the main build, the command would look like the following (the bold part is new):
The unit test will be like:
We should add this after #453 is merged in because maven will crash at the missing project
resource-managers/kubernetes/token-refresh-serverin the-plparameter of the unit test build command line. (Missing profiles, on the other hand, are simply ignored)Please feel free to ask questions or make suggestions. (This was discussed in the SIG meeting this morning)