Many small fixes mostly related to config files:
Added optional workersPerPod parameter to TeraSort for running at k8s
A small bug is fixed in the script to push the base Kubernetes Docker image to Docker Hub.
Set the default value for mpijob restarts on failures to 1.
Set the default checkpointing timeout to 600 seconds. Checkpointing may take a long time in large jobs.
Moved all cluster independent JobMaster config parameters to common/core.yaml. Deleted them from other cluster specific yamls.
Moved WorkerController config parameters to common/core.yaml. Deleted them from cluster specific yamls.
Commented out default config values.
Many small fixes mostly related to config files: Added optional workersPerPod parameter to TeraSort for running at k8s A small bug is fixed in the script to push the base Kubernetes Docker image to Docker Hub. Set the default value for mpijob restarts on failures to 1. Set the default checkpointing timeout to 600 seconds. Checkpointing may take a long time in large jobs. Moved all cluster independent JobMaster config parameters to common/core.yaml. Deleted them from other cluster specific yamls. Moved WorkerController config parameters to common/core.yaml. Deleted them from cluster specific yamls. Commented out default config values.