...
Pipelines are configured by default to run on a remote Dataproc cluster. When you run your pipeline, Cloud Data Fusion runs the pipeline on a Dataproc cluster by SSHing to the cluster’s master node and launching a Hadoop job from the node. If Cloud Data Fusion is unable to SSH to the master node due to lack of network connectivity or authentication failure, the pipeline run will fail and a JSchException will appear in the pipeline logs.
...