I believe this is where the pipeline failed.
Steps to reproduce,
Create 6.2.2/6.2.3 instance
Create pipeline with trigger/schedule
Kick off the schedule/trigger
Upgrade CDAP to 6.3 without doing anything to trigger/schedule
Post upgrade the pipeline failed with no logs.
Viewing advanced logs seems to indicate that there was some issue creating the dataproc cluster. But we don't seem to be surfacing the underlying cause as to why it fails.
Without logs I was not even sure if it is a problem during upgrade or if the problem was while creating dataproc cluster.
Attaching available logs for reference.