Noted in #1537 (comment), the retrieve jobs notebook is failing when the most recent job is cancelled. We should probably filter by job state, something like:
# Get ID of most recent successful job for demonstration.
# This will not work if you've never successfully run a job.
successful_jobs = service.jobs().filter(lambda j: j.status() == "DONE")
job_id = successful_jobs[0].job_id()
print(job_id)