When doing the lab with the smallest instance, it took a while for the bucket to get the log from the instance. The reason is because the script does other things like install updates and the "stress" utility before actually copying the log file to the bucket. I had to wait about 10 minutes before the log showed up.
This command helps to monitor the output of the startup script after ssh-ing into CE instance:
sudo journalctl -u google-startup-scripts.service