We've finally managed to reproduce the problem on our setups and identify the root cause of this issue (thanks to details of your setups posted here! 
).
It appears that on RHEL (Docker 1.12/13), when Docker daemon logging driver is set to journald
neither bootstrap nor upgrade monitor does not end streaming logs from the process and it times out.
Therefore, at the end it produces a false negative - throws an exception, even though the
bootstrap/upgrade process has been completed successfully.
We'll update our documentation with this item as a known issue and work on the fix which
future ECE release is for sure going to provide.
Apologies for any inconveniences and thanks again for your priceless input in this discussion!