The advice, which is specifically for virtual machines using Azure, shows that sometimes the solution to a catastrophic failure is turn it off and on again. And again.
“We have received feedback from customers that several reboots (as many as 15 have been reported) may be required, but overall feedback is that reboots are an effective troubleshooting step at this stage."
A lot of liberties taken with that headline.