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.