I'm also experiencing a slight elevation in billing rates - got alarms for 10x consumption and I can't check on them...
Edit: also API access is failing, terraform can't take anything down because "connection was forcibly closed"
Imagine triggering a big instance for machine learning or a huge EMR cluster that would otherwise be short lived and not being able to scale it down.
I am quite sure the AWS support will be getting many refund requests over the course of the week.
ECR and API are fucked so it’s impossible to scale anything to the point fire can come out :)
Maybe when the error rate hits 100% they’ll say “error rate now stable”.