>>106963675
Accidentally taking down 1 pod with a bad update happens. Rolling that bad change out to am entire region and fucking your companies reputation to the tune of hundreds of billions of dollars in a single day is far more than a simple oversight. It implies there are no guardrails for preventing things like this.
Worse yet it shows that fuck ups are only caught when they burn the entire building down. Just imagine how much jank is under the hood that isn't big enough to cause a global stink and is just mildly inconvenient so they call it a p3 and move on. Now imagine that every team at Amazon has at least a dozen of those. And the iceberg is even bigger for the ones that customers haven't reported yet. It's fucked.