They will have to write a detailed post-mortem (with many people with titles starting from director watching it every week). Based on what comes out from the post-mortem automation/testing will be implemented to remove/mitigate the failure from the equation in the future.

Unless this is something egregious (ex: a manager not allowing the team to "scale up" in preparation for the event) no one will get fired. Tempers may flare a bit if it is something stupid (it's usually not).

Nothing different from what happened with the (very public) DynamoDB and S3 failures of yesteryear.