We're mostly deployed on eu-west-1 but still seeing weird STS and IAM failures, likely due to internal AWS dependencies.
Also we use Docker Hub, NPM and a bunch of other services that are hosted by their vendors on us-east-1 so even non AWS customers often can't avoid the blast radius of us-east-1 (though the NPM issue mostly affects devs updating/adding dependencies, our CI builds use our internal mirror)
FYI:
1. AWS IAM mutations all go through us-east-1 before being replicated to other public/commercial regions. Read/List operations should use local regional stacks. I expect you'll see a concept of "home region" give you flexibility on the write path in the future.
2. STS has both global and regional endpoints. Make sure you're setup to use regional endpoints in your clients https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credenti...
Also we use Docker Hub, NPM and a bunch of other services that are hosted by their vendors on us-east-1 so even non AWS customers often can't avoid the blast radius of us-east-1 (though the NPM issue mostly affects devs updating/adding dependencies, our CI builds use our internal mirror)