Amazon Lightsail Containers exists for teams who want to deploy something without becoming infrastructure specialists. Those are exactly the teams least likely to have a resilience practice, and the platform gives them no host to test against even if they wanted one.

What tends to happen next is familiar. The service works, so it stays. It picks up responsibilities it wasn't designed for. Nobody revisits the error handling written on day one, because nothing has forced them to, and the container keeps running whether or not the thing it depends on is healthy.

Testing at the code level is the only option here, and it's a cheap one. Add Failure Flags to your container and inject latency or errors into a specific dependency call. You'll learn how the application responds when its database or an upstream API stops answering, whether it recovers on its own, and how long that takes.

For a workload with no runbook and no on-call rotation, knowing the answer in advance is most of the value.