The entrypoint script probably tries to create the logs folder and because it already exists fails to do so. This would only be a problem the very first time the container starts with an empty userdata folder. Each subsequent start of a container with that userdata folder will see that it’s already populated and not try to create folders that already exists.
This should be fixable in the script but I question whether it’d be worth the effort. It’s an error that only occurs under this exact situation which should not be a problem when following the instructions. But I know that a PR would appreciated if one were submitted.