A separate questions. Are their any common practices for recovering the system from message bus failures (such as SQS in your case).

Seeing your post made me realize if SQS goes down, a lot of random things will be happening in my application.