How to architect long-running processes on EB workers?

I have a few processes that I've offloaded onto EB Worker environments, but there are some that take a long time, such as daily backups or file imports. However, what happens is that the SQS message times out waiting to hear a response and then it refires again later, causing duplicated backups or imports. I have extended the queue visibility timeout and the receive message wait time is 0 (infinite?) and updated the proxy nginx server config timeout to a long time, but I was wondering if there was general architecture guidance for something like this? Like if your worker is going to take a long time (15 minutes?), how do you set that up? Do you need to worry if it was successful?