I've been playing around with a Raspberry Pi hosted with Mythic Beasts and wanted a way to keep a large number of infrequently-accessed files on the server. A cost-effective place to store such data is on Amazon S3. After looking around at ways to mount an S3 bucket on the Raspberry Pi and have it appear as a regular filesystem, I came across s3backer. This provides a single file, divided into blocks, each of which is stored as an S3 object. A filesystem is mounted onto this file using a loopback mount. This setup has many benefits which are explained well in the README. Here's how to set it up on the Raspberry Pi.
I'm going to assume that you have an S3 bucket mybucket created and a user with credentials to access it. Although not explicitly specified, some of the commands below need to be run as root.
First, install prerequisites (the first line is