That's step one...
Back to the original question, it's worth asking what is the downside of configuring a humongous size in the first place, more than you'll ever need?
The answer depends entirely on the upper layer filesystem you're using...
- All filesystems have a maximum size - is it big enough?
- How big is the filesystem meta-data overhead when the size is very large?
I believe resizing does not do a whole lot of I/O but that of course depends on the filesystem and resize tool
Running fsck probably does a lot more I/O, but again how much is dependent on the filesystem, and things like the average file size (fsck reads filesystem meta-data but usually not actual file data).
One of the nice features of s3backer is the simplicity of the service it performs. Unfortunately, that means most practical questions must pass through to the upper filesystem to be answered :)
-Archie