Users can skip the assign step and simply post to the weed-fs cluster with a POST containing their own unique ID.

example:

> curl -F file=@/home/chris/myphoto.jpg http://127.0.0.1:8080/myphoto.jpg

Can later be retrieved

> http://localhost:8080/myphoto.jpg

Or deleted

> curl -X DELETE http://127.0.0.1:8080/myphoto.jpg

This would eliminate double calls and function more like a volume. Static assets, for example, could be called as easily as <img src="http://weedfs-cluster.com:8080/myphoto_thumb.jpg"&gt;

This can be achieved with a simple server in front of weed fs. The server will use a key-value store to map url to the file id, and fetch the file id from weed-fs, and pass the content out.

The simple server in front of weed-fs requires me to cluster that as well if I want to provide scale-out without a single point of failure. I want to use weed as the key value store! :)

In our proof-of-concept setup, we use Redis as the (persistant) key-value store; works like a charm

Hi, Ernest, it will be great if you can share your code to do this. Or maybe create an open source project for it.

A "weed filer" option is added.