From bug 1253051 comment 16, in the case of network errors we might leave batches around on the server for no reason. We shouldn't do this. Implementing this would require a server API for destroying the data associated with a batch, and updates to the clients to actually do this.

Ryan, this bug came up in our triage today. We felt that since the server would have to clean up anyway, then there would be little value in doing this from the client. If you agree, then close please. :)

In the case of network errors, we are unlikely to be able to tell the server to cleanup, and we definitely aren't going to start persisting these and attempt to clean up later. Also, in bug 1250189 comment 69, rfkelly said that he feels this isn't necessary. I'll leave the needinfo on rfkelly to decide, but it sounds to me like this should be WONTFIX.

Since the server is going to need to garbage-collect abandoned batches anyway, I'm doubtful that it'll be worth the effort to add explicit API support for this.