@ev yes, this would be great. The simple way would be to pipe the flumelog into another instance of the log, but filtering out the feeds to be deleted. put that into another flume instance, and when it completes, move all the new flume files over the original (easy since it's all in one directory) then reload the scuttlebot instance (which will regenerate the indexes). It would help a lot if more people understood flume in detail, so I'd be very happy to advise someone working on this.