I'm thinking that I'll periodically snapshot this relay and upload it to a blossom server. Then the server page can post a hash with the download link to grab the data.

Then anyone could run the graph if they want on their own relay, and they'll already get seeded all of the scrobble events. 

 Heads up I think the stats job hit a glitch and duplicated everything. 

 Ha it sure did. Right now the server just stores everything in memory, probably a concurrency thing