Last month I had around 1,500 users and served 8.7million requests. It’s a fraction of the requests the main mirror gets, but it’s been enough to burn through a hard drive (and as of this morning, another) and pretty much peg the CPU at 100%.
I don’t have an endless supply of spare hardware to throw at it, and the hardware I do have is spare for a reason — whatever was leftover from an upgrade or kinda-sorta-working. Definitely not server-level stuff.
Last night I configured a mirror on a VPS I have and briefly shifted traffic to it. It worked well, until the VPS went down and I got a nasty-gram from the host. I was hoping that would solve some of the issues, but I think it was just too much for the VPS.
I’m going to try to get it up and running again (migrating the vm to a new datastore right now) but since I don’t even use the MusicBrainz database myself, I really can’t see myself buying hardware to keep it running if it continues to cause issues.
I hope you guys understand. If you have any questions, I’m more than happy to answer them.