@p @cjd It would also align with the issue I had with Gitea. They also processed responses and followed up with requests, which ended up with the repo sizes balooning thanks to them downloading the zip and tar archives along with git bundles and every commit diff page they found out about for every public repo.

I didn't see EC2 and Akamai, but I did see a bunch of IPs belonging to Google's usercontent separation last month.