I asked admins@ for help, and riastradh stepped in - again, there were a lot of httpd and git processes running, too much for the machine. Killed now, and we're thinking of how to best rate-limit the service. Thomas