A few things have conspired against mirrormanger in the last 12 hours, but appear to be going OK now.
1) httpd re-reads its mod_python targets when re-forking. I had moved the report_mirrorlist mod_python CGI, so that broke between midnight and 3:15am. Thanks to lmacken for restarting httpd on publictest7 for me there.
2) publictest7 went belly up about 10am. Pings ok, but can't connect. I moved the report_mirrorlist CGI over to app3 and got it running there serving traffic now. It belongs there anyhow.
We need to get app3 httpd into supervisord to make sure that stays up now.
Thanks, Matt
infrastructure@lists.fedoraproject.org