Re: handling round robin errors

Hello, I have one comment if I may. When switching from Apache to nginx, the server_alias setting is lost. Maybe the script should be "improved" to check if the mirror is answering to "XX.php.net" and "XX[0-9].php.net" and not to care which one is first. I don't know if there's a good reason behind the forcing of the ro.php.net vs ro1.php.net. Thank you, Stefan. On Mon, Mar 18, 2013 at 2:16 AM, Ferenc Kovacs <tyra3l@gmail.com> wrote: > Hi Dan, > > After thinking about what would be the best way to handle the round-robin > mirror-test, I think that it should be handles as a normal mirror error, > which after a couple of retries disables the mirror itself. > Currently only 3 mirrors doesn't have round robin enabled (cn1, hu2 both > disabled for other reasons and www.php.net, which also intentionally left > out from the round robin), so we can say that everybody who complies to the > mirroring rules also complies for this requirement. > If a new mirror fails on this, they should fix it before we enable them, > and if an existing mirror fails this test (which means that they have > changed something in their configuration, and there could be other possible > problems) it is ok that we temporary disable the mirror, and we can help > the hoster fix the issue, or worst case we can simply disable the round > robin flag for that particular mirror and the test should pass again. > This change would make this test to fall in line with every other mirror > testsand would also eliminate/greatly reduce the current false positives. > If you think that for some reason it would be a bad idea to merge this with > the other mirror-test errors, then please tell me why, and if we drop this > idea I will add a new field to the mirrors table where I can store the > retry counts for reducing the false positives for the current approach. > > -- > Ferenc Kovács > @Tyr43l - http://tyrael.hu >