I think we do agree about everything so no worries. Priorization of work is the biggest question and I'm sure IPv6 must not be on the top of that list. You know the best.
I'm aware that the pool attracts also audience with exotic setups, but the thing is that if anyone has IPv6 problem they will see it not by failing NTP but earlier with some other use cases and get it fixed for that. But for NTP, even current bad solutions are mostly good enough, so I think that most problems won't really be as bad (or even visible) as thought. Or the problems are just temporary and no one will even see it. I'm excluding servers from this discussion as I assume people running those are smart enough to handle things properly. I just don't like IPv6 being blamed for theoretical problems as I've seen rather popular web and streaming services getting IPv6 enabled permanently overnight and resulting no complains at all. Most of the seen problems are coming from history and the real ones are mostly easy to solve. -- Markku Miettinen Ask Bjørn Hansen <[email protected]> kirjoitti: > >On Mar 13, 2013, at 1:30 AM, Markku Miettinen <[email protected]> wrote: > >> Still, it's not IPv6 to blame, but more ignorant developers. It's just too >> easy to leave things as they are than to add somewhat complex networking >> logic, even that it would improve service level. > >I don't disagree with you in principle, but I have to work with reality. :-) > >> If the pool would enable IPv6 for everything I'd bet that pretty much no one >> would even notice at the user level. Even the tunnels work reliably enough >> for end user NTP needs. (Excluding the zones with very few IPv6 servers of >> course, which might get to a real problem, but it's no different than having >> too little severs in general?) > >Many more zones have "too little IPv6 servers" than "too little severs in >general". > >I have some work on my todo to better deal with "too little servers" in >general, when I've implemented that I can make it also handle IPv6 and setup >the data appropriately for countries with enough IPv6 servers. > >> Additionally to below, the tunnel users won't even start using IPv6 as the >> OS resolves prefer native IPv4 instead. With native IPv6 one is _very_ >> unlikely to see any difference if there are enough servers available. He >> might even get better results with it most of the time. > >The NTP Pool users are using all sorts of unusual operating systems, software >and who knows what – I don't think you can be so sure that they'll all know if >their IPv6 is tunneled/working/reliable or not. > >Obviously long term IPv6 and IPv4 will be on "even footing" in the system; I >am just explaining why I'm threading very cautiously. > >Another problem is that the geo targeting for IPv6 is (still) way way way >behind the quality of the IPv4 targeting. > > >Ask > >_______________________________________________ >pool mailing list >[email protected] >http://lists.ntp.org/listinfo/pool _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
