That's not correct, if that was the case, TCP applications in general would not work in anycast setups, which obviously isn't the case.
It's correct that if you'd end up getting rerouted all of a sudden, it would trigger a reset, but the application would (likely) still succeed after establishing the TCP connection again. Running DNS (and accepting TCP queries) in an anycasted setup is rather common :) I run a few of such setups. On 2 Apr 2023, at 14:08, Marco Mangione <marco.mangi...@gmail.com> wrote: becouse of split path routing, there are 2 or more anycast node and network load balancing causes packet transmitted to the anycast address to alternate between those nodes Il giorno dom 2 apr 2023 alle ore 08:59 Lucas Rolff <lu...@lucasrolff.com<mailto:lu...@lucasrolff.com>> ha scritto: > Because if traffic is forced on TCP, for example with a TC=1, not everything > could work correctly. Why not? > On 2 Apr 2023, at 13:55, Marco Mangione via dnsdist > <dnsdist@mailman.powerdns.com<mailto:dnsdist@mailman.powerdns.com>> wrote: > > Hi everyone, > I would like to open a conversation about a dnsdist+pdns setup in anycast > mode, therefore multiple locations with multiple instances inside. > All OK if we talk about UDP, but what about TCP? Has anyone had any > experience? > > Because if traffic is forced on TCP, for example with a TC=1, not everything > could work correctly. > > M1 > _______________________________________________ > dnsdist mailing list > dnsdist@mailman.powerdns.com<mailto:dnsdist@mailman.powerdns.com> > https://mailman.powerdns.com/mailman/listinfo/dnsdist
_______________________________________________ dnsdist mailing list dnsdist@mailman.powerdns.com https://mailman.powerdns.com/mailman/listinfo/dnsdist
