Bug #37308 for bind9-public: -4 but disable6 not used in some places

On Thu Sep 25 04:34:53 2014, jreed@isc.org wrote: > > named manual says about -4: > -4 > Use IPv4 only even if the host machine is capable of IPv6. > > > I used named -4 but log output said: > > 22-Sep-2014 14:20:18.922 using default UDP/IPv6 port range: [1024, > 65535] > 22-Sep-2014 14:20:18.923 no IPv6 interfaces found > ... > 22-Sep-2014 14:20:24.899 no IPv6 interfaces found > > It said that 81 times within 16 minutes. Presumable you have interfaces coming and going and named is re-probing the interface states as a result of those comings and goings. > On another named -4 run, I had it configured with: > listen-on-v6 { none; }; > and it had the same behaviour. > I didn't configure all the various other transfer etc tunables for v6 > to try to turn it off. > > So why is use-v6-udp-ports done if -4 switch sets disable6 to true? Because it is simpler and less error prone to always setup the port tables than to conditionally set them up. > And why is isc_net_probeipv6() done if disable6 is true? Because that is the generic mechanism used to see if IPv6 is supported either by the OS or as the result of a command line switch. Doing it one way in named and a different way in the libraries is more error prone. > Either let's use disable6 or let's fix documentation. We don't use IPv6 with 'named -4' to send or receive queries. That doesn't mean we still don't need to discover stuff. > Thanks

---