Re: [OpenWrt-Devel] [LEDE-DEV] RFC netifd: UCI parameter to sort name servers in resolv.conf.auto

One other point about provider order in dnsmasq. dnsmasq added long ago the facility (particularly for reverse lookups on ipv6) to bind source addresses to destination dns servers. This solves a portion of this problem. Stuff coming in from one ipv6 network gets dns lookups out the right network

Re: [OpenWrt-Devel] [LEDE-DEV] RFC netifd: UCI parameter to sort name servers in resolv.conf.auto

On Mon, Sep 5, 2016 at 3:23 PM, Kevin Darbyshire-Bryant wrote: > > > On 05/09/16 13:49, Hans Dedecker wrote: > > On Mon, Sep 5, 2016 at 1:49 PM, Jo-Philipp Wich wrote: > > Hi Hans, > > imho it would also make sense to take any existing metric setting

Re: [OpenWrt-Devel] [LEDE-DEV] RFC netifd: UCI parameter to sort name servers in resolv.conf.auto

On 05/09/16 13:49, Hans Dedecker wrote: On Mon, Sep 5, 2016 at 1:49 PM, Jo-Philipp Wich wrote: Hi Hans, imho it would also make sense to take any existing metric setting into account as well. At least I'd expect that if I have a wan1 with metric 10 and a wan2 with metric 20

Re: [OpenWrt-Devel] [LEDE-DEV] RFC netifd: UCI parameter to sort name servers in resolv.conf.auto

On Mon, Sep 5, 2016 at 1:49 PM, Jo-Philipp Wich wrote: > Hi Hans, > > imho it would also make sense to take any existing metric setting into > account as well. At least I'd expect that if I have a wan1 with metric > 10 and a wan2 with metric 20 that the DNS server entries "inherit"

Re: [OpenWrt-Devel] [LEDE-DEV] RFC netifd: UCI parameter to sort name servers in resolv.conf.auto

Hi Hans, imho it would also make sense to take any existing metric setting into account as well. At least I'd expect that if I have a wan1 with metric 10 and a wan2 with metric 20 that the DNS server entries "inherit" the same weight/order. So priority wise I'd first order by dns metric, then by