Need suggestion for dns balancing

Hi there,

currently im having a issue where my nextdns sometimes goes down or even has huge latency spikes perhaps it could be that I was seeing a ipv6 entry enabled inside the stubby.yml configuration when I have ipv6 disabled (I have uncommented this line after my foundings and used stubby -g), it could be that it sometimes tries to connect to the ipv6 with this enabled for some reason, but not sure but that should now not happen anymore.

but I’m searching for a solution to counter this problem in case it is indeed a problematic outage of the dns.

now I explored a few ways of doing it but none seem like really nice:

in luci there is a option to add another dns inside the dns forwardings field where 127.0.0.1#53535 is used for nextdns, if I put my own secondairy dns and make sure strict dns is set I still get a huge delay if nextdns is down, this is not really useable for what I want + it could also give buggy results inside the gl-inet web ui.

then there is also a way to advertise the dns through dhcp but I also want to enforce dns everywhere for smart devices cus they can easily ignore device dns settings.

none of the above approaches are really working great since the delay it gives, I would love to add some kind of QoS implementation or failover over the dns since im experiencing alot of dns jitter for a unknown reason, can somebody suggest me a trick or package which can do this better? :slight_smile:

I’m using flint 1800ax and using 3.208 release build.

1 Like

Thanks for the suggestions. DNS is really difficult.

We are planning to re-design the DNS module and still not finalized.

1 Like

I am doing something really weird, I do it more for backup than balancing but could help - better than nothing after all.

Two identical routers, incidentally far apart so that on the wireless side of things, some clients connect to one and others to another. They both have DNS identically configured. One could set one to forward certain domain queries to another.

Both routers have DHCP, with different ranges of course. The DHCP protocol does account for competing DHCP servers, so doing this is no issue from my standpoint. One of those has the “force” flag but actually I should take it out.

Each router’s configuration has its own address as DNS 1 and the other router as DNS 2. There is a third router around doing nothing so thats DNS 3. All DNS routers forward to different upstream servers and go via different Internet connections. Also, totally irrelevant to the discussion, each DHCP points to itself as the primary gateway and the other as the secondary (secondary gateway configuration is quite ignored by all clients I have including Windows, so whatever…).

I am never on site so if one router blows, the transition to the other one is more or less seamless without having to run there to fix things.

My setup is 3…8 PCs, cell phones/tables, 3 printers, 2 plotters, some IP phones.

Both DNS/DHCP routers have fixed DHCP addresses/names for all hardware things in order to avoid a major name resolution mess, and just let the PCs and phones get dynamic addresses.

So, this is all quite the patchwork. Something similar could help in a pinch.