docker-pihole-unbound VS dnsperftest

I'm running the chriscrowe 'one container' build for pihole for ad-blocking (https://github.com/chriscrowe/docker-pihole-unbound), and unbound for private recursive DNS which is running great on my Zima board.

I believe my problem stems from when I tried to install the pihole-unbound container (https://github.com/chriscrowe/docker-pihole-unbound)

I've got lucky enough that I could get in Nov-19 a Raspberry Pi 4 with 4GB of RAM that I use as a NAS+Pi-Hole + Unbound +Other stuff (running Void MUSL of course) and instead of an SD I use an HDD for the OS + a 5TB HDD for the data and even fully loaded all cores it barely consumes more than 22W.

I was only able to get pihole macvlan approach working with an ssh access + docker-compose approach. I used the crowe one-container approach Still had to enable Open vSwitch for the network in Synology first.

You can follow the doc to do those install steps within your own Dockerfile. Or you can use the commonly recommended chriscrowe image, which does what I’m describing.

When I started experimenting with this topic chriscrowe's was a baseline for stuff that I tryed out, so kudos to chris!

It seems you’re looking at a wrong file. Here is a docker file for one-container solution: https://github.com/chriscrowe/docker-pihole-unbound/blob/55c88afaf8d76958923adc38f4c12a80e1cb9084/one-container/pihole-unbound/Dockerfile

2) you can use something like this maintained GitHub repo. Just decide if you want to use 1 or 2 containers and modify the ip and gateways, etc accordingly.

Running DNSPerfTest on a linux machine, I see:

When I run a DNS performance test (which will detect your device settings and add a predefined list for comparison I get these results:

Namebench and DNSPerfTest

So I looked into how I could improve performance, and found the Unbound option serve-expired, and on the Pi-hole side, the dnsmasq option use-stale-cache. They seem to do the same thing, just at different points in the DNS lookup chain, and either one gives me a huge, noticeable speedup. Using dnsperftest, my setup was always the slowest resolver, and now it's usually at the top.

Link to script for anyone interested: https://github.com/cleanbrowsing/dnsperftest

The fellas of Cleanbrowsing created a nifty little script

With a DNS test, it's always far behind opendns, google or cloudflare.

I still believe that Quad9 is better, but I also noticed that when I run this script https://github.com/cleanbrowsing/dnsperftest only against Quad9 and Cloudflare I get many timeouts for Quad9 ... :/