Support strict order for upstreams (fallback DNS servers)

[timkgh]

I use 2 upstreams from different providers, Cloudflare and Quad9. Cloudflare is fast from my location, Quad9 is about 2x slower. So I'd like to use Cloudflare all the time, unless it's unavailable (which happened before). When it's not available I'd like AGH to switch to the second provider (Quad9 in this case), then periodically check the first provider and switch back to it when it is available again.

For now I can use parallel queries but it seems wasteful and also don't want both providers to have a log of my queries all the time.

This is somewhat related to #3672, if the load balancing feature worked correctly it would be close to equivalent to strict order, unless someone wanted the slower upstream to be the main one.

[agneevX]

This is a good feature request... that way primary and backup servers can be used.

[timkgh]

This seems very similar to the dnsproxy --upstream and --fallback options. AGH uses dnsproxy, is it just a matter of exposing the config options in AGH?

[ghost]

This is a good feature request... that way primary and backup servers can be used.

I agree. This would also allow for upstream dns resolution to prefer dns over a vpn tunnel if the tunnel is up, where it only falls back to a public dns server if the tunnel is down. So say:

10.1.1.1
8.8.8.8

If 10.1.1.1 is reachable, only that gets used. Otherwise 8.8.8.8 will be used as a fallback.

[timkgh]

@ameshkov How difficult would it be to expose the dnsproxy feature through AGH? Thanks!

[ameshkov]

Not too complicated, UI changes is what slows this down.

[ghost]

Hi there, is there any update regarding the implementation of this option? 🙏

[ainar-g]

@komic, it's still unplanned, as you can see by the absence of a set milestone. Please use the 👍 reaction to show your support for the feature.

[agneevX]

Hello @ainar-g and team, today I faced an issue where this feature request would've come in really handy and potentially mitigated the issue.

Google DNS was returning REFUSED and SERVFAIL intermittently for a specific domain while other DNS servers were responding properly.

I'd like to therefore ask that this feature be prioritized.

[rrfeng]

#5731
I have made a pull request to support fallback option which dnsproxy already supported.
When upstream is not reachable (include the domain specific upstream), the fallback upstreams are used.

[ainar-g]

Now that we've improved the fallback upstreams in dnsproxy, I think we can implement this using the fallback mechanism. That is, if the main upstreams fail, use the new additional configuration field instead.

[duckxx]

fallback DNS支持dot或doh吗？

[ainar-g]

@duckxx, the same types of upstreams as in the main upstream fields are supported.

The fallback feature is going to be released in the upcoming v0.107.37.