question: how to use random resolving?

[monperrus]

Hi!

the code suggests that one can randomize the upstream servers:

Random   = "random"
NginxWRR = "weighted_round_robin"
LVSWRR   = "lvs_weighted_round_robin"

How to configure dns-over-https to use it?

[m13253]

In the doh-client.conf, find the following lines:

# available selector: random or weighted_round_robin or lvs_weighted_round_robin
upstream_selector = "random"

The default is already "random".

When you have more than one upstream servers, a random upstream server will be chosen for each request.

[monperrus]

thanks, see #67

[monperrus]

One problem with pure randomization is that "over time (days, weeks) any resolver you use (either in round robin or with a random distribution) will likely acquire enough information about an end user to profile", which is a problem for privacy (source)

One solution for that is a pseudo-random choice: "the choice of the server is consistent: if the same domain is resolved twice, the same DoH server for both queries is used". This is implemented for instance in nss-tls.

What about supporting upstream_selector = "pseudo-random"?

[m13253]

It's a good idea. And I suggest we name the selector "hostname_hash" or "hash" to match the naming of other load balancers.

I currently do not have time to implement new feature. So please help me if you are willing to. Pull requests are welcome. <3