<a>.<b>.<token>.min.dnsp.co, then <token>.minreport.dnsp.co
Whether your resolver sends only what each server needs.
What it does
QNAME minimisation means a resolver asks each server in the chain only for the labels that server is responsible for, instead of the whole name every time. Without it, every server in the delegation chain learns exactly what you were looking for.
This test takes two queries. The first is an ordinary lookup of <a>.<b>.<token>.min.dnsp.co, then <token>.minreport.dnsp.co: it returns a normal address, and its only purpose is to be watched. Every query that reaches us underneath your token is recorded.
The second query reads that recording back. Ask for <token>.minreport.dnsp.co as TXT and we report how many queries arrived and how much of the name each one carried. The token is the label immediately in front of min, and it is what ties the two queries together.
Use a token you have not used before. A resolver with a warm cache answers from memory without asking us, and the report will then describe an earlier run rather than yours.
Example
Reading the answer
Several queries arrived at growing depth. Your resolver minimises, so servers above us learn less about you.
One query carried the whole name. Every server in the delegation chain sees everything you asked for.
Nothing reached us under that token. Either the first query has not run yet, or a cache answered it. Use a fresh token.
Notes
- The token is the label immediately in front of min. The labels before it are the ladder a minimising resolver walks down, and more of them make the behaviour easier to see.
- Use a token you have not used before, every run. A cached answer never reaches us and reports nothing.
- Aim it through your resolver. Asked directly there is no chain to minimise, so the result only describes your own client.