A Settings section in docs/exposure.md: the counting model, a table of all
15 rate-limit settings from the BIND 9.20 ARM, and what was checked
against real named (TCP never limited, NXDOMAIN falls back to
responses-per-second and is counted per zone across names, all-per-second
only drops, slipped answers are TC or BADCOOKIE depending on DNS cookies).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Jarl works out which views are reachable from the internet (match-clients
ending in a catch-all, not private-looking addresses: an internal view of a
global IPv6 prefix is still yours; a server without views: open allow-query
and a non-private listen address), with a per-view override (schema 9), and
checks public ones: rate limiting, open resolver, minimal-any, cache
snooping, open NOTIFY and transfers, version and host name, management
interfaces. Fixes are staged like any change, including rate limiting from
ISC's example in log-only mode with the secondaries exempt, then enforcing.
named logs what rate limiting limits (or would, in log-only mode, where its
counters don't move) as query-errors; Jarl routes those lines to its log
and counts them (rrl breakdown, ratelimited top lists, Prometheus
jarl_ratelimit_logged_total), shown on the new Exposure page.
A change that newly opens a public view up must be confirmed (accept_risk,
--accept-risk, a checkbox on the Changes page; 409 confirm_risk
otherwise). Findings can be acknowledged with a reason and stay visible as
info. GET /exposure, POST /exposure/fix|ack|unack, jarl exposure,
docs/exposure.md.
Also fixes query statistics skipping TSIG-signed queries ("IP#port/key
NAME" in named's log).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
adamandClaude Opus 5.5