health-drain-resume-returns

Purpose

Draining the only answering backend yields a forward-timeout SERVFAIL; resume on that backend (with the sibling held down) restores successful answers. Proves drain/resume round-trip without relying on fail-open alone.

Results appear on the Conduit behavior matrix (one stub peer). Requires a host conduitctl binary (see interop README).

How it works

1. Pool: live (stub) and dead (unused IP); slow probes so operator controls own applied state for the proof.
2. Drain live → dig gets SERVFAIL (only dead remains eligible and times out).
3. Resume live, then drain dead so fail-open cannot re-admit a blackhole.
4. Second dig succeeds on live.
5. Checks require SERVFAIL then NOERROR, one peer query overall, one dead timeout attempt, and one live success.

Outcomes

Outcome	Meaning for operators
pass	Drain caused SERVFAIL; resume restored live answers.
fail	Unexpected rcodes, peer traffic, or missing drain/resume effect.
skip	This peer/profile combination is out of scope.
characterized	Not used for this case today. If it appeared, it would mean a documented peer-specific quirk rather than a Conduit regression.

Matrix: conduit (Conduit behavior)

Suites: full

Oracles: sequence, property, peer-query-count, metrics-delta