health-drain-avoids-backend

Purpose

An operator drain (conduitctl health set down) removes a backend from Route while a sibling stays Up, so client traffic goes only to the live backend.

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 peer) and dead (unused IP); probes are slow so they do not race the drain proof.
2. Control plane listens on 0.0.0.0:5199 (published to the host).
3. The harness runs conduitctl health set down --pool default --backend dead, then digs www.smoke.test.
4. Checks require NOERROR, one peer query, and forward success only on live.

Outcomes

Outcome	Meaning for operators
pass	After draining dead, traffic stayed on live.
fail	Drain did not stick, digs failed, or forwards still hit dead.
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: property, peer-query-count, metrics-delta