TODO.md

@@ -30,7 +30,12 @@ confirm make check still passes.
   `internal/resolver/livedns_test.go` adds a package-wide concurrency
   gate (so parallel tests stop bursting at the first root server),
   retry with exponential backoff on transport failures only, and
-  quorum instead of unanimity for multi-nameserver assertions. The
+  quorum instead of unanimity for multi-nameserver assertions. Quorum
+  tolerates silence only: every per-nameserver status must be in a
+  closed allowlist (`ok`/`timeout`/`error`, or
+  `nxdomain`/`timeout`/`error`), so a wrong answer from a minority —
+  `nodata` today, any status added later — fails the test instead of
+  sliding through under the majority. The
   `make test` cap moved to the new org-wide 60s hard cap / 20s target
   with a 90s `-timeout` backstop; `REPO_POLICIES.md` re-vendored
   byte-identical from `sneak/prompts`. No mocks, no `-short`, no build

internal/resolver/livedns_harness_test.go

@@ -16,6 +16,16 @@ import (
 // perform no DNS resolution of any kind, so they neither mock DNS
 // nor depend on it.
 
+// Names for the synthetic status maps below. Nothing is ever queried
+// at them: they are map keys handed to the package's pure counting
+// helpers, not a stand-in for a nameserver.
+const (
+	nsExample1 = "ns1.example."
+	nsExample2 = "ns2.example."
+	nsExample3 = "ns3.example."
+	nsExample4 = "ns4.example."
+)
+
 func TestLiveQuorumIsStrictMajority(t *testing.T) {
 	t.Parallel()
 
@@ -41,20 +51,20 @@ func TestStatusCountingIgnoresSilentNameservers(t *testing.T) {
 	t.Parallel()
 
 	results := map[string]*resolver.NameserverResponse{
-		"ns1.example.": {
-			Nameserver: "ns1.example.",
+		nsExample1: {
+			Nameserver: nsExample1,
 			Status:     resolver.StatusOK,
 		},
-		"ns2.example.": {
-			Nameserver: "ns2.example.",
+		nsExample2: {
+			Nameserver: nsExample2,
 			Status:     resolver.StatusOK,
 		},
-		"ns3.example.": {
-			Nameserver: "ns3.example.",
+		nsExample3: {
+			Nameserver: nsExample3,
 			Status:     resolver.StatusTimeout,
 		},
-		"ns4.example.": {
-			Nameserver: "ns4.example.",
+		nsExample4: {
+			Nameserver: nsExample4,
 			Status:     resolver.StatusError,
 		},
 	}
@@ -106,14 +116,126 @@ func TestRetryLiveGivesEachAttemptADeadline(t *testing.T) {
 	retryLive(t, "deadline", func(ctx context.Context) error {
 		deadline, ok := ctx.Deadline()
 		assert.True(t, ok, "attempt should carry a deadline")
-		assert.LessOrEqual(
-			t, time.Until(deadline), liveAttemptTimeout,
-		)
+
+		remaining := time.Until(deadline)
+
+		assert.LessOrEqual(t, remaining, liveAttemptTimeout)
+
+		// Lower bound too: without one this passes for a
+		// deadline far shorter than intended, which would
+		// silently turn every live attempt into an instant
+		// timeout.
+		assert.Greater(t, remaining, liveAttemptTimeout/2)
 
 		return nil
 	})
 }
 
+// TestUnsanctionedStatusesRejectsWrongAnswers is the regression test
+// for the defect this allowlist exists to prevent: a minority of
+// nameservers answering WRONGLY while quorum keeps the suite green.
+// nodata is the case that motivated it — it is a wrong answer, not
+// silence, and it was previously banned by neither test.
+func TestUnsanctionedStatusesRejectsWrongAnswers(t *testing.T) {
+	t.Parallel()
+
+	// Four nameservers, three OK and one answering nodata: a
+	// quorum of three is satisfied and no NXDOMAIN is present, so
+	// the old blocklist assertions both passed on this input.
+	results := map[string]*resolver.NameserverResponse{
+		nsExample1: {
+			Nameserver: nsExample1,
+			Status:     resolver.StatusOK,
+		},
+		nsExample2: {
+			Nameserver: nsExample2,
+			Status:     resolver.StatusOK,
+		},
+		nsExample3: {
+			Nameserver: nsExample3,
+			Status:     resolver.StatusOK,
+		},
+		nsExample4: {
+			Nameserver: nsExample4,
+			Status:     resolver.StatusNoData,
+		},
+	}
+
+	assert.GreaterOrEqual(
+		t,
+		countStatus(results, resolver.StatusOK),
+		liveQuorum(len(results)),
+	)
+	assert.Zero(t, countStatus(results, resolver.StatusNXDomain))
+
+	// nodata is an ANSWER, so it never triggers a retry: nothing
+	// but the allowlist stands between it and a false green.
+	assert.Equal(t, len(results), answeredCount(results))
+
+	assert.Equal(
+		t,
+		[]string{nsExample4 + "=nodata"},
+		unsanctionedStatuses(
+			results,
+			resolver.StatusOK,
+			resolver.StatusTimeout,
+			resolver.StatusError,
+		),
+		"nodata must be reported as an unsanctioned status",
+	)
+}
+
+func TestUnsanctionedStatusesToleratesSilenceOnly(t *testing.T) {
+	t.Parallel()
+
+	results := map[string]*resolver.NameserverResponse{
+		nsExample1: {
+			Nameserver: nsExample1,
+			Status:     resolver.StatusNXDomain,
+		},
+		nsExample2: {
+			Nameserver: nsExample2,
+			Status:     resolver.StatusTimeout,
+		},
+		nsExample3: {
+			Nameserver: nsExample3,
+			Status:     resolver.StatusError,
+		},
+	}
+
+	allowed := []string{
+		resolver.StatusNXDomain,
+		resolver.StatusTimeout,
+		resolver.StatusError,
+	}
+
+	assert.Empty(
+		t,
+		unsanctionedStatuses(results, allowed...),
+		"timeout and error are non-answers and are tolerated",
+	)
+
+	// The same silent nameservers do not count towards a quorum.
+	assert.Equal(t, 1, answeredCount(results))
+
+	// An unknown status is treated as silence by answeredCount —
+	// so it retries and fails loudly — and is unsanctioned by the
+	// allowlist rather than quietly permitted.
+	const laterStatus = "some-status-added-later"
+
+	results[nsExample4] = &resolver.NameserverResponse{
+		Nameserver: nsExample4,
+		Status:     laterStatus,
+	}
+
+	assert.Equal(t, 1, answeredCount(results))
+	assert.Equal(
+		t,
+		[]string{nsExample4 + "=" + laterStatus},
+		unsanctionedStatuses(results, allowed...),
+	)
+}
+
 func TestRunLiveBoundsConcurrency(t *testing.T) {
 	t.Parallel()
 

internal/resolver/livedns_test.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"slices"
 	"sort"
 	"strings"
 	"testing"
@@ -44,6 +45,15 @@ import (
 //     nameservers, a strict majority answering as expected is
 //     enough; a server that fails to answer is tolerated, while a
 //     server that answers *wrongly* still fails the test.
+//
+// The tolerance in (3) is expressed as an ALLOWLIST of sanctioned
+// statuses, never as a blocklist of known-bad ones. A blocklist bans
+// the one wrong answer its author thought of and silently admits
+// every other status, including any added to the resolver later; an
+// allowlist fails on anything nobody explicitly sanctioned. Silence
+// (timeout, error) is the only thing quorum exists to tolerate. A
+// *wrong answer* — nxdomain for a name that exists, ok for one that
+// does not, nodata for either — is never tolerated at any count.
 
 const (
 	// liveAttempts is how many times a live DNS operation is
@@ -172,14 +182,62 @@ func countStatus(
 	return n
 }
 
+// liveAnswerStatuses is the closed set of statuses that count as a
+// nameserver having ANSWERED at all, whether or not the test agrees
+// with the answer. It is deliberately an allowlist: a status added
+// to the resolver later is treated as silence, so it can only ever
+// cause a retry and then a loud failure, never a quiet pass.
+func liveAnswerStatuses() []string {
+	return []string{
+		resolver.StatusOK,
+		resolver.StatusNXDomain,
+		resolver.StatusNoData,
+	}
+}
+
 // answeredCount counts the nameservers that produced an answer of
 // any kind, as opposed to failing or timing out.
 func answeredCount(
 	results map[string]*resolver.NameserverResponse,
 ) int {
-	return len(results) -
-		countStatus(results, resolver.StatusError) -
-		countStatus(results, resolver.StatusTimeout)
+	answers := liveAnswerStatuses()
+
+	n := 0
+
+	for _, resp := range results {
+		if slices.Contains(answers, resp.Status) {
+			n++
+		}
+	}
+
+	return n
+}
+
+// unsanctionedStatuses returns "nameserver=status" for every result
+// whose status the caller did not explicitly sanction, sorted for a
+// stable failure message. Callers pass the full closed set they will
+// accept — the expected answer plus whichever non-answers (timeout,
+// error) quorum is allowed to tolerate — so that any status outside
+// it fails the test by name.
+func unsanctionedStatuses(
+	results map[string]*resolver.NameserverResponse,
+	allowed ...string,
+) []string {
+	offenders := make([]string, 0, len(results))
+
+	for ns, resp := range results {
+		if slices.Contains(allowed, resp.Status) {
+			continue
+		}
+
+		offenders = append(
+			offenders, fmt.Sprintf("%s=%s", ns, resp.Status),
+		)
+	}
+
+	sort.Strings(offenders)
+
+	return offenders
 }
 
 // describeStatuses renders per-nameserver statuses for use in

internal/resolver/resolver_test.go

@@ -324,12 +324,21 @@ func TestQueryAllNameservers_AllReturnOK(t *testing.T) {
 		describeStatuses(results),
 	)
 
-	// Any nameserver claiming google.com does not exist is a
-	// real failure and is never tolerated.
-	assert.Zero(
+	// Quorum tolerates SILENCE only. Every individual result must
+	// be either the expected answer or a non-answer: ok, timeout
+	// or error, and nothing else. Stated as a closed allowlist so
+	// that a wrong answer no one thought to ban — nxdomain and
+	// nodata today, any status added later — fails here rather
+	// than sliding through under the quorum.
+	assert.Empty(
 		t,
-		countStatus(results, resolver.StatusNXDomain),
-		"no nameserver should report NXDOMAIN: %s",
+		unsanctionedStatuses(
+			results,
+			resolver.StatusOK,
+			resolver.StatusTimeout,
+			resolver.StatusError,
+		),
+		"every nameserver must answer OK or not answer at all: %s",
 		describeStatuses(results),
 	)
 }
@@ -352,12 +361,20 @@ func TestQueryAllNameservers_NXDomainFromAllNS(
 		describeStatuses(results),
 	)
 
-	// Silence is tolerated; a positive answer for a name that
-	// does not exist is not.
-	assert.Zero(
+	// Silence is tolerated; any actual answer other than NXDOMAIN
+	// is not. Closed allowlist for the same reason as above: a
+	// server answering `ok` or `nodata` for a name that must not
+	// exist is a wrong answer, not a slow one.
+	assert.Empty(
 		t,
-		countStatus(results, resolver.StatusOK),
-		"no nameserver should answer OK: %s",
+		unsanctionedStatuses(
+			results,
+			resolver.StatusNXDomain,
+			resolver.StatusTimeout,
+			resolver.StatusError,
+		),
+		"every nameserver must report NXDOMAIN or not answer "+
+			"at all: %s",
 		describeStatuses(results),
 	)
 }