#19: fix F1-F3 defects found auditing the phases against PLAN.md

Every fix here is a case where the code and the plan disagreed, and the
code was wrong.

Certificates (section 8.2, 8.4). The parser documented that it skipped
expired entries but never checked, so an expired certificate would be
served until every handshake failed. LastReload() only advanced when
acme.json changed - and it does not change for sixty days - which made
the timestamp risk R2 monitors indistinguishable from a watcher gone
blind. Expiry is now checked on every pass, LastReload records the check
and LastChange the swap, and the <7d/<24h alerts of section 8.4 exist.

ARC (section 5). buildArcSealer resolved the DKIM key at startup, before
runDnsReconciler had provisioned one. On a first boot ARC was disabled
until someone restarted the process, which the log even admitted while
offering no way for it to start. A LazySealer binds on first use instead.

Health (section 8.4). There was no degraded state at all. /health now
reports one - still 200, because restarting cannot fix a certificate
Traefik never issued.

Bounces. DSNs were sent From postmaster@lambdamail.local, a domain that
does not resolve, so the notification was itself undeliverable. They now
carry the configured mail host, a Message-ID, and Auto-Submitted so a
vacation responder cannot loop against them.

Fan-out atomicity (section 6.1). Persist ran one transaction per
recipient, so a failure halfway through an alias fan-out left earlier
recipients with a copy while the sender was told to retry - and the retry
delivered to them twice. PersistAll commits the whole delivery or none of
it. Verified against Postgres by reverting the repository to the old
behaviour and confirming the new test fails on it.

SMTP replies. The use case encoded replies in error text ("550 5.7.1 ...")
which the adapter re-parsed by prefix, so rewording a message could turn a
permanent refusal into a retry. SmtpRejection carries the decision and the
adapter renders it.

Repository tests. These only run with TEST_DATABASE_URL, which CI does
set - so the job was red: copy and expunge seeded a 68-character digest
into a CHAR(64) column, and three others used fixed digests that collide
with the UNIQUE index on any re-run. A shared testBlobDigest() helper
fixes both; the suite now passes twice in a row against one database.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

services/protocols/internal/presentation/http/router.go

@@ -22,6 +22,11 @@ type Router struct {
 	reportUseCase *appusecase.IngestReportsUseCase
 	pingFunc      func() error
 	mtaStsMode    valueobject.MtaStsMode
+	// degradedFunc reports a condition that leaves the service running but
+	// not fit for production, the clearest case being a self-signed
+	// certificate standing in for one Traefik never issued
+	// (PLAN.md section 8.4).
+	degradedFunc func() error
 }
 
 func NewRouter(reportUseCase *appusecase.IngestReportsUseCase, pingFunc func() error) *Router {
@@ -62,6 +67,13 @@ func (r *Router) registerRoutes() {
 	r.mux.HandleFunc("/api/v1/reports/tlsrpt", r.handleTlsRptIngest)
 }
 
+// SetDegradedCheck registers the probe behind the "degraded" health state.
+// Returning an error means the process is serving but something an operator
+// must fix is wrong.
+func (r *Router) SetDegradedCheck(check func() error) {
+	r.degradedFunc = check
+}
+
 func (r *Router) handleHealth(w http.ResponseWriter, _ *http.Request) {
 	if r.pingFunc != nil {
 		if err := r.pingFunc(); err != nil {
@@ -69,6 +81,21 @@ func (r *Router) handleHealth(w http.ResponseWriter, _ *http.Request) {
 			return
 		}
 	}
+
+	// Degraded still answers 200: the container is alive and must not be
+	// restarted in a loop over a certificate problem a restart cannot fix.
+	// The distinction is carried in the body and the header so monitoring can
+	// alert on it (PLAN.md section 8.4).
+	if r.degradedFunc != nil {
+		if err := r.degradedFunc(); err != nil {
+			w.Header().Set("X-LambdaMail-Health", "degraded")
+			w.WriteHeader(http.StatusOK)
+			fmt.Fprintf(w, "DEGRADED: %v\n", err)
+			return
+		}
+	}
+
+	w.Header().Set("X-LambdaMail-Health", "ok")
 	w.WriteHeader(http.StatusOK)
 	w.Write([]byte("OK\n"))
 }

services/protocols/internal/presentation/http/router_test.go

@@ -2,6 +2,7 @@ package httppresentation
 
 import (
 	"bytes"
+	"errors"
 	"net/http"
 	"net/http/httptest"
 	"strings"
@@ -67,3 +68,38 @@ func TestRouter_MtaSts_Autoconfig_SecurityTxt_Ingest(t *testing.T) {
 		t.Errorf("dmarc ingest status = %d, body = %q", w.Code, w.Body.String())
 	}
 }
+
+// A certificate problem must be visible without taking the container down:
+// restarting cannot fix a certificate Traefik never issued, so /health stays
+// 200 and carries the degraded state instead (PLAN.md section 8.4).
+func TestRouter_HealthReportsDegradedWithoutFailing(t *testing.T) {
+	router := NewRouter(appusecase.NewIngestReportsUseCase(nil), func() error { return nil })
+	router.SetDegradedCheck(func() error { return errors.New("self-signed certificate in use") })
+
+	req := httptest.NewRequest(http.MethodGet, "/health", nil)
+	w := httptest.NewRecorder()
+	router.ServeHTTP(w, req)
+
+	if w.Code != http.StatusOK {
+		t.Errorf("degraded health status = %d, want 200 so the orchestrator does not restart-loop", w.Code)
+	}
+	if got := w.Header().Get("X-LambdaMail-Health"); got != "degraded" {
+		t.Errorf("X-LambdaMail-Health = %q, want degraded", got)
+	}
+	if !strings.Contains(w.Body.String(), "self-signed certificate in use") {
+		t.Errorf("the degraded body does not name the cause: %q", w.Body.String())
+	}
+}
+
+// A dependency that is actually down is a different state: that one must fail.
+func TestRouter_HealthFailsWhenDependencyIsDown(t *testing.T) {
+	router := NewRouter(appusecase.NewIngestReportsUseCase(nil), func() error { return errors.New("database unreachable") })
+
+	req := httptest.NewRequest(http.MethodGet, "/health", nil)
+	w := httptest.NewRecorder()
+	router.ServeHTTP(w, req)
+
+	if w.Code != http.StatusServiceUnavailable {
+		t.Errorf("health status = %d, want 503", w.Code)
+	}
+}