test(ingestion): OIDC discovery 502s then 200 against a real issuer (#83)

The retry landed in 9a232ad but was tested only through startupretry with
fake funcs; nothing exercised chassisauth.NewJWTValidator's real 502 path.
Extracts the main() closure into buildJWTValidator (behaviour unchanged)
and tests it against an httptest issuer: 3x502 then 200 succeeds with 4
discoveries; a persistent 502 stops at exactly the 8-attempt budget with
the last error; cancelling ctx mid-backoff returns context.Canceled.
Flip-verified: Attempts=1 turns all three red.

Lane: risk/medium issue, committed to main because the behaviour change
already shipped in 9a232ad; this is a test plus a pure extraction.

Closes #83

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

ingestion/cmd/server/jwt_validator.go

@@ -0,0 +1,25 @@
+package main
+
+import (
+	"context"
+	"log/slog"
+
+	chassisauth "git.d-ma.be/mathias/mcp-chassis/auth"
+
+	"github.com/mathiasbq/hyperguild/ingestion/internal/startupretry"
+)
+
+// buildJWTValidator runs OIDC discovery under policy, so a 502 from an auth
+// upstream that is not serving yet on a cold boot is retried instead of
+// exiting (#83). It honours ctx between attempts.
+func buildJWTValidator(
+	ctx context.Context, logger *slog.Logger, issuer, audience string, policy startupretry.Policy,
+) (*chassisauth.JWTValidator, error) {
+	return startupretry.Do(ctx, policy, func(ctx context.Context) (*chassisauth.JWTValidator, error) {
+		v, err := chassisauth.NewJWTValidator(ctx, issuer, audience)
+		if err != nil {
+			logger.Warn("oidc discovery not ready, retrying", "err", err)
+		}
+		return v, err
+	})
+}

ingestion/cmd/server/jwt_validator_test.go

@@ -0,0 +1,86 @@
+package main
+
+import (
+	"context"
+	"io"
+	"log/slog"
+	"net/http"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/mathiasbq/hyperguild/ingestion/internal/startupretry"
+)
+
+// fakeIssuer answers OIDC discovery with 502 for the first fail requests, as
+// ingress does while dex is not yet serving on a cold boot (#83), then with a
+// real discovery document and an empty JWKS.
+func fakeIssuer(t *testing.T, fail int32) (*httptest.Server, *atomic.Int32) {
+	t.Helper()
+	var discoveries atomic.Int32
+	mux := http.NewServeMux()
+	var srv *httptest.Server
+	mux.HandleFunc("/.well-known/openid-configuration", func(w http.ResponseWriter, _ *http.Request) {
+		if discoveries.Add(1) <= fail {
+			w.WriteHeader(http.StatusBadGateway)
+			return
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = io.WriteString(w, `{"issuer":"`+srv.URL+`","jwks_uri":"`+srv.URL+`/keys"}`)
+	})
+	mux.HandleFunc("/keys", func(w http.ResponseWriter, _ *http.Request) {
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = io.WriteString(w, `{"keys":[]}`)
+	})
+	srv = httptest.NewServer(mux)
+	t.Cleanup(srv.Close)
+	return srv, &discoveries
+}
+
+var fastPolicy = startupretry.Policy{Attempts: 8, Base: time.Millisecond, Max: 5 * time.Millisecond}
+
+func TestBuildJWTValidatorRetriesDiscovery502sThenSucceeds(t *testing.T) {
+	srv, discoveries := fakeIssuer(t, 3)
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	v, err := buildJWTValidator(ctx, slog.New(slog.DiscardHandler), srv.URL, "brain", fastPolicy)
+	require.NoError(t, err)
+	assert.NotNil(t, v)
+	assert.Equal(t, int32(4), discoveries.Load(), "three 502s, then the successful discovery")
+}
+
+func TestBuildJWTValidatorGivesUpAfterTheBudgetWithTheLastError(t *testing.T) {
+	srv, discoveries := fakeIssuer(t, 1000)
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	_, err := buildJWTValidator(ctx, slog.New(slog.DiscardHandler), srv.URL, "brain", fastPolicy)
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "status 502")
+	assert.Equal(t, int32(8), discoveries.Load(), "bounded: exactly the attempt budget")
+}
+
+func TestBuildJWTValidatorStopsRetryingWhenTheContextIsCancelled(t *testing.T) {
+	srv, _ := fakeIssuer(t, 1000)
+	ctx, cancel := context.WithCancel(context.Background())
+	slow := startupretry.Policy{Attempts: 8, Base: time.Hour, Max: time.Hour}
+
+	done := make(chan error, 1)
+	go func() {
+		_, err := buildJWTValidator(ctx, slog.New(slog.DiscardHandler), srv.URL, "brain", slow)
+		done <- err
+	}()
+	time.Sleep(50 * time.Millisecond)
+	cancel()
+	select {
+	case err := <-done:
+		assert.ErrorIs(t, err, context.Canceled)
+	case <-time.After(3 * time.Second):
+		t.Fatal("did not return after the context was cancelled")
+	}
+}

ingestion/cmd/server/main.go

@@ -381,15 +381,9 @@ func main() {
 	//
 	// The budget stays bounded: retrying forever would turn a misconfiguration
 	// into a process that looks alive and never serves.
-	jwtValidator, err := startupretry.Do(ctx,
-		startupretry.Policy{Attempts: 8, Base: time.Second, Max: 30 * time.Second},
-		func(ctx context.Context) (*chassisauth.JWTValidator, error) {
-			v, err := chassisauth.NewJWTValidator(ctx, os.Getenv("DEX_ISSUER_URL"), os.Getenv("MCP_AUDIENCE"))
-			if err != nil {
-				logger.Warn("oidc discovery not ready, retrying", "err", err)
-			}
-			return v, err
-		})
+	jwtValidator, err := buildJWTValidator(ctx, logger,
+		os.Getenv("DEX_ISSUER_URL"), os.Getenv("MCP_AUDIENCE"),
+		startupretry.Policy{Attempts: 8, Base: time.Second, Max: 30 * time.Second})
 	if err != nil {
 		logger.Error("build jwt validator", "err", err)
 		os.Exit(1)