controller v0.73.0: DR recipe — emit secret-free customer+apps half in hub report

DR recipe slice (controller half), grounded in SPIKE-dr-recipe-2026-06-16. The
controller emitter is the BOUNDARY enforcement point: v1 ships an explicit
allowlist {catalog_ref, enabled, storage_bindings} and reads NOTHING from
AppConfig.Env, so no ENC:/token/password can leak. storage_bindings parsed from
the compose (${HDD_PATH}/${USERDATA_PATH} volume binds -> {container_path,
drive, subpath}).

Load-bearing tests: TestBuildAppRecipe_NoSecrets (synthetic-secret app -> none
leak) + TestBuildAppRecipe_AllowlistIsLoadBearing (red-proof companion) +
TestAppStorageBindings + TestBuildDRRecipeAppHalf. Red-proofed live: forcing the
emitter to dump Env makes the boundary test fail. recipe_version=1.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

controller/README.md

@@ -1248,6 +1248,7 @@ Periodic JSON push (default every 15 min) to the central felhom-hub service:
 - **Geo-restriction (always present, v0.70.0):** `geo_restriction` is always populated — `Enabled=false` with an empty country list when never configured — so the Hub always renders the geo section ("Inaktív" when off) instead of hiding it. `buildGeoRestrictionReport` in `internal/report/builder.go`.
 - **App telemetry** (v0.28.0+): Per-stack memory (current/avg/peak) and CPU averages from the last 15 minutes of metrics data, plus log scan results (error/warning counts with deduplicated issues). Only non-protected, deployed stacks are included. Backward-compatible: old Hub versions silently ignore this field.
 - **Controller telemetry** (v0.32.4+): The controller's own container (`felhom-controller`) is included as a special entry in the `app_telemetry` array. Its memory/CPU metrics come from the same metrics collector, and its log warnings/errors are scanned via `docker logs` using the same pipeline as app containers. This reuses all existing Hub telemetry infrastructure (memory trend charts, known issues, fleet aggregation) with zero Hub-side changes.
+- **DR recipe — customer + apps half (v0.73.0):** `dr_recipe` is the controller half of the secret-free reconstruction recipe (`SPIKE-dr-recipe-2026-06-16.md`) that complements escrow (keys) + PBS/restic (bytes). `BuildDRRecipeAppHalf` (`internal/report/dr_recipe.go`) emits `{recipe_version, customer{id,display,domain}, apps[]}`; each deployed, non-protected app contributes `AppRecipe{catalog_ref, enabled, storage_bindings}` where bindings are parsed from the compose (`${HDD_PATH}`/`${USERDATA_PATH}` volume binds → `{container_path, drive, subpath}`, e.g. romm → `felhom-flash:userdata/roms`). **THE BOUNDARY:** the emitter is the enforcement point — it ships an explicit allowlist of those three fields and reads NOTHING from `AppConfig.Env`, so no `ENC:`/token/password can leak (allowlist, not denylist → new fields excluded by default). The load-bearing `TestBuildAppRecipe_NoSecrets` + its red-proof companion live here. The hub assembles this half with the agent's storage/guest/PBS half into one customer recipe. `recipe_version=1`, ignore-unknown on read.
 
 Bearer token authentication, 3-attempt retry with 5-second backoff. Push status tracked via `PushStatus` struct (LastAttempt, LastSuccess, LastError, consecutive failures) — used by the monitoring page and alert system to show Hub connection health.
 

controller/internal/report/builder.go

@@ -161,6 +161,11 @@ func BuildReport(
 	// "Inaktív" hub-side.
 	r.GeoRestriction = buildGeoRestrictionReport(geoRestriction)
 
+	// DR recipe app-half — customer identity + per-app {catalog_ref, enabled, storage_bindings}.
+	// Allowlist-only (the boundary): NO env/secret fields. The hub assembles it with the agent half.
+	r.DRRecipe = BuildDRRecipeAppHalf(cfg.Customer.ID, cfg.Customer.Name, cfg.Customer.Domain,
+		stackMgr.GetStacks(), readComposeFile)
+
 	if debug && logger != nil {
 		logger.Printf("[DEBUG] [report] BuildReport: complete — containers=%d, health=%s, deployed=%d, available=%d, app_telemetry=%d",
 			r.Containers.Total, r.Health.Status, len(r.Stacks.Deployed), len(r.Stacks.Available), len(r.AppTelemetry))

controller/internal/report/dr_recipe.go

@@ -0,0 +1,161 @@
+package report
+
+import (
+	"bufio"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"gitea.dooplex.hu/admin/felhom-controller/internal/stacks"
+)
+
+// readComposeFile is the production composeReader: best-effort read of a stack's docker-compose.yml
+// (an unreadable file yields "" → no storage bindings for that app, never a failure).
+func readComposeFile(path string) string {
+	data, err := os.ReadFile(path)
+	if err != nil {
+		return ""
+	}
+	return string(data)
+}
+
+// DR recipe — the controller (customer + apps) HALF of the secret-free reconstruction recipe
+// (SPIKE-dr-recipe-2026-06-16). The recipe is the non-secret re-provision plan that complements escrow
+// (keys) + PBS/restic (bytes). This emitter is THE BOUNDARY ENFORCEMENT POINT: the controller is the
+// component that distinguishes secret from non-secret deploy fields (it encrypts ENC: secrets in
+// app.yaml), so v1 ships ONLY an explicit allowlist — {catalog_ref, enabled, storage_bindings} — and
+// NEVER touches AppConfig.Env. A new field cannot leak a secret because it is excluded by default
+// (allowlist, not denylist). TestBuildAppRecipe_NoSecrets is the load-bearing proof.
+//
+// recipe_version=1; read is ignore-unknown (forward-compat). The hub assembles this half with the
+// agent's storage/guest/PBS half into one customer recipe keyed by customer + recipe_version.
+const DRRecipeVersion = 1
+
+// DRRecipeAppHalf is the controller-emitted half.
+type DRRecipeAppHalf struct {
+	RecipeVersion int         `json:"recipe_version"`
+	Customer      DRCustomer  `json:"customer"`
+	Apps          []AppRecipe `json:"apps"`
+}
+
+// DRCustomer is the customer identity — public identifiers only.
+type DRCustomer struct {
+	ID      string `json:"id"`
+	Display string `json:"display"`
+	Domain  string `json:"domain"`
+}
+
+// AppRecipe is the v1 per-app allowlist. NOTHING from AppConfig.Env is emitted — the three fields below
+// are the entire surface, so no env secret (ENC: value, token, password) can ride along.
+type AppRecipe struct {
+	CatalogRef      string           `json:"catalog_ref"`
+	Enabled         bool             `json:"enabled"`
+	StorageBindings []StorageBinding `json:"storage_bindings"`
+}
+
+// StorageBinding names WHERE an app's data lives on a user-data drive: the container mount target, the
+// drive name, and the path under the drive. Identifiers/paths only — e.g. romm.library → felhom-flash
+// : userdata/roms. The hub correlates `drive` to the agent half's durable-id.
+type StorageBinding struct {
+	ContainerPath string `json:"container_path"`
+	Drive         string `json:"drive"`
+	Subpath       string `json:"subpath"`
+}
+
+// BuildDRRecipeAppHalf assembles the controller half: customer identity + an AppRecipe per DEPLOYED,
+// non-protected stack. composeReader returns a stack's docker-compose.yml content (seam: tests inject a
+// fake; production passes os.ReadFile-backed). Pure given the reader → unit-tested directly.
+func BuildDRRecipeAppHalf(custID, custName, domain string, all []stacks.Stack, composeReader func(path string) string) *DRRecipeAppHalf {
+	half := &DRRecipeAppHalf{
+		RecipeVersion: DRRecipeVersion,
+		Customer:      DRCustomer{ID: custID, Display: custName, Domain: domain},
+		Apps:          []AppRecipe{},
+	}
+	for _, s := range all {
+		if s.Protected || !s.Deployed {
+			continue
+		}
+		var composeYAML string
+		if composeReader != nil && s.ComposePath != "" {
+			composeYAML = composeReader(s.ComposePath)
+		}
+		half.Apps = append(half.Apps, buildAppRecipe(s, composeYAML))
+	}
+	return half
+}
+
+// buildAppRecipe constructs ONE app's recipe from the allowlist ONLY. This function deliberately reads
+// NOTHING from s.AppConfig.Env — that is the boundary. catalog_ref = the catalog slug (falls back to the
+// stack name for an orphaned app); storage_bindings parsed from the compose against the app's HDD_PATH.
+func buildAppRecipe(s stacks.Stack, composeYAML string) AppRecipe {
+	catalogRef := s.Meta.Slug
+	if catalogRef == "" {
+		catalogRef = s.Name
+	}
+	var hddPath string
+	if s.AppConfig != nil {
+		hddPath = s.AppConfig.Env["HDD_PATH"] // a path identifier, NOT a secret (secrets are ENC: in Env)
+	}
+	return AppRecipe{
+		CatalogRef:      catalogRef,
+		Enabled:         s.Deployed,
+		StorageBindings: appStorageBindings(composeYAML, hddPath),
+	}
+}
+
+// appStorageBindings parses a docker-compose.yml for volume binds that land on the app's user-data drive
+// (under HDD_PATH, or its USERDATA_PATH=<HDD_PATH>/userdata sibling) and returns {container_path, drive,
+// subpath}. Pure (operates on the compose text), so it is unit-tested without files. The drive name is
+// the basename of HDD_PATH (e.g. /mnt/felhom-drives/felhom-flash → "felhom-flash").
+func appStorageBindings(composeYAML, hddPath string) []StorageBinding {
+	if hddPath == "" || composeYAML == "" {
+		return []StorageBinding{}
+	}
+	cleanHDD := filepath.ToSlash(filepath.Clean(hddPath))
+	userdata := cleanHDD + "/userdata"
+	drive := filepath.Base(cleanHDD)
+
+	bindings := []StorageBinding{}
+	seen := map[string]bool{}
+	scanner := bufio.NewScanner(strings.NewReader(composeYAML))
+	inVolumes := false
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if strings.HasPrefix(line, "volumes:") {
+			inVolumes = true
+			continue
+		}
+		if inVolumes && !strings.HasPrefix(line, "-") && !strings.HasPrefix(line, "#") && line != "" {
+			inVolumes = false
+		}
+		if !inVolumes || !strings.HasPrefix(line, "- ") {
+			continue
+		}
+		mountStr := strings.Trim(strings.TrimPrefix(line, "- "), "\"'")
+		parts := strings.SplitN(mountStr, ":", 3)
+		if len(parts) < 2 {
+			continue
+		}
+		host := parts[0]
+		host = strings.ReplaceAll(host, "${USERDATA_PATH}", userdata)
+		host = strings.ReplaceAll(host, "${HDD_PATH}", cleanHDD)
+		// Also tolerate the un-braced $VAR form.
+		host = strings.ReplaceAll(host, "$USERDATA_PATH", userdata)
+		host = strings.ReplaceAll(host, "$HDD_PATH", cleanHDD)
+		host = filepath.ToSlash(filepath.Clean(host))
+
+		if host != cleanHDD && !strings.HasPrefix(host, cleanHDD+"/") {
+			continue // not on this drive
+		}
+		subpath := strings.TrimPrefix(host, cleanHDD)
+		subpath = strings.TrimPrefix(subpath, "/")
+		container := parts[1]
+		key := container + "\x00" + subpath
+		if seen[key] {
+			continue
+		}
+		seen[key] = true
+		bindings = append(bindings, StorageBinding{ContainerPath: container, Drive: drive, Subpath: subpath})
+	}
+	return bindings
+}

controller/internal/report/dr_recipe_test.go

@@ -0,0 +1,201 @@
+package report
+
+import (
+	"encoding/json"
+	"regexp"
+	"strings"
+	"testing"
+
+	"gitea.dooplex.hu/admin/felhom-controller/internal/stacks"
+)
+
+// secretNameRe matches any JSON key that smells like a credential — mirrors the agent half.
+var secretNameRe = regexp.MustCompile(`(?i)(password|secret|token|hash|passphrase|api[_-]?key|\bkey\b|enc:)`)
+
+// rommCompose is a realistic catalog compose: user-data binds via ${USERDATA_PATH}/${HDD_PATH}, plus a
+// secret-laden environment section (which must NEVER reach the recipe — bindings come from volumes only).
+const rommCompose = `services:
+  romm:
+    image: romm:latest
+    environment:
+      - DB_PASSWORD=${DB_PASSWORD}
+      - IGDB_CLIENT_SECRET=${IGDB_CLIENT_SECRET}
+    volumes:
+      - ${USERDATA_PATH}/roms:/roms
+      - ${HDD_PATH}/appdata/romm/resources:/romm/resources
+      - romm_redis_data:/data
+volumes:
+  romm_redis_data:
+`
+
+// secretLadenStack builds a romm Stack whose persisted Env carries synthetic secrets (an ENC: value and
+// a token-shaped value) — exactly what the recipe must keep out.
+func secretLadenStack() stacks.Stack {
+	return stacks.Stack{
+		Name:        "romm",
+		Meta:        stacks.Metadata{Slug: "romm", DisplayName: "RomM"},
+		ComposePath: "/stacks/romm/docker-compose.yml",
+		Deployed:    true,
+		AppConfig: &stacks.AppConfig{
+			Deployed: true,
+			Env: map[string]string{
+				"HDD_PATH":           "/mnt/felhom-drives/felhom-flash",
+				"DB_PASSWORD":        "ENC:U2FsdGVkX1+DEADBEEFsecret==",
+				"IGDB_CLIENT_SECRET": "tok_live_SUPERSECRET_must_not_leak",
+				"SECRET_KEY":         "ENC:another_encrypted_blob",
+			},
+		},
+	}
+}
+
+// TestBuildAppRecipe_NoSecrets is THE load-bearing boundary test: the emitted AppRecipe for an app whose
+// deploy config contains secrets (ENC: + token-shaped) carries NONE of those values and NO
+// credential-shaped key — only the {catalog_ref, enabled, storage_bindings} allowlist.
+func TestBuildAppRecipe_NoSecrets(t *testing.T) {
+	s := secretLadenStack()
+	rec := buildAppRecipe(s, rommCompose)
+
+	b, err := json.Marshal(rec)
+	if err != nil {
+		t.Fatal(err)
+	}
+	out := string(b)
+
+	// (1) none of the secret VALUES survived.
+	for _, leak := range []string{"ENC:", "U2FsdGVkX1", "DEADBEEF", "tok_live_SUPERSECRET_must_not_leak", "another_encrypted_blob"} {
+		if strings.Contains(out, leak) {
+			t.Errorf("SECRET LEAK: emitted recipe contains %q\n recipe: %s", leak, out)
+		}
+	}
+	// (2) no credential-shaped KEY survived (DB_PASSWORD / SECRET_KEY / IGDB_CLIENT_SECRET as keys).
+	assertNoSecretKeys(t, b)
+
+	// (3) positive: we DID emit the allowlisted facts (not a vacuous pass).
+	if rec.CatalogRef != "romm" || !rec.Enabled {
+		t.Errorf("expected catalog_ref=romm enabled=true, got %+v", rec)
+	}
+	if len(rec.StorageBindings) != 2 {
+		t.Fatalf("expected 2 storage bindings (roms + resources), got %+v", rec.StorageBindings)
+	}
+}
+
+// TestBuildAppRecipe_AllowlistIsLoadBearing is the companion (red-proof of the boundary test): a NAIVE
+// emitter that dumps AppConfig.Env (i.e. the allowlist guard removed) WOULD leak the token — proving the
+// no-secrets assertion above is real, not vacuous. The production emitter must NOT leak it.
+func TestBuildAppRecipe_AllowlistIsLoadBearing(t *testing.T) {
+	s := secretLadenStack()
+	const token = "tok_live_SUPERSECRET_must_not_leak"
+
+	// The "guard removed" shape — dumping Env alongside the app. This is what the boundary forbids.
+	unsafe, _ := json.Marshal(map[string]any{"catalog_ref": s.Meta.Slug, "env": s.AppConfig.Env})
+	if !strings.Contains(string(unsafe), token) {
+		t.Fatal("companion is broken: the unsafe (guard-removed) shape should contain the secret token")
+	}
+
+	// The REAL emitter must keep it out — same fixture, allowlist intact.
+	real, _ := json.Marshal(buildAppRecipe(s, rommCompose))
+	if strings.Contains(string(real), token) {
+		t.Fatalf("BOUNDARY VIOLATION: production emitter leaked the token: %s", real)
+	}
+}
+
+func TestAppStorageBindings(t *testing.T) {
+	got := appStorageBindings(rommCompose, "/mnt/felhom-drives/felhom-flash")
+	want := map[string]StorageBinding{
+		"/roms":           {ContainerPath: "/roms", Drive: "felhom-flash", Subpath: "userdata/roms"},
+		"/romm/resources": {ContainerPath: "/romm/resources", Drive: "felhom-flash", Subpath: "appdata/romm/resources"},
+	}
+	if len(got) != len(want) {
+		t.Fatalf("got %d bindings, want %d: %+v", len(got), len(want), got)
+	}
+	for _, b := range got {
+		w, ok := want[b.ContainerPath]
+		if !ok || b != w {
+			t.Errorf("binding %+v unexpected (want %+v)", b, w)
+		}
+	}
+	// The named volume (romm_redis_data) is NOT a drive bind → excluded.
+	for _, b := range got {
+		if strings.Contains(b.Subpath, "redis") {
+			t.Errorf("named volume leaked into bindings: %+v", b)
+		}
+	}
+}
+
+// TestAppStorageBindings_NoHDD: an app with no HDD_PATH (rootfs-only) yields no bindings, non-nil slice.
+func TestAppStorageBindings_NoHDD(t *testing.T) {
+	got := appStorageBindings(rommCompose, "")
+	if got == nil || len(got) != 0 {
+		t.Errorf("no HDD_PATH should yield empty (non-nil) bindings, got %+v", got)
+	}
+}
+
+func TestBuildDRRecipeAppHalf(t *testing.T) {
+	reader := func(path string) string {
+		if path == "/stacks/romm/docker-compose.yml" {
+			return rommCompose
+		}
+		return ""
+	}
+	all := []stacks.Stack{
+		secretLadenStack(),
+		{Name: "traefik", Protected: true, Deployed: true},                                  // protected → excluded
+		{Name: "vikunja", Meta: stacks.Metadata{Slug: "vikunja"}, Deployed: false},          // not deployed → excluded
+		{Name: "actualbudget", Meta: stacks.Metadata{Slug: "actualbudget"}, Deployed: true}, // rootfs app, no compose path
+	}
+	half := BuildDRRecipeAppHalf("cust-demo", "Demo Customer", "demo-felhom.eu", all, reader)
+
+	if half.RecipeVersion != 1 {
+		t.Errorf("recipe_version=%d want 1", half.RecipeVersion)
+	}
+	if half.Customer.ID != "cust-demo" || half.Customer.Display != "Demo Customer" || half.Customer.Domain != "demo-felhom.eu" {
+		t.Errorf("customer = %+v", half.Customer)
+	}
+	// Only romm + actualbudget (deployed, non-protected). traefik (protected) + vikunja (not deployed) out.
+	if len(half.Apps) != 2 {
+		t.Fatalf("expected 2 apps, got %d: %+v", len(half.Apps), half.Apps)
+	}
+	byRef := map[string]AppRecipe{}
+	for _, a := range half.Apps {
+		byRef[a.CatalogRef] = a
+	}
+	if r, ok := byRef["romm"]; !ok || len(r.StorageBindings) != 2 {
+		t.Errorf("romm recipe wrong: %+v", r)
+	}
+	if r, ok := byRef["actualbudget"]; !ok || len(r.StorageBindings) != 0 {
+		t.Errorf("actualbudget (rootfs) should have 0 bindings: %+v", r)
+	}
+
+	// Whole-half no-secrets sweep.
+	b, _ := json.Marshal(half)
+	assertNoSecretKeys(t, b)
+	if strings.Contains(string(b), "tok_live_SUPERSECRET_must_not_leak") {
+		t.Errorf("SECRET LEAK in assembled app-half: %s", b)
+	}
+}
+
+// assertNoSecretKeys walks decoded JSON and fails on any object key matching secretNameRe.
+func assertNoSecretKeys(t *testing.T, jsonBytes []byte) {
+	t.Helper()
+	var v any
+	if err := json.Unmarshal(jsonBytes, &v); err != nil {
+		t.Fatal(err)
+	}
+	var walk func(prefix string, node any)
+	walk = func(prefix string, node any) {
+		switch n := node.(type) {
+		case map[string]any:
+			for k, child := range n {
+				if secretNameRe.MatchString(k) {
+					t.Errorf("secret-shaped key %q at %s — the recipe must carry no credential field", k, prefix)
+				}
+				walk(prefix+"."+k, child)
+			}
+		case []any:
+			for _, child := range n {
+				walk(prefix, child)
+			}
+		}
+	}
+	walk("<root>", v)
+}

controller/internal/report/types.go

@@ -24,6 +24,10 @@ type Report struct {
 	Stacks            StacksReport     `json:"stacks"`
 	AppTelemetry      []AppTelemetry          `json:"app_telemetry,omitempty"`
 	GeoRestriction    *GeoRestrictionReport   `json:"geo_restriction,omitempty"`
+
+	// DR recipe — the controller (customer + apps) half of the secret-free reconstruction recipe
+	// (SPIKE-dr-recipe-2026-06-16). The hub assembles it with the agent's storage/guest/PBS half.
+	DRRecipe *DRRecipeAppHalf `json:"dr_recipe,omitempty"`
 }
 
 // SystemReport holds host-level system info.