| 1 | package edge |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "context" |
| 6 | "encoding/hex" |
| 7 | "log" |
| 8 | "strings" |
| 9 | "testing" |
| 10 | "time" |
| 11 | |
| 12 | "github.com/carlosframework/platform/internal/release" |
| 13 | ) |
| 14 | |
| 15 | // captureLog redirects the standard logger for the duration of a test — |
| 16 | // synchronous callers only (mintMeshKey, loadMeshKeys direct calls here are |
| 17 | // never run concurrently with the read), same shape certsync_test.go uses. |
| 18 | func captureLog(t *testing.T) *bytes.Buffer { |
| 19 | t.Helper() |
| 20 | var buf bytes.Buffer |
| 21 | prev := log.Writer() |
| 22 | log.SetOutput(&buf) |
| 23 | t.Cleanup(func() { log.SetOutput(prev) }) |
| 24 | return &buf |
| 25 | } |
| 26 | |
| 27 | // meshTestStore is a fresh directory-backed store, the same DirStore every |
| 28 | // other edge test in this package uses in place of live S3. |
| 29 | func meshTestStore(t *testing.T) release.Store { |
| 30 | t.Helper() |
| 31 | return release.NewDirStore(t.TempDir()) |
| 32 | } |
| 33 | |
| 34 | func TestMintMeshKeyIsCreateOnly(t *testing.T) { |
| 35 | s := meshTestStore(t) |
| 36 | if err := mintMeshKey(context.Background(), s); err != nil { |
| 37 | t.Fatal(err) |
| 38 | } |
| 39 | first, _, _ := s.Get(context.Background(), meshKeyKey) |
| 40 | if err := mintMeshKey(context.Background(), s); err != nil { |
| 41 | t.Fatalf("second mint must be success-by-other: %v", err) |
| 42 | } |
| 43 | second, _, _ := s.Get(context.Background(), meshKeyKey) |
| 44 | if !bytes.Equal(first, second) { |
| 45 | t.Fatal("second mint replaced the key") |
| 46 | } |
| 47 | } |
| 48 | |
| 49 | // The exported wrapper is the seam internal/host's housekeeper duty calls |
| 50 | // through — prove it actually reaches the same key, not a copy of the logic. |
| 51 | func TestMintMeshKeyExportedWrapperMintsTheSameKey(t *testing.T) { |
| 52 | s := meshTestStore(t) |
| 53 | if err := MintMeshKey(context.Background(), s); err != nil { |
| 54 | t.Fatal(err) |
| 55 | } |
| 56 | body, _, err := s.Get(context.Background(), meshKeyKey) |
| 57 | if err != nil { |
| 58 | t.Fatalf("get %s: %v", meshKeyKey, err) |
| 59 | } |
| 60 | if len(body) != meshKeyBytes*2 { // hex-encoded |
| 61 | t.Fatalf("mesh key body length = %d, want %d (hex of %d random bytes)", len(body), meshKeyBytes*2, meshKeyBytes) |
| 62 | } |
| 63 | } |
| 64 | |
| 65 | // A lost race is the steady state of every tick but the first, forever — it |
| 66 | // must not log anything, matching housekeeper.go's own silence on election |
| 67 | // contention. Only the first, real mint gets a line. |
| 68 | func TestMintMeshKeySecondMintIsSilent(t *testing.T) { |
| 69 | buf := captureLog(t) |
| 70 | s := meshTestStore(t) |
| 71 | if err := mintMeshKey(context.Background(), s); err != nil { |
| 72 | t.Fatal(err) |
| 73 | } |
| 74 | first := buf.String() |
| 75 | if !strings.Contains(first, "minted") { |
| 76 | t.Fatalf("first mint log = %q, want it to mention minting", first) |
| 77 | } |
| 78 | buf.Reset() |
| 79 | if err := mintMeshKey(context.Background(), s); err != nil { |
| 80 | t.Fatalf("second mint must be success-by-other: %v", err) |
| 81 | } |
| 82 | if got := buf.String(); got != "" { |
| 83 | t.Fatalf("second mint (a lost CAS race) logged %q, want silence", got) |
| 84 | } |
| 85 | } |
| 86 | |
| 87 | func TestHopSignatureRoundTripsAndBoundsSkew(t *testing.T) { |
| 88 | key := []byte("0123456789abcdef0123456789abcdef") |
| 89 | now := time.Unix(1_766_000_000, 0) |
| 90 | h := signHop(key, "app.example.com", now) |
| 91 | if !verifyHop(key, "app.example.com", h, now.Add(29*time.Second)) { |
| 92 | t.Fatal("within skew must verify") |
| 93 | } |
| 94 | if verifyHop(key, "app.example.com", h, now.Add(31*time.Second)) { |
| 95 | t.Fatal("past skew must refuse") |
| 96 | } |
| 97 | if verifyHop(key, "other.example.com", h, now) { |
| 98 | t.Fatal("host is bound into the mac") |
| 99 | } |
| 100 | if verifyHop([]byte("wrong-key-wrong-key-wrong-key-00"), "app.example.com", h, now) { |
| 101 | t.Fatal("wrong key must refuse") |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | // Skew is bounded in both directions — a clock that runs fast must be |
| 106 | // refused exactly like one that runs slow. |
| 107 | func TestHopSignatureRefusesFutureSkewToo(t *testing.T) { |
| 108 | key := []byte("0123456789abcdef0123456789abcdef") |
| 109 | now := time.Unix(1_766_000_000, 0) |
| 110 | h := signHop(key, "app.example.com", now) |
| 111 | if verifyHop(key, "app.example.com", h, now.Add(-31*time.Second)) { |
| 112 | t.Fatal("verifying 31s before the signature's timestamp must refuse") |
| 113 | } |
| 114 | if !verifyHop(key, "app.example.com", h, now.Add(-29*time.Second)) { |
| 115 | t.Fatal("verifying 29s before the signature's timestamp must still verify") |
| 116 | } |
| 117 | } |
| 118 | |
| 119 | // A malformed header (no ":", a non-numeric timestamp, non-hex mac) must |
| 120 | // refuse, never panic — it arrives off the wire from a hop this box does not |
| 121 | // control. |
| 122 | func TestVerifyHopRefusesMalformedHeaders(t *testing.T) { |
| 123 | key := []byte("0123456789abcdef0123456789abcdef") |
| 124 | now := time.Unix(1_766_000_000, 0) |
| 125 | for _, header := range []string{ |
| 126 | "", |
| 127 | "nocolon", |
| 128 | "notanumber:deadbeef", |
| 129 | "1766000000:not-hex-zzzz", |
| 130 | "1766000000:", |
| 131 | } { |
| 132 | if verifyHop(key, "app.example.com", header, now) { |
| 133 | t.Fatalf("verifyHop(%q) = true, want false", header) |
| 134 | } |
| 135 | } |
| 136 | } |
| 137 | |
| 138 | // The MAC is over the CANONICAL host, so a signer's and a verifier's |
| 139 | // differing-but-equivalent forms of the same host must both verify against |
| 140 | // the same signature: Task 4 signs one form (e.g. a home-map lookup key, |
| 141 | // unlikely to carry a port), Task 3 verifies a literal net/http request Host |
| 142 | // (may carry a port, whatever case and trailing dot the client sent) — |
| 143 | // divergence here would be spurious refusals presenting as regional |
| 144 | // flakiness, not a security property either side intends. |
| 145 | func TestHopMACCanonicalizesHost(t *testing.T) { |
| 146 | key := []byte("0123456789abcdef0123456789abcdef") |
| 147 | now := time.Unix(1_766_000_000, 0) |
| 148 | canonical := signHop(key, "app.example.com", now) |
| 149 | |
| 150 | // A signature made for the canonical host verifies against every |
| 151 | // equivalent form a verifier might see the request host in. |
| 152 | for _, host := range []string{ |
| 153 | "app.example.com", |
| 154 | "APP.EXAMPLE.COM", |
| 155 | "App.Example.Com", |
| 156 | "app.example.com.", |
| 157 | "app.example.com:443", |
| 158 | "app.example.com.:8443", |
| 159 | "APP.EXAMPLE.COM.:443", |
| 160 | } { |
| 161 | if !verifyHop(key, host, canonical, now) { |
| 162 | t.Fatalf("verifyHop(%q) against a signature made for the canonical host = false, want true", host) |
| 163 | } |
| 164 | } |
| 165 | |
| 166 | // And signing under any of those same equivalent forms produces a |
| 167 | // signature that verifies against the canonical form too — the |
| 168 | // canonicalization is symmetric, not a one-sided allowance. |
| 169 | for _, host := range []string{"APP.EXAMPLE.COM", "app.example.com.", "app.example.com:9443"} { |
| 170 | h := signHop(key, host, now) |
| 171 | if !verifyHop(key, "app.example.com", h, now) { |
| 172 | t.Fatalf("a signature made for host %q must verify against the canonical form", host) |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | // A genuinely different host must still refuse. |
| 177 | if verifyHop(key, "other.example.com", canonical, now) { |
| 178 | t.Fatal("a different host must still refuse — canonicalization must not collapse distinct hosts") |
| 179 | } |
| 180 | } |
| 181 | |
| 182 | // ---- rotation dual-accept (plan's global constraint: the READ side ships |
| 183 | // now — verifyHopEitherKey and signingKey are what Task 3/4's request path |
| 184 | // calls once it has loaded both mirrored keys) -------------------------------- |
| 185 | |
| 186 | func TestSignHopPrefersNextKeyWhenPresent(t *testing.T) { |
| 187 | current := []byte("current-key-current-key-current") |
| 188 | next := []byte("next-key-next-key-next-key-next") |
| 189 | if got := signingKey(current, next); !bytes.Equal(got, next) { |
| 190 | t.Fatal("signingKey must prefer .next when it exists") |
| 191 | } |
| 192 | if got := signingKey(current, nil); !bytes.Equal(got, current) { |
| 193 | t.Fatal("signingKey must fall back to current when there is no .next") |
| 194 | } |
| 195 | } |
| 196 | |
| 197 | func TestVerifyHopEitherKeyAcceptsSignaturesUnderCurrentOrNext(t *testing.T) { |
| 198 | current := []byte("current-key-current-key-current") |
| 199 | next := []byte("next-key-next-key-next-key-next-") |
| 200 | wrong := []byte("wrong-key-wrong-key-wrong-key-00") |
| 201 | now := time.Unix(1_766_000_000, 0) |
| 202 | host := "app.example.com" |
| 203 | |
| 204 | // sign with .next -> verifies |
| 205 | hNext := signHop(signingKey(current, next), host, now) |
| 206 | if !verifyHopEitherKey(current, next, host, hNext, now) { |
| 207 | t.Fatal("a signature made with .next must verify while .next exists") |
| 208 | } |
| 209 | |
| 210 | // sign with old key while .next exists -> still verifies (dual-accept) |
| 211 | hOld := signHop(current, host, now) |
| 212 | if !verifyHopEitherKey(current, next, host, hOld, now) { |
| 213 | t.Fatal("a signature made with the pre-rotation key must still verify while .next exists") |
| 214 | } |
| 215 | |
| 216 | // wrong key -> refuses |
| 217 | hWrong := signHop(wrong, host, now) |
| 218 | if verifyHopEitherKey(current, next, host, hWrong, now) { |
| 219 | t.Fatal("a signature made with neither key must refuse") |
| 220 | } |
| 221 | |
| 222 | // no rotation in flight (next is nil): only current verifies. |
| 223 | if !verifyHopEitherKey(current, nil, host, hOld, now) { |
| 224 | t.Fatal("current key must verify with no rotation in flight") |
| 225 | } |
| 226 | if verifyHopEitherKey(current, nil, host, hNext, now) { |
| 227 | t.Fatal("a .next signature must not verify once .next is gone") |
| 228 | } |
| 229 | } |
| 230 | |
| 231 | func TestLoadMeshKeysReadsBothMirroredObjects(t *testing.T) { |
| 232 | ctx := context.Background() |
| 233 | s := meshTestStore(t) |
| 234 | |
| 235 | // Before the housekeeper's first tick: neither object exists, and that |
| 236 | // is quiet — not an error. |
| 237 | current, next, err := loadMeshKeys(ctx, s) |
| 238 | if err != nil { |
| 239 | t.Fatalf("loadMeshKeys before minting: %v", err) |
| 240 | } |
| 241 | if current != nil || next != nil { |
| 242 | t.Fatalf("loadMeshKeys before minting = (%x, %x), want (nil, nil)", current, next) |
| 243 | } |
| 244 | |
| 245 | if err := mintMeshKey(ctx, s); err != nil { |
| 246 | t.Fatal(err) |
| 247 | } |
| 248 | current, next, err = loadMeshKeys(ctx, s) |
| 249 | if err != nil { |
| 250 | t.Fatalf("loadMeshKeys after minting: %v", err) |
| 251 | } |
| 252 | if len(current) != meshKeyBytes { |
| 253 | t.Fatalf("current key length = %d, want %d raw bytes", len(current), meshKeyBytes) |
| 254 | } |
| 255 | if next != nil { |
| 256 | t.Fatal("next key must be nil outside a rotation window") |
| 257 | } |
| 258 | |
| 259 | // A rotation in flight: .next mirrored alongside the current key. |
| 260 | nextBody := []byte(hex.EncodeToString(bytes.Repeat([]byte{0xab}, meshKeyBytes))) |
| 261 | if _, err := s.Put(ctx, meshKeyNextKey, nextBody, ""); err != nil { |
| 262 | t.Fatal(err) |
| 263 | } |
| 264 | current, next, err = loadMeshKeys(ctx, s) |
| 265 | if err != nil { |
| 266 | t.Fatalf("loadMeshKeys during rotation: %v", err) |
| 267 | } |
| 268 | if len(current) != meshKeyBytes || len(next) != meshKeyBytes { |
| 269 | t.Fatalf("key lengths = (%d, %d), want both %d", len(current), len(next), meshKeyBytes) |
| 270 | } |
| 271 | } |
| 272 | |
| 273 | // A malformed .next — hand-written during a rotation, so a bad upload is a |
| 274 | // real possibility — must not take the live current key down with it: the |
| 275 | // safe response is to treat it as no rotation in flight (loudly logged), |
| 276 | // never to refuse every hop fleet-wide over one bad object. |
| 277 | func TestLoadMeshKeysToleratesAnUnreadableNext(t *testing.T) { |
| 278 | ctx := context.Background() |
| 279 | s := meshTestStore(t) |
| 280 | if err := mintMeshKey(ctx, s); err != nil { |
| 281 | t.Fatal(err) |
| 282 | } |
| 283 | if _, err := s.Put(ctx, meshKeyNextKey, []byte("not-valid-hex-zzzz"), ""); err != nil { |
| 284 | t.Fatal(err) |
| 285 | } |
| 286 | |
| 287 | buf := captureLog(t) |
| 288 | current, next, err := loadMeshKeys(ctx, s) |
| 289 | if err != nil { |
| 290 | t.Fatalf("loadMeshKeys with a malformed .next must not fail: %v", err) |
| 291 | } |
| 292 | if len(current) != meshKeyBytes { |
| 293 | t.Fatalf("current key length = %d, want %d — a malformed .next took the live key down with it", len(current), meshKeyBytes) |
| 294 | } |
| 295 | if next != nil { |
| 296 | t.Fatal("a malformed .next must be treated as absent (no rotation in flight), not surfaced") |
| 297 | } |
| 298 | if got := buf.String(); !strings.Contains(got, meshKeyNextKey) { |
| 299 | t.Fatalf("log = %q, want a loud mention of %s — this is an operator mistake worth paging on", got, meshKeyNextKey) |
| 300 | } |
| 301 | |
| 302 | // And a signature made under the still-good current key verifies fine — |
| 303 | // the tolerance is not merely "does not error", it's "hop auth keeps |
| 304 | // working". |
| 305 | now := time.Unix(1_766_000_000, 0) |
| 306 | h := signHop(current, "app.example.com", now) |
| 307 | if !verifyHopEitherKey(current, next, "app.example.com", h, now) { |
| 308 | t.Fatal("hop auth must keep working off the current key despite the malformed .next") |
| 309 | } |
| 310 | } |
| 311 | |
| 312 | // ---- the key half of the home-map mirror ------------------------------------ |
| 313 | // |
| 314 | // Task 3 folds the mesh-key read into the home map's 30s poll rather than |
| 315 | // giving it a loop of its own: one list-and-get pass keeps both halves of a |
| 316 | // hop's answer (who homes this host, and what key proves the hop) from ever |
| 317 | // being read from two different moments in the bucket's life. |
| 318 | |
| 319 | func TestHomeMapRefreshMirrorsTheMeshKeys(t *testing.T) { |
| 320 | ctx := context.Background() |
| 321 | s := meshTestStore(t) |
| 322 | m := newHomeMap(s, "annie") |
| 323 | |
| 324 | // Before the housekeeper's first tick: no key, and every hop fails |
| 325 | // closed — refreshing is still not an error. |
| 326 | if err := m.Refresh(ctx); err != nil { |
| 327 | t.Fatalf("refresh before minting: %v", err) |
| 328 | } |
| 329 | if cur, next := m.meshKeys(); cur != nil || next != nil { |
| 330 | t.Fatalf("keys before minting = (%x, %x), want (nil, nil)", cur, next) |
| 331 | } |
| 332 | |
| 333 | if err := mintMeshKey(ctx, s); err != nil { |
| 334 | t.Fatal(err) |
| 335 | } |
| 336 | if err := m.Refresh(ctx); err != nil { |
| 337 | t.Fatalf("refresh after minting: %v", err) |
| 338 | } |
| 339 | cur, next := m.meshKeys() |
| 340 | if len(cur) != meshKeyBytes || next != nil { |
| 341 | t.Fatalf("keys after minting = (%d bytes, %v), want the current key alone", len(cur), next) |
| 342 | } |
| 343 | |
| 344 | // A rotation in flight: dual-accept means a signature under EITHER key is |
| 345 | // good, which is what the request path calls (verifyHopEitherKey, never |
| 346 | // the single-key verifyHop). |
| 347 | nextBody := []byte(hex.EncodeToString(bytes.Repeat([]byte{0xcd}, meshKeyBytes))) |
| 348 | if _, err := s.Put(ctx, meshKeyNextKey, nextBody, ""); err != nil { |
| 349 | t.Fatal(err) |
| 350 | } |
| 351 | if err := m.Refresh(ctx); err != nil { |
| 352 | t.Fatalf("refresh during rotation: %v", err) |
| 353 | } |
| 354 | cur, next = m.meshKeys() |
| 355 | if len(cur) != meshKeyBytes || len(next) != meshKeyBytes { |
| 356 | t.Fatalf("keys during rotation = (%d, %d), want both mirrored", len(cur), len(next)) |
| 357 | } |
| 358 | now := time.Now() |
| 359 | for _, k := range [][]byte{cur, next} { |
| 360 | if !verifyHopEitherKey(cur, next, "app.example.com", signHop(k, "app.example.com", now), now) { |
| 361 | t.Fatal("a hop signed under either mirrored key must verify while both are mirrored") |
| 362 | } |
| 363 | } |
| 364 | } |
| 365 | |
| 366 | // The keys ride the same last-good posture as the host fold: a bucket blip |
| 367 | // must not disarm hop verification fleet-wide. |
| 368 | func TestHomeMapKeepsLastGoodMeshKeysWhenTheBucketIsDown(t *testing.T) { |
| 369 | ctx := context.Background() |
| 370 | inner := meshTestStore(t) |
| 371 | s := &listFailingStore{Store: inner} |
| 372 | if err := mintMeshKey(ctx, inner); err != nil { |
| 373 | t.Fatal(err) |
| 374 | } |
| 375 | m := newHomeMap(s, "annie") |
| 376 | if err := m.Refresh(ctx); err != nil { |
| 377 | t.Fatalf("refresh: %v", err) |
| 378 | } |
| 379 | good, _ := m.meshKeys() |
| 380 | |
| 381 | s.setFail(true) |
| 382 | if err := m.Refresh(ctx); err == nil { |
| 383 | t.Fatal("a bucket blip must be reported, not swallowed") |
| 384 | } |
| 385 | cur, _ := m.meshKeys() |
| 386 | if !bytes.Equal(cur, good) { |
| 387 | t.Fatal("the last good mesh key must stand through a bucket blip") |
| 388 | } |
| 389 | } |
| 390 | |
amadan