| 1 | package console |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "context" |
| 6 | "fmt" |
| 7 | "io" |
| 8 | "net/http" |
| 9 | "net/http/httptest" |
| 10 | "net/http/httputil" |
| 11 | "net/url" |
| 12 | "strings" |
| 13 | "sync/atomic" |
| 14 | "testing" |
| 15 | "time" |
| 16 | |
| 17 | "github.com/carlosframework/platform/internal/geoip" |
| 18 | "github.com/carlosframework/platform/internal/status" |
| 19 | ) |
| 20 | |
| 21 | func putGeoip(t *testing.T, base, token string, body io.Reader, size int64) (int, string) { |
| 22 | t.Helper() |
| 23 | req, _ := http.NewRequest(http.MethodPut, base+"/api/cli/system/geoip", body) |
| 24 | req.ContentLength = size |
| 25 | req.Header.Set("Content-Type", "application/octet-stream") |
| 26 | req.Header.Set("Authorization", "Bearer "+token) |
| 27 | res, err := http.DefaultClient.Do(req) |
| 28 | if err != nil { |
| 29 | t.Fatal(err) |
| 30 | } |
| 31 | defer res.Body.Close() |
| 32 | b, _ := io.ReadAll(res.Body) |
| 33 | return res.StatusCode, string(b) |
| 34 | } |
| 35 | |
| 36 | func jobID(t *testing.T, body string) string { |
| 37 | t.Helper() |
| 38 | i := strings.Index(body, `"job":"`) |
| 39 | if i < 0 { |
| 40 | t.Fatalf("no job id in %q", body) |
| 41 | } |
| 42 | rest := body[i+len(`"job":"`):] |
| 43 | return rest[:strings.IndexByte(rest, '"')] |
| 44 | } |
| 45 | |
| 46 | func TestGeoipPublishIsAJobAndReturnsItsPointer(t *testing.T) { |
| 47 | withFixtureProbes(t) |
| 48 | fx := newConsoleFixture(t) |
| 49 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-put") |
| 50 | body := fixture(t, "GeoIP2-City-Test.mmdb") |
| 51 | st, out := putGeoip(t, fx.ts.URL, owner, bytes.NewReader(body), int64(len(body))) |
| 52 | if st != http.StatusAccepted { |
| 53 | t.Fatalf("PUT = %d %s, want 202", st, out) |
| 54 | } |
| 55 | job := waitJob(t, fx, owner, jobID(t, out)) |
| 56 | if job["error"] != "" { |
| 57 | t.Fatalf("job error = %v", job["error"]) |
| 58 | } |
| 59 | res, _ := job["result"].(map[string]any) |
| 60 | if res == nil || res["database_type"] != "GeoIP2-City" { |
| 61 | t.Fatalf("job result = %v, want the published pointer", job["result"]) |
| 62 | } |
| 63 | if _, _, err := fx.rel.Get(context.Background(), geoip.PointerKey); err != nil { |
| 64 | t.Fatalf("pointer not written: %v", err) |
| 65 | } |
| 66 | } |
| 67 | |
| 68 | func TestGeoipPublishRefusalIsTheJobsError(t *testing.T) { |
| 69 | withFixtureProbes(t) |
| 70 | fx := newConsoleFixture(t) |
| 71 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-junk") |
| 72 | st, out := putGeoip(t, fx.ts.URL, owner, strings.NewReader("junk"), 4) |
| 73 | if st != http.StatusAccepted { |
| 74 | t.Fatalf("PUT junk = %d %s, want 202 (verification is the job's)", st, out) |
| 75 | } |
| 76 | job := waitJob(t, fx, owner, jobID(t, out)) |
| 77 | if e, _ := job["error"].(string); !strings.Contains(e, "not a valid MaxMind-format database") { |
| 78 | t.Fatalf("job error = %q", e) |
| 79 | } |
| 80 | } |
| 81 | |
| 82 | func TestGeoipDoorsAreForOperatorsOnly(t *testing.T) { |
| 83 | fx := newConsoleFixture(t) |
| 84 | if err := fx.st.AddMember(OpsAccountID, "member@keymail.dev", RoleMember, time.Now()); err != nil { |
| 85 | t.Fatal(err) |
| 86 | } |
| 87 | member := cliToken(t, fx, "member@keymail.dev", "code-geoip-member") |
| 88 | _, want := getRaw(t, fx.ts.URL+"/api/cli/system/no-such-door", member) |
| 89 | if st, body := getRaw(t, fx.ts.URL+"/api/cli/system/geoip", member); st != http.StatusNotFound || body != want { |
| 90 | t.Errorf("GET as a member = %d %q, want the bare 404", st, body) |
| 91 | } |
| 92 | if st, _ := putGeoip(t, fx.ts.URL, member, strings.NewReader("x"), 1); st != http.StatusNotFound { |
| 93 | t.Errorf("PUT as a member = %d, want 404", st) |
| 94 | } |
| 95 | } |
| 96 | |
| 97 | // Codex R5-1: the slot is reserved before any of the body is read. |
| 98 | func TestSecondUploadIsRefusedBeforeItsBodyIsRead(t *testing.T) { |
| 99 | fx := newConsoleFixture(t) |
| 100 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-slot") |
| 101 | pr, pw := io.Pipe() |
| 102 | defer pw.Close() |
| 103 | go func() { pw.Write(make([]byte, 1024)) }() // then stall |
| 104 | first := make(chan struct{}) |
| 105 | go func() { |
| 106 | // Not putGeoip: t.Fatal from a goroutine would exit it without |
| 107 | // closing first. The first upload ends in an error by design. |
| 108 | defer close(first) |
| 109 | req, _ := http.NewRequest(http.MethodPut, fx.ts.URL+"/api/cli/system/geoip", pr) |
| 110 | req.ContentLength = 1 << 20 |
| 111 | req.Header.Set("Authorization", "Bearer "+owner) |
| 112 | if res, err := http.DefaultClient.Do(req); err == nil { |
| 113 | res.Body.Close() |
| 114 | } |
| 115 | }() |
| 116 | deadline := time.Now().Add(5 * time.Second) |
| 117 | for { |
| 118 | if !fx.srv.geoip.mu.TryLock() { |
| 119 | break // the first upload holds the slot |
| 120 | } |
| 121 | fx.srv.geoip.mu.Unlock() |
| 122 | if time.Now().After(deadline) { |
| 123 | t.Fatal("the first upload never took the slot") |
| 124 | } |
| 125 | time.Sleep(5 * time.Millisecond) |
| 126 | } |
| 127 | var read atomic.Bool |
| 128 | second := readSpy{r: strings.NewReader("second"), read: &read} |
| 129 | if st, out := putGeoip(t, fx.ts.URL, owner, second, 6); st != http.StatusConflict { |
| 130 | t.Fatalf("second PUT = %d %s, want 409", st, out) |
| 131 | } |
| 132 | // The client may have sent the bytes; what matters is the handler never |
| 133 | // read them, which an unread body plus a 409 shows from the server's side: |
| 134 | // the slot is still the first upload's. |
| 135 | if fx.srv.geoip.mu.TryLock() { |
| 136 | fx.srv.geoip.mu.Unlock() |
| 137 | t.Fatal("the refusal released the first upload's slot") |
| 138 | } |
| 139 | pw.CloseWithError(io.ErrUnexpectedEOF) |
| 140 | <-first |
| 141 | _ = read.Load() |
| 142 | } |
| 143 | |
| 144 | type readSpy struct { |
| 145 | r io.Reader |
| 146 | read *atomic.Bool |
| 147 | } |
| 148 | |
| 149 | func (s readSpy) Read(p []byte) (int, error) { s.read.Store(true); return s.r.Read(p) } |
| 150 | |
| 151 | // Codex R4-2: the 202 arrives through a proxy with a short response-header |
| 152 | // timeout even when verification takes longer. |
| 153 | func TestGeoipPublishAnswersBeforeVerificationThroughABackhaul(t *testing.T) { |
| 154 | withFixtureProbes(t) |
| 155 | fx := newConsoleFixture(t) |
| 156 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-backhaul") |
| 157 | geoipCommitDelay = 300 * time.Millisecond |
| 158 | t.Cleanup(func() { geoipCommitDelay = 0 }) |
| 159 | target, _ := url.Parse(fx.ts.URL) |
| 160 | rp := httputil.NewSingleHostReverseProxy(target) |
| 161 | rp.Transport = &http.Transport{ResponseHeaderTimeout: 100 * time.Millisecond} |
| 162 | proxy := httptest.NewServer(rp) |
| 163 | t.Cleanup(proxy.Close) |
| 164 | body := fixture(t, "GeoIP2-City-Test.mmdb") |
| 165 | st, out := putGeoip(t, proxy.URL, owner, bytes.NewReader(body), int64(len(body))) |
| 166 | if st != http.StatusAccepted { |
| 167 | t.Fatalf("PUT through the proxy = %d %s, want 202 before verification", st, out) |
| 168 | } |
| 169 | if job := waitJob(t, fx, owner, jobID(t, out)); job["error"] != "" { |
| 170 | t.Fatalf("job = %v", job) |
| 171 | } |
| 172 | } |
| 173 | |
| 174 | func seedGeoBox(t *testing.T, fx *consoleFixture, label string, g *status.GeoIPState, at time.Time) { |
| 175 | t.Helper() |
| 176 | b := status.CollectBox(nil, status.Probes{Release: "v1", GeoIP: g}, at) |
| 177 | b.Label = label |
| 178 | if err := status.WriteBox(context.Background(), fx.rel, b); err != nil { |
| 179 | t.Fatal(err) |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | func seedEdgesRecord(t *testing.T, fx *consoleFixture, label string) { |
| 184 | t.Helper() |
| 185 | if _, err := fx.rel.Put(context.Background(), "control/edges/"+label+".json", []byte(`{"v":1,"box":"`+label+`","eip":"192.0.2.1","hosts":{}}`), ""); err != nil { |
| 186 | t.Fatal(err) |
| 187 | } |
| 188 | } |
| 189 | |
| 190 | func TestGeoipStatusVerdict(t *testing.T) { |
| 191 | withFixtureProbes(t) |
| 192 | fx := newConsoleFixture(t) |
| 193 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-status") |
| 194 | ptr, err := fx.srv.geoip.Publish(context.Background(), bytes.NewReader(fixture(t, "GeoIP2-City-Test.mmdb")), geoipPublishOpts{Origin: "manual"}) |
| 195 | if err != nil { |
| 196 | t.Fatal(err) |
| 197 | } |
| 198 | now := time.Now() |
| 199 | seedEdgesRecord(t, fx, "ie-1") |
| 200 | seedEdgesRecord(t, fx, "uk-1") |
| 201 | seedGeoBox(t, fx, "ie-1", &status.GeoIPState{Mode: "mirror", SHA256: ptr.SHA256}, now) |
| 202 | // Codex R3-4: a heartbeat just inside the next jittered tick is fresh. |
| 203 | seedGeoBox(t, fx, "uk-1", &status.GeoIPState{Mode: "mirror", SHA256: ptr.SHA256}, now.Add(-(status.SnapshotPeriod + 59*time.Second))) |
| 204 | st, out := getJSON(t, fx.ts.URL+"/api/cli/system/geoip", owner) |
| 205 | if st != http.StatusOK || out["ready"] != true { |
| 206 | t.Fatalf("status = %d %v, want ready", st, out) |
| 207 | } |
| 208 | // An old binary, an expected box with no heartbeat, and a stale one. |
| 209 | seedEdgesRecord(t, fx, "us-1") |
| 210 | seedGeoBox(t, fx, "us-1", nil, now) |
| 211 | seedEdgesRecord(t, fx, "au-1") |
| 212 | seedEdgesRecord(t, fx, "sg-1") |
| 213 | seedGeoBox(t, fx, "sg-1", &status.GeoIPState{Mode: "mirror", SHA256: ptr.SHA256}, now.Add(-2*time.Hour)) |
| 214 | _, out = getJSON(t, fx.ts.URL+"/api/cli/system/geoip", owner) |
| 215 | if out["ready"] != false { |
| 216 | t.Fatalf("status = %v, want not ready", out) |
| 217 | } |
| 218 | blocking, _ := out["blocking"].([]any) |
| 219 | joined := "" |
| 220 | for _, b := range blocking { |
| 221 | joined += b.(string) + "\n" |
| 222 | } |
| 223 | for _, want := range []string{"us-1: runs a binary", "au-1: no heartbeat", "sg-1: no heartbeat since", "control/edges/sg-1.json"} { |
| 224 | if !strings.Contains(joined, want) { |
| 225 | t.Errorf("blocking lacks %q:\n%s", want, joined) |
| 226 | } |
| 227 | } |
| 228 | } |
| 229 | |
| 230 | func TestGeoipVerdictIncompleteReadIsNeverReady(t *testing.T) { |
| 231 | ready, _, blocking := geoipVerdict(geoipWant{state: geoipOff}, geoipInventoryRead{expected: map[geoipBoxKey]bool{}, boxes: map[geoipBoxKey]status.Box{}, expectedComplete: false, heartbeatsComplete: true}, time.Now()) |
| 232 | if ready || len(blocking) == 0 { |
| 233 | t.Fatalf("incomplete read = ready %v, blocking %v", ready, blocking) |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | func TestGeoipOff(t *testing.T) { |
| 238 | withFixtureProbes(t) |
| 239 | fx := newConsoleFixture(t) |
| 240 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-off") |
| 241 | if _, err := fx.srv.geoip.Publish(context.Background(), bytes.NewReader(fixture(t, "GeoIP2-City-Test.mmdb")), geoipPublishOpts{Origin: "manual"}); err != nil { |
| 242 | t.Fatal(err) |
| 243 | } |
| 244 | req, _ := http.NewRequest(http.MethodDelete, fx.ts.URL+"/api/cli/system/geoip", nil) |
| 245 | req.Header.Set("Authorization", "Bearer "+owner) |
| 246 | res, err := http.DefaultClient.Do(req) |
| 247 | if err != nil { |
| 248 | t.Fatal(err) |
| 249 | } |
| 250 | res.Body.Close() |
| 251 | if res.StatusCode != http.StatusOK { |
| 252 | t.Fatalf("DELETE = %d", res.StatusCode) |
| 253 | } |
| 254 | if _, _, err := fx.rel.Get(context.Background(), geoip.PointerKey); err == nil { |
| 255 | t.Fatal("pointer survived off") |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | // Codex R8-1: identity is (namespace, label). A fresh platform box must not |
| 260 | // stand in for a fleet box of the same label, nor one fleet's for another's. |
| 261 | func TestGeoipVerdictKeysBoxesByNamespace(t *testing.T) { |
| 262 | now := time.Now() |
| 263 | fresh := func(label string) status.Box { |
| 264 | b := status.CollectBox(nil, status.Probes{Release: "v1", GeoIP: &status.GeoIPState{Mode: "mirror", SHA256: "s"}}, now) |
| 265 | b.Label = label |
| 266 | return b |
| 267 | } |
| 268 | old := status.CollectBox(nil, status.Probes{Release: "v0"}, now) |
| 269 | old.Label = "node-1" |
| 270 | inv := geoipInventoryRead{ |
| 271 | expected: map[geoipBoxKey]bool{{"", "node-1"}: true, {"tito", "node-1"}: true, {"bae", "node-1"}: true}, |
| 272 | boxes: map[geoipBoxKey]status.Box{{"", "node-1"}: fresh("node-1"), {"tito", "node-1"}: fresh("node-1"), {"bae", "node-1"}: old}, |
| 273 | expectedComplete: true, heartbeatsComplete: true, |
| 274 | } |
| 275 | ready, _, blocking := geoipVerdict(geoipWant{state: geoipLive, sha: "s"}, inv, now) |
| 276 | if ready || len(blocking) != 1 || !strings.HasPrefix(blocking[0], "bae/node-1: runs a binary") { |
| 277 | t.Fatalf("ready=%v blocking=%v, want only bae/node-1 blocking", ready, blocking) |
| 278 | } |
| 279 | delete(inv.boxes, geoipBoxKey{"tito", "node-1"}) |
| 280 | _, _, blocking = geoipVerdict(geoipWant{state: geoipLive, sha: "s"}, inv, now) |
| 281 | found := false |
| 282 | for _, b := range blocking { |
| 283 | found = found || strings.HasPrefix(b, "tito/node-1: no heartbeat") |
| 284 | } |
| 285 | if !found { |
| 286 | t.Fatalf("a missing fleet box was satisfied by a same-label box elsewhere: %v", blocking) |
| 287 | } |
| 288 | } |
| 289 | |
| 290 | // Codex R8 on R1-16: a box with no edges record (retired) is listed but |
| 291 | // never blocks. |
| 292 | func TestGeoipVerdictRetiredBoxDoesNotBlock(t *testing.T) { |
| 293 | now := time.Now() |
| 294 | b := status.CollectBox(nil, status.Probes{Release: "v0"}, now.Add(-time.Hour)) |
| 295 | b.Label = "gone-1" |
| 296 | inv := geoipInventoryRead{expected: map[geoipBoxKey]bool{}, boxes: map[geoipBoxKey]status.Box{{"", "gone-1"}: b}, expectedComplete: true, heartbeatsComplete: false} |
| 297 | ready, rows, _ := geoipVerdict(geoipWant{state: geoipOff}, inv, now) |
| 298 | if !ready || len(rows) != 1 { |
| 299 | t.Fatalf("ready=%v rows=%v, want ready with the retired box listed", ready, rows) |
| 300 | } |
| 301 | } |
| 302 | |
| 303 | // Codex R9-1: an unreadable pointer is never ready, and off requires every |
| 304 | // mirror to have dropped its generation. |
| 305 | func TestGeoipVerdictPointerStates(t *testing.T) { |
| 306 | now := time.Now() |
| 307 | mirror := func(sha string) status.Box { |
| 308 | b := status.CollectBox(nil, status.Probes{Release: "v1", GeoIP: &status.GeoIPState{Mode: "mirror", SHA256: sha}}, now) |
| 309 | b.Label = "ie-1" |
| 310 | return b |
| 311 | } |
| 312 | inv := func(sha string) geoipInventoryRead { |
| 313 | return geoipInventoryRead{expected: map[geoipBoxKey]bool{{"", "ie-1"}: true}, |
| 314 | boxes: map[geoipBoxKey]status.Box{{"", "ie-1"}: mirror(sha)}, expectedComplete: true, heartbeatsComplete: true} |
| 315 | } |
| 316 | if ready, _, _ := geoipVerdict(geoipWant{state: geoipUnknown}, inv("a"), now); ready { |
| 317 | t.Fatal("an unreadable pointer reported ready") |
| 318 | } |
| 319 | if ready, _, b := geoipVerdict(geoipWant{state: geoipOff}, inv("a"), now); ready || len(b) != 1 || !strings.Contains(b[0], "still has") { |
| 320 | t.Fatalf("off with a mirror still loaded = %v %v", ready, b) |
| 321 | } |
| 322 | if ready, _, _ := geoipVerdict(geoipWant{state: geoipOff}, inv(""), now); !ready { |
| 323 | t.Fatal("off with every mirror empty must be ready") |
| 324 | } |
| 325 | if ready, _, _ := geoipVerdict(geoipWant{state: geoipLive, sha: "b"}, inv("a"), now); ready { |
| 326 | t.Fatal("a mirror on another sha reported ready") |
| 327 | } |
| 328 | } |
| 329 | |
| 330 | // Codex R9-2: off never waits behind a publish. |
| 331 | func TestGeoipOffDuringAPublishIs409(t *testing.T) { |
| 332 | withFixtureProbes(t) |
| 333 | fx := newConsoleFixture(t) |
| 334 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-offbusy") |
| 335 | st, err := fx.srv.geoip.Stage(context.Background(), bytes.NewReader(fixture(t, "GeoIP2-City-Test.mmdb"))) |
| 336 | if err != nil { |
| 337 | t.Fatal(err) |
| 338 | } |
| 339 | defer fx.srv.geoip.Abandon(st) |
| 340 | req, _ := http.NewRequest(http.MethodDelete, fx.ts.URL+"/api/cli/system/geoip", nil) |
| 341 | req.Header.Set("Authorization", "Bearer "+owner) |
| 342 | done := make(chan int, 1) |
| 343 | go func() { |
| 344 | res, err := http.DefaultClient.Do(req) |
| 345 | if err != nil { |
| 346 | done <- 0 |
| 347 | return |
| 348 | } |
| 349 | res.Body.Close() |
| 350 | done <- res.StatusCode |
| 351 | }() |
| 352 | select { |
| 353 | case code := <-done: |
| 354 | if code != http.StatusConflict { |
| 355 | t.Fatalf("off during a publish = %d, want 409", code) |
| 356 | } |
| 357 | case <-time.After(5 * time.Second): |
| 358 | t.Fatal("off waited behind the publish") |
| 359 | } |
| 360 | } |
| 361 | |
| 362 | // Codex R10 on R9-1: a pointer past D1 by the clock is "off" for the |
| 363 | // verdict even before the daily job marks it. |
| 364 | func TestGeoipStatusTreatsAnUnmarkedExpiredPointerAsOff(t *testing.T) { |
| 365 | fx := newConsoleFixture(t) |
| 366 | owner := cliToken(t, fx, "paul@keymail.dev", "code-geoip-expired") |
| 367 | sha := strings.Repeat("c", 64) |
| 368 | ptr := geoip.Pointer{V: 1, SHA256: sha, Key: geoip.BlobKey(sha), Size: 1, DatabaseType: "GeoLite2-City", |
| 369 | BuildEpoch: time.Now().Add(-40 * 24 * time.Hour).Unix(), Origin: "manual"} |
| 370 | b, _ := marshalPointer(ptr) |
| 371 | if _, err := fx.rel.Put(context.Background(), geoip.PointerKey, b, ""); err != nil { |
| 372 | t.Fatal(err) |
| 373 | } |
| 374 | seedEdgesRecord(t, fx, "ie-1") |
| 375 | seedGeoBox(t, fx, "ie-1", &status.GeoIPState{Mode: "mirror", SHA256: sha}, time.Now()) |
| 376 | _, out := getJSON(t, fx.ts.URL+"/api/cli/system/geoip", owner) |
| 377 | if out["ready"] != false || !strings.Contains(fmt.Sprint(out["blocking"]), "still has") { |
| 378 | t.Fatalf("expired pointer with a loaded mirror = %v", out) |
| 379 | } |
| 380 | seedGeoBox(t, fx, "ie-1", &status.GeoIPState{Mode: "mirror"}, time.Now()) |
| 381 | if _, out = getJSON(t, fx.ts.URL+"/api/cli/system/geoip", owner); out["ready"] != true { |
| 382 | t.Fatalf("expired pointer with every mirror empty = %v, want ready", out) |
| 383 | } |
| 384 | } |
| 385 | |
amadan