services/keepstore/azure_blob_volume_test.go

   1 package main
   2
   3 import (
   4         "bytes"
   5         "crypto/md5"
   6         "encoding/base64"
   7         "encoding/xml"
   8         "flag"
   9         "fmt"
  10         "io/ioutil"
  11         "log"
  12         "math/rand"
  13         "net"
  14         "net/http"
  15         "net/http/httptest"
  16         "regexp"
  17         "sort"
  18         "strconv"
  19         "strings"
  20         "sync"
  21         "testing"
  22         "time"
  23
  24         "github.com/curoverse/azure-sdk-for-go/storage"
  25 )
  26
  27 const (
  28         // The same fake credentials used by Microsoft's Azure emulator
  29         emulatorAccountName = "devstoreaccount1"
  30         emulatorAccountKey  = "Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw=="
  31 )
  32
  33 var azureTestContainer string
  34
  35 func init() {
  36         flag.StringVar(
  37                 &azureTestContainer,
  38                 "test.azure-storage-container-volume",
  39                 "",
  40                 "Name of Azure container to use for testing. Do not use a container with real data! Use -azure-storage-account-name and -azure-storage-key-file arguments to supply credentials.")
  41 }
  42
  43 type azBlob struct {
  44         Data        []byte
  45         Etag        string
  46         Metadata    map[string]string
  47         Mtime       time.Time
  48         Uncommitted map[string][]byte
  49 }
  50
  51 type azStubHandler struct {
  52         sync.Mutex
  53         blobs map[string]*azBlob
  54         race  chan chan struct{}
  55 }
  56
  57 func newAzStubHandler() *azStubHandler {
  58         return &azStubHandler{
  59                 blobs: make(map[string]*azBlob),
  60         }
  61 }
  62
  63 func (h *azStubHandler) TouchWithDate(container, hash string, t time.Time) {
  64         blob, ok := h.blobs[container+"|"+hash]
  65         if !ok {
  66                 return
  67         }
  68         blob.Mtime = t
  69 }
  70
  71 func (h *azStubHandler) PutRaw(container, hash string, data []byte) {
  72         h.Lock()
  73         defer h.Unlock()
  74         h.blobs[container+"|"+hash] = &azBlob{
  75                 Data:        data,
  76                 Mtime:       time.Now(),
  77                 Metadata:    make(map[string]string),
  78                 Uncommitted: make(map[string][]byte),
  79         }
  80 }
  81
  82 func (h *azStubHandler) unlockAndRace() {
  83         if h.race == nil {
  84                 return
  85         }
  86         h.Unlock()
  87         // Signal caller that race is starting by reading from
  88         // h.race. If we get a channel, block until that channel is
  89         // ready to receive. If we get nil (or h.race is closed) just
  90         // proceed.
  91         if c := <-h.race; c != nil {
  92                 c <- struct{}{}
  93         }
  94         h.Lock()
  95 }
  96
  97 var rangeRegexp = regexp.MustCompile(`^bytes=(\d+)-(\d+)$`)
  98
  99 func (h *azStubHandler) ServeHTTP(rw http.ResponseWriter, r *http.Request) {
 100         h.Lock()
 101         defer h.Unlock()
 102         // defer log.Printf("azStubHandler: %+v", r)
 103
 104         path := strings.Split(r.URL.Path, "/")
 105         container := path[1]
 106         hash := ""
 107         if len(path) > 2 {
 108                 hash = path[2]
 109         }
 110
 111         if err := r.ParseForm(); err != nil {
 112                 log.Printf("azStubHandler(%+v): %s", r, err)
 113                 rw.WriteHeader(http.StatusBadRequest)
 114                 return
 115         }
 116
 117         body, err := ioutil.ReadAll(r.Body)
 118         if err != nil {
 119                 return
 120         }
 121
 122         type blockListRequestBody struct {
 123                 XMLName     xml.Name `xml:"BlockList"`
 124                 Uncommitted []string
 125         }
 126
 127         blob, blobExists := h.blobs[container+"|"+hash]
 128
 129         switch {
 130         case r.Method == "PUT" && r.Form.Get("comp") == "":
 131                 // "Put Blob" API
 132                 if _, ok := h.blobs[container+"|"+hash]; !ok {
 133                         // Like the real Azure service, we offer a
 134                         // race window during which other clients can
 135                         // list/get the new blob before any data is
 136                         // committed.
 137                         h.blobs[container+"|"+hash] = &azBlob{
 138                                 Mtime:       time.Now(),
 139                                 Uncommitted: make(map[string][]byte),
 140                                 Metadata:    make(map[string]string),
 141                                 Etag:        makeEtag(),
 142                         }
 143                         h.unlockAndRace()
 144                 }
 145                 metadata := make(map[string]string)
 146                 for k, v := range r.Header {
 147                         if strings.HasPrefix(strings.ToLower(k), "x-ms-meta-") {
 148                                 name := k[len("x-ms-meta-"):]
 149                                 metadata[strings.ToLower(name)] = v[0]
 150                         }
 151                 }
 152                 h.blobs[container+"|"+hash] = &azBlob{
 153                         Data:        body,
 154                         Mtime:       time.Now(),
 155                         Uncommitted: make(map[string][]byte),
 156                         Metadata:    metadata,
 157                         Etag:        makeEtag(),
 158                 }
 159                 rw.WriteHeader(http.StatusCreated)
 160         case r.Method == "PUT" && r.Form.Get("comp") == "block":
 161                 // "Put Block" API
 162                 if !blobExists {
 163                         log.Printf("Got block for nonexistent blob: %+v", r)
 164                         rw.WriteHeader(http.StatusBadRequest)
 165                         return
 166                 }
 167                 blockID, err := base64.StdEncoding.DecodeString(r.Form.Get("blockid"))
 168                 if err != nil || len(blockID) == 0 {
 169                         log.Printf("Invalid blockid: %+q", r.Form.Get("blockid"))
 170                         rw.WriteHeader(http.StatusBadRequest)
 171                         return
 172                 }
 173                 blob.Uncommitted[string(blockID)] = body
 174                 rw.WriteHeader(http.StatusCreated)
 175         case r.Method == "PUT" && r.Form.Get("comp") == "blocklist":
 176                 // "Put Block List" API
 177                 bl := &blockListRequestBody{}
 178                 if err := xml.Unmarshal(body, bl); err != nil {
 179                         log.Printf("xml Unmarshal: %s", err)
 180                         rw.WriteHeader(http.StatusBadRequest)
 181                         return
 182                 }
 183                 for _, encBlockID := range bl.Uncommitted {
 184                         blockID, err := base64.StdEncoding.DecodeString(encBlockID)
 185                         if err != nil || len(blockID) == 0 || blob.Uncommitted[string(blockID)] == nil {
 186                                 log.Printf("Invalid blockid: %+q", encBlockID)
 187                                 rw.WriteHeader(http.StatusBadRequest)
 188                                 return
 189                         }
 190                         blob.Data = blob.Uncommitted[string(blockID)]
 191                         blob.Etag = makeEtag()
 192                         blob.Mtime = time.Now()
 193                         delete(blob.Uncommitted, string(blockID))
 194                 }
 195                 rw.WriteHeader(http.StatusCreated)
 196         case r.Method == "PUT" && r.Form.Get("comp") == "metadata":
 197                 // "Set Metadata Headers" API. We don't bother
 198                 // stubbing "Get Metadata Headers": AzureBlobVolume
 199                 // sets metadata headers only as a way to bump Etag
 200                 // and Last-Modified.
 201                 if !blobExists {
 202                         log.Printf("Got metadata for nonexistent blob: %+v", r)
 203                         rw.WriteHeader(http.StatusBadRequest)
 204                         return
 205                 }
 206                 blob.Metadata = make(map[string]string)
 207                 for k, v := range r.Header {
 208                         if strings.HasPrefix(strings.ToLower(k), "x-ms-meta-") {
 209                                 name := k[len("x-ms-meta-"):]
 210                                 blob.Metadata[strings.ToLower(name)] = v[0]
 211                         }
 212                 }
 213                 blob.Mtime = time.Now()
 214                 blob.Etag = makeEtag()
 215         case (r.Method == "GET" || r.Method == "HEAD") && r.Form.Get("comp") == "metadata" && hash != "":
 216                 // "Get Blob Metadata" API
 217                 if !blobExists {
 218                         rw.WriteHeader(http.StatusNotFound)
 219                         return
 220                 }
 221                 for k, v := range blob.Metadata {
 222                         rw.Header().Set(fmt.Sprintf("x-ms-meta-%s", k), v)
 223                 }
 224                 return
 225         case (r.Method == "GET" || r.Method == "HEAD") && hash != "":
 226                 // "Get Blob" API
 227                 if !blobExists {
 228                         rw.WriteHeader(http.StatusNotFound)
 229                         return
 230                 }
 231                 data := blob.Data
 232                 if rangeSpec := rangeRegexp.FindStringSubmatch(r.Header.Get("Range")); rangeSpec != nil {
 233                         b0, err0 := strconv.Atoi(rangeSpec[1])
 234                         b1, err1 := strconv.Atoi(rangeSpec[2])
 235                         if err0 != nil || err1 != nil || b0 >= len(data) || b1 >= len(data) || b0 > b1 {
 236                                 rw.Header().Set("Content-Range", fmt.Sprintf("bytes */%d", len(data)))
 237                                 rw.WriteHeader(http.StatusRequestedRangeNotSatisfiable)
 238                                 return
 239                         }
 240                         rw.Header().Set("Content-Range", fmt.Sprintf("bytes %d-%d/%d", b0, b1, len(data)))
 241                         rw.WriteHeader(http.StatusPartialContent)
 242                         data = data[b0 : b1+1]
 243                 }
 244                 rw.Header().Set("Last-Modified", blob.Mtime.Format(time.RFC1123))
 245                 rw.Header().Set("Content-Length", strconv.Itoa(len(data)))
 246                 if r.Method == "GET" {
 247                         if _, err := rw.Write(data); err != nil {
 248                                 log.Printf("write %+q: %s", data, err)
 249                         }
 250                 }
 251                 h.unlockAndRace()
 252         case r.Method == "DELETE" && hash != "":
 253                 // "Delete Blob" API
 254                 if !blobExists {
 255                         rw.WriteHeader(http.StatusNotFound)
 256                         return
 257                 }
 258                 delete(h.blobs, container+"|"+hash)
 259                 rw.WriteHeader(http.StatusAccepted)
 260         case r.Method == "GET" && r.Form.Get("comp") == "list" && r.Form.Get("restype") == "container":
 261                 // "List Blobs" API
 262                 prefix := container + "|" + r.Form.Get("prefix")
 263                 marker := r.Form.Get("marker")
 264
 265                 maxResults := 2
 266                 if n, err := strconv.Atoi(r.Form.Get("maxresults")); err == nil && n >= 1 && n <= 5000 {
 267                         maxResults = n
 268                 }
 269
 270                 resp := storage.BlobListResponse{
 271                         Marker:     marker,
 272                         NextMarker: "",
 273                         MaxResults: int64(maxResults),
 274                 }
 275                 var hashes sort.StringSlice
 276                 for k := range h.blobs {
 277                         if strings.HasPrefix(k, prefix) {
 278                                 hashes = append(hashes, k[len(container)+1:])
 279                         }
 280                 }
 281                 hashes.Sort()
 282                 for _, hash := range hashes {
 283                         if len(resp.Blobs) == maxResults {
 284                                 resp.NextMarker = hash
 285                                 break
 286                         }
 287                         if len(resp.Blobs) > 0 || marker == "" || marker == hash {
 288                                 blob := h.blobs[container+"|"+hash]
 289                                 bmeta := map[string]string(nil)
 290                                 if r.Form.Get("include") == "metadata" {
 291                                         bmeta = blob.Metadata
 292                                 }
 293                                 b := storage.Blob{
 294                                         Name: hash,
 295                                         Properties: storage.BlobProperties{
 296                                                 LastModified:  blob.Mtime.Format(time.RFC1123),
 297                                                 ContentLength: int64(len(blob.Data)),
 298                                                 Etag:          blob.Etag,
 299                                         },
 300                                         Metadata: bmeta,
 301                                 }
 302                                 resp.Blobs = append(resp.Blobs, b)
 303                         }
 304                 }
 305                 buf, err := xml.Marshal(resp)
 306                 if err != nil {
 307                         log.Print(err)
 308                         rw.WriteHeader(http.StatusInternalServerError)
 309                 }
 310                 rw.Write(buf)
 311         default:
 312                 log.Printf("azStubHandler: not implemented: %+v Body:%+q", r, body)
 313                 rw.WriteHeader(http.StatusNotImplemented)
 314         }
 315 }
 316
 317 // azStubDialer is a net.Dialer that notices when the Azure driver
 318 // tries to connect to "devstoreaccount1.blob.127.0.0.1:46067", and
 319 // in such cases transparently dials "127.0.0.1:46067" instead.
 320 type azStubDialer struct {
 321         net.Dialer
 322 }
 323
 324 var localHostPortRe = regexp.MustCompile(`(127\.0\.0\.1|localhost|\[::1\]):\d+`)
 325
 326 func (d *azStubDialer) Dial(network, address string) (net.Conn, error) {
 327         if hp := localHostPortRe.FindString(address); hp != "" {
 328                 log.Println("azStubDialer: dial", hp, "instead of", address)
 329                 address = hp
 330         }
 331         return d.Dialer.Dial(network, address)
 332 }
 333
 334 type TestableAzureBlobVolume struct {
 335         *AzureBlobVolume
 336         azHandler *azStubHandler
 337         azStub    *httptest.Server
 338         t         TB
 339 }
 340
 341 func NewTestableAzureBlobVolume(t TB, readonly bool, replication int) *TestableAzureBlobVolume {
 342         azHandler := newAzStubHandler()
 343         azStub := httptest.NewServer(azHandler)
 344
 345         var azClient storage.Client
 346
 347         container := azureTestContainer
 348         if container == "" {
 349                 // Connect to stub instead of real Azure storage service
 350                 stubURLBase := strings.Split(azStub.URL, "://")[1]
 351                 var err error
 352                 if azClient, err = storage.NewClient(emulatorAccountName, emulatorAccountKey, stubURLBase, storage.DefaultAPIVersion, false); err != nil {
 353                         t.Fatal(err)
 354                 }
 355                 container = "fakecontainername"
 356         } else {
 357                 // Connect to real Azure storage service
 358                 accountKey, err := readKeyFromFile(azureStorageAccountKeyFile)
 359                 if err != nil {
 360                         t.Fatal(err)
 361                 }
 362                 azClient, err = storage.NewBasicClient(azureStorageAccountName, accountKey)
 363                 if err != nil {
 364                         t.Fatal(err)
 365                 }
 366         }
 367
 368         v := NewAzureBlobVolume(azClient, container, readonly, replication)
 369
 370         return &TestableAzureBlobVolume{
 371                 AzureBlobVolume: v,
 372                 azHandler:       azHandler,
 373                 azStub:          azStub,
 374                 t:               t,
 375         }
 376 }
 377
 378 func TestAzureBlobVolumeWithGeneric(t *testing.T) {
 379         defer func(t http.RoundTripper) {
 380                 http.DefaultTransport = t
 381         }(http.DefaultTransport)
 382         http.DefaultTransport = &http.Transport{
 383                 Dial: (&azStubDialer{}).Dial,
 384         }
 385         azureWriteRaceInterval = time.Millisecond
 386         azureWriteRacePollTime = time.Nanosecond
 387         DoGenericVolumeTests(t, func(t TB) TestableVolume {
 388                 return NewTestableAzureBlobVolume(t, false, azureStorageReplication)
 389         })
 390 }
 391
 392 func TestAzureBlobVolumeConcurrentRanges(t *testing.T) {
 393         defer func(b int) {
 394                 azureMaxGetBytes = b
 395         }(azureMaxGetBytes)
 396
 397         defer func(t http.RoundTripper) {
 398                 http.DefaultTransport = t
 399         }(http.DefaultTransport)
 400         http.DefaultTransport = &http.Transport{
 401                 Dial: (&azStubDialer{}).Dial,
 402         }
 403         azureWriteRaceInterval = time.Millisecond
 404         azureWriteRacePollTime = time.Nanosecond
 405         // Test (BlockSize mod azureMaxGetBytes)==0 and !=0 cases
 406         for _, azureMaxGetBytes = range []int{2 << 22, 2<<22 - 1} {
 407                 DoGenericVolumeTests(t, func(t TB) TestableVolume {
 408                         return NewTestableAzureBlobVolume(t, false, azureStorageReplication)
 409                 })
 410         }
 411 }
 412
 413 func TestReadonlyAzureBlobVolumeWithGeneric(t *testing.T) {
 414         defer func(t http.RoundTripper) {
 415                 http.DefaultTransport = t
 416         }(http.DefaultTransport)
 417         http.DefaultTransport = &http.Transport{
 418                 Dial: (&azStubDialer{}).Dial,
 419         }
 420         azureWriteRaceInterval = time.Millisecond
 421         azureWriteRacePollTime = time.Nanosecond
 422         DoGenericVolumeTests(t, func(t TB) TestableVolume {
 423                 return NewTestableAzureBlobVolume(t, true, azureStorageReplication)
 424         })
 425 }
 426
 427 func TestAzureBlobVolumeRangeFenceposts(t *testing.T) {
 428         defer func(t http.RoundTripper) {
 429                 http.DefaultTransport = t
 430         }(http.DefaultTransport)
 431         http.DefaultTransport = &http.Transport{
 432                 Dial: (&azStubDialer{}).Dial,
 433         }
 434
 435         v := NewTestableAzureBlobVolume(t, false, 3)
 436         defer v.Teardown()
 437
 438         for _, size := range []int{
 439                 2<<22 - 1, // one <max read
 440                 2 << 22,   // one =max read
 441                 2<<22 + 1, // one =max read, one <max
 442                 2 << 23,   // two =max reads
 443                 BlockSize - 1,
 444                 BlockSize,
 445         } {
 446                 data := make([]byte, size)
 447                 for i := range data {
 448                         data[i] = byte((i + 7) & 0xff)
 449                 }
 450                 hash := fmt.Sprintf("%x", md5.Sum(data))
 451                 err := v.Put(hash, data)
 452                 if err != nil {
 453                         t.Error(err)
 454                 }
 455                 gotData := make([]byte, len(data))
 456                 gotLen, err := v.Get(hash, gotData)
 457                 if err != nil {
 458                         t.Error(err)
 459                 }
 460                 gotHash := fmt.Sprintf("%x", md5.Sum(gotData))
 461                 if gotLen != size {
 462                         t.Error("length mismatch: got %d != %d", gotLen, size)
 463                 }
 464                 if gotHash != hash {
 465                         t.Error("hash mismatch: got %s != %s", gotHash, hash)
 466                 }
 467         }
 468 }
 469
 470 func TestAzureBlobVolumeReplication(t *testing.T) {
 471         for r := 1; r <= 4; r++ {
 472                 v := NewTestableAzureBlobVolume(t, false, r)
 473                 defer v.Teardown()
 474                 if n := v.Replication(); n != r {
 475                         t.Errorf("Got replication %d, expected %d", n, r)
 476                 }
 477         }
 478 }
 479
 480 func TestAzureBlobVolumeCreateBlobRace(t *testing.T) {
 481         defer func(t http.RoundTripper) {
 482                 http.DefaultTransport = t
 483         }(http.DefaultTransport)
 484         http.DefaultTransport = &http.Transport{
 485                 Dial: (&azStubDialer{}).Dial,
 486         }
 487
 488         v := NewTestableAzureBlobVolume(t, false, 3)
 489         defer v.Teardown()
 490
 491         azureWriteRaceInterval = time.Second
 492         azureWriteRacePollTime = time.Millisecond
 493
 494         allDone := make(chan struct{})
 495         v.azHandler.race = make(chan chan struct{})
 496         go func() {
 497                 err := v.Put(TestHash, TestBlock)
 498                 if err != nil {
 499                         t.Error(err)
 500                 }
 501         }()
 502         continuePut := make(chan struct{})
 503         // Wait for the stub's Put to create the empty blob
 504         v.azHandler.race <- continuePut
 505         go func() {
 506                 buf := make([]byte, len(TestBlock))
 507                 _, err := v.Get(TestHash, buf)
 508                 if err != nil {
 509                         t.Error(err)
 510                 }
 511                 close(allDone)
 512         }()
 513         // Wait for the stub's Get to get the empty blob
 514         close(v.azHandler.race)
 515         // Allow stub's Put to continue, so the real data is ready
 516         // when the volume's Get retries
 517         <-continuePut
 518         // Wait for volume's Get to return the real data
 519         <-allDone
 520 }
 521
 522 func TestAzureBlobVolumeCreateBlobRaceDeadline(t *testing.T) {
 523         defer func(t http.RoundTripper) {
 524                 http.DefaultTransport = t
 525         }(http.DefaultTransport)
 526         http.DefaultTransport = &http.Transport{
 527                 Dial: (&azStubDialer{}).Dial,
 528         }
 529
 530         v := NewTestableAzureBlobVolume(t, false, 3)
 531         defer v.Teardown()
 532
 533         azureWriteRaceInterval = 2 * time.Second
 534         azureWriteRacePollTime = 5 * time.Millisecond
 535
 536         v.PutRaw(TestHash, nil)
 537
 538         buf := new(bytes.Buffer)
 539         v.IndexTo("", buf)
 540         if buf.Len() != 0 {
 541                 t.Errorf("Index %+q should be empty", buf.Bytes())
 542         }
 543
 544         v.TouchWithDate(TestHash, time.Now().Add(-1982*time.Millisecond))
 545
 546         allDone := make(chan struct{})
 547         go func() {
 548                 defer close(allDone)
 549                 buf := make([]byte, BlockSize)
 550                 n, err := v.Get(TestHash, buf)
 551                 if err != nil {
 552                         t.Error(err)
 553                         return
 554                 }
 555                 if n != 0 {
 556                         t.Errorf("Got %+q, expected empty buf", buf[:n])
 557                 }
 558         }()
 559         select {
 560         case <-allDone:
 561         case <-time.After(time.Second):
 562                 t.Error("Get should have stopped waiting for race when block was 2s old")
 563         }
 564
 565         buf.Reset()
 566         v.IndexTo("", buf)
 567         if !bytes.HasPrefix(buf.Bytes(), []byte(TestHash+"+0")) {
 568                 t.Errorf("Index %+q should have %+q", buf.Bytes(), TestHash+"+0")
 569         }
 570 }
 571
 572 func (v *TestableAzureBlobVolume) PutRaw(locator string, data []byte) {
 573         v.azHandler.PutRaw(v.containerName, locator, data)
 574 }
 575
 576 func (v *TestableAzureBlobVolume) TouchWithDate(locator string, lastPut time.Time) {
 577         v.azHandler.TouchWithDate(v.containerName, locator, lastPut)
 578 }
 579
 580 func (v *TestableAzureBlobVolume) Teardown() {
 581         v.azStub.Close()
 582 }
 583
 584 func makeEtag() string {
 585         return fmt.Sprintf("0x%x", rand.Int63())
 586 }