services/keepproxy/keepproxy.go

   1 package main
   2
   3 import (
   4         "errors"
   5         "flag"
   6         "fmt"
   7         "io"
   8         "io/ioutil"
   9         "log"
  10         "net"
  11         "net/http"
  12         "os"
  13         "os/signal"
  14         "regexp"
  15         "strings"
  16         "sync"
  17         "syscall"
  18         "time"
  19
  20         "git.curoverse.com/arvados.git/sdk/go/arvados"
  21         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  22         "git.curoverse.com/arvados.git/sdk/go/config"
  23         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  24         "github.com/coreos/go-systemd/daemon"
  25         "github.com/ghodss/yaml"
  26         "github.com/gorilla/mux"
  27 )
  28
  29 type Config struct {
  30         Client          arvados.Client
  31         Listen          string
  32         DisableGet      bool
  33         DisablePut      bool
  34         DefaultReplicas int
  35         Timeout         arvados.Duration
  36         PIDFile         string
  37         Debug           bool
  38 }
  39
  40 func DefaultConfig() *Config {
  41         return &Config{
  42                 Listen:  ":25107",
  43                 Timeout: arvados.Duration(15 * time.Second),
  44         }
  45 }
  46
  47 var (
  48         listener net.Listener
  49         router   http.Handler
  50 )
  51
  52 func main() {
  53         cfg := DefaultConfig()
  54
  55         flagset := flag.NewFlagSet("keepproxy", flag.ExitOnError)
  56         flagset.Usage = usage
  57
  58         const deprecated = " (DEPRECATED -- use config file instead)"
  59         flagset.StringVar(&cfg.Listen, "listen", cfg.Listen, "Local port to listen on."+deprecated)
  60         flagset.BoolVar(&cfg.DisableGet, "no-get", cfg.DisableGet, "Disable GET operations."+deprecated)
  61         flagset.BoolVar(&cfg.DisablePut, "no-put", cfg.DisablePut, "Disable PUT operations."+deprecated)
  62         flagset.IntVar(&cfg.DefaultReplicas, "default-replicas", cfg.DefaultReplicas, "Default number of replicas to write if not specified by the client. If 0, use site default."+deprecated)
  63         flagset.StringVar(&cfg.PIDFile, "pid", cfg.PIDFile, "Path to write pid file."+deprecated)
  64         timeoutSeconds := flagset.Int("timeout", int(time.Duration(cfg.Timeout)/time.Second), "Timeout (in seconds) on requests to internal Keep services."+deprecated)
  65
  66         var cfgPath string
  67         const defaultCfgPath = "/etc/arvados/keepproxy/keepproxy.yml"
  68         flagset.StringVar(&cfgPath, "config", defaultCfgPath, "Configuration file `path`")
  69         dumpConfig := flagset.Bool("dump-config", false, "write current configuration to stdout and exit")
  70         flagset.Parse(os.Args[1:])
  71
  72         err := config.LoadFile(cfg, cfgPath)
  73         if err != nil {
  74                 h := os.Getenv("ARVADOS_API_HOST")
  75                 t := os.Getenv("ARVADOS_API_TOKEN")
  76                 if h == "" || t == "" || !os.IsNotExist(err) || cfgPath != defaultCfgPath {
  77                         log.Fatal(err)
  78                 }
  79                 log.Print("DEPRECATED: No config file found, but ARVADOS_API_HOST and ARVADOS_API_TOKEN environment variables are set. Please use a config file instead.")
  80                 cfg.Client.APIHost = h
  81                 cfg.Client.AuthToken = t
  82                 if regexp.MustCompile("^(?i:1|yes|true)$").MatchString(os.Getenv("ARVADOS_API_HOST_INSECURE")) {
  83                         cfg.Client.Insecure = true
  84                 }
  85                 if y, err := yaml.Marshal(cfg); err == nil && !*dumpConfig {
  86                         log.Print("Current configuration:\n", string(y))
  87                 }
  88                 cfg.Timeout = arvados.Duration(time.Duration(*timeoutSeconds) * time.Second)
  89         }
  90
  91         if *dumpConfig {
  92                 log.Fatal(config.DumpAndExit(cfg))
  93         }
  94
  95         arv, err := arvadosclient.New(&cfg.Client)
  96         if err != nil {
  97                 log.Fatalf("Error setting up arvados client %s", err.Error())
  98         }
  99
 100         if cfg.Debug {
 101                 keepclient.DebugPrintf = log.Printf
 102         }
 103         kc, err := keepclient.MakeKeepClient(arv)
 104         if err != nil {
 105                 log.Fatalf("Error setting up keep client %s", err.Error())
 106         }
 107
 108         if cfg.PIDFile != "" {
 109                 f, err := os.Create(cfg.PIDFile)
 110                 if err != nil {
 111                         log.Fatal(err)
 112                 }
 113                 defer f.Close()
 114                 err = syscall.Flock(int(f.Fd()), syscall.LOCK_EX|syscall.LOCK_NB)
 115                 if err != nil {
 116                         log.Fatalf("flock(%s): %s", cfg.PIDFile, err)
 117                 }
 118                 defer os.Remove(cfg.PIDFile)
 119                 err = f.Truncate(0)
 120                 if err != nil {
 121                         log.Fatalf("truncate(%s): %s", cfg.PIDFile, err)
 122                 }
 123                 _, err = fmt.Fprint(f, os.Getpid())
 124                 if err != nil {
 125                         log.Fatalf("write(%s): %s", cfg.PIDFile, err)
 126                 }
 127                 err = f.Sync()
 128                 if err != nil {
 129                         log.Fatal("sync(%s): %s", cfg.PIDFile, err)
 130                 }
 131         }
 132
 133         if cfg.DefaultReplicas > 0 {
 134                 kc.Want_replicas = cfg.DefaultReplicas
 135         }
 136         go kc.RefreshServices(5*time.Minute, 3*time.Second)
 137
 138         listener, err = net.Listen("tcp", cfg.Listen)
 139         if err != nil {
 140                 log.Fatalf("listen(%s): %s", cfg.Listen, err)
 141         }
 142         if _, err := daemon.SdNotify(false, "READY=1"); err != nil {
 143                 log.Printf("Error notifying init daemon: %v", err)
 144         }
 145         log.Println("Listening at", listener.Addr())
 146
 147         // Shut down the server gracefully (by closing the listener)
 148         // if SIGTERM is received.
 149         term := make(chan os.Signal, 1)
 150         go func(sig <-chan os.Signal) {
 151                 s := <-sig
 152                 log.Println("caught signal:", s)
 153                 listener.Close()
 154         }(term)
 155         signal.Notify(term, syscall.SIGTERM)
 156         signal.Notify(term, syscall.SIGINT)
 157
 158         // Start serving requests.
 159         router = MakeRESTRouter(!cfg.DisableGet, !cfg.DisablePut, kc, time.Duration(cfg.Timeout))
 160         http.Serve(listener, router)
 161
 162         log.Println("shutting down")
 163 }
 164
 165 type ApiTokenCache struct {
 166         tokens     map[string]int64
 167         lock       sync.Mutex
 168         expireTime int64
 169 }
 170
 171 // Cache the token and set an expire time.  If we already have an expire time
 172 // on the token, it is not updated.
 173 func (this *ApiTokenCache) RememberToken(token string) {
 174         this.lock.Lock()
 175         defer this.lock.Unlock()
 176
 177         now := time.Now().Unix()
 178         if this.tokens[token] == 0 {
 179                 this.tokens[token] = now + this.expireTime
 180         }
 181 }
 182
 183 // Check if the cached token is known and still believed to be valid.
 184 func (this *ApiTokenCache) RecallToken(token string) bool {
 185         this.lock.Lock()
 186         defer this.lock.Unlock()
 187
 188         now := time.Now().Unix()
 189         if this.tokens[token] == 0 {
 190                 // Unknown token
 191                 return false
 192         } else if now < this.tokens[token] {
 193                 // Token is known and still valid
 194                 return true
 195         } else {
 196                 // Token is expired
 197                 this.tokens[token] = 0
 198                 return false
 199         }
 200 }
 201
 202 func GetRemoteAddress(req *http.Request) string {
 203         if xff := req.Header.Get("X-Forwarded-For"); xff != "" {
 204                 return xff + "," + req.RemoteAddr
 205         }
 206         return req.RemoteAddr
 207 }
 208
 209 func CheckAuthorizationHeader(kc *keepclient.KeepClient, cache *ApiTokenCache, req *http.Request) (pass bool, tok string) {
 210         var auth string
 211         if auth = req.Header.Get("Authorization"); auth == "" {
 212                 return false, ""
 213         }
 214
 215         _, err := fmt.Sscanf(auth, "OAuth2 %s", &tok)
 216         if err != nil {
 217                 // Scanning error
 218                 return false, ""
 219         }
 220
 221         if cache.RecallToken(tok) {
 222                 // Valid in the cache, short circuit
 223                 return true, tok
 224         }
 225
 226         arv := *kc.Arvados
 227         arv.ApiToken = tok
 228         if err := arv.Call("HEAD", "users", "", "current", nil, nil); err != nil {
 229                 log.Printf("%s: CheckAuthorizationHeader error: %v", GetRemoteAddress(req), err)
 230                 return false, ""
 231         }
 232
 233         // Success!  Update cache
 234         cache.RememberToken(tok)
 235
 236         return true, tok
 237 }
 238
 239 type proxyHandler struct {
 240         http.Handler
 241         *keepclient.KeepClient
 242         *ApiTokenCache
 243         timeout   time.Duration
 244         transport *http.Transport
 245 }
 246
 247 // MakeRESTRouter returns an http.Handler that passes GET and PUT
 248 // requests to the appropriate handlers.
 249 func MakeRESTRouter(enable_get bool, enable_put bool, kc *keepclient.KeepClient, timeout time.Duration) http.Handler {
 250         rest := mux.NewRouter()
 251         h := &proxyHandler{
 252                 Handler:    rest,
 253                 KeepClient: kc,
 254                 timeout:    timeout,
 255                 transport: &http.Transport{
 256                         Dial: (&net.Dialer{
 257                                 Timeout:   20 * time.Second,
 258                                 KeepAlive: 10 * time.Second,
 259                         }).Dial,
 260                         TLSClientConfig:     arvadosclient.MakeTLSConfig(kc.Arvados.ApiInsecure),
 261                         TLSHandshakeTimeout: 10 * time.Second,
 262                 },
 263                 ApiTokenCache: &ApiTokenCache{
 264                         tokens:     make(map[string]int64),
 265                         expireTime: 300,
 266                 },
 267         }
 268
 269         go func(t *http.Transport) {
 270                 for range time.NewTicker(5 * time.Minute).C {
 271                         t.CloseIdleConnections()
 272                 }
 273         }(h.transport)
 274
 275         if enable_get {
 276                 rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Get).Methods("GET", "HEAD")
 277                 rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Get).Methods("GET", "HEAD")
 278
 279                 // List all blocks
 280                 rest.HandleFunc(`/index`, h.Index).Methods("GET")
 281
 282                 // List blocks whose hash has the given prefix
 283                 rest.HandleFunc(`/index/{prefix:[0-9a-f]{0,32}}`, h.Index).Methods("GET")
 284         }
 285
 286         if enable_put {
 287                 rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Put).Methods("PUT")
 288                 rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Put).Methods("PUT")
 289                 rest.HandleFunc(`/`, h.Put).Methods("POST")
 290                 rest.HandleFunc(`/{any}`, h.Options).Methods("OPTIONS")
 291                 rest.HandleFunc(`/`, h.Options).Methods("OPTIONS")
 292         }
 293
 294         rest.NotFoundHandler = InvalidPathHandler{}
 295         return h
 296 }
 297
 298 var errLoopDetected = errors.New("loop detected")
 299
 300 func (*proxyHandler) checkLoop(resp http.ResponseWriter, req *http.Request) error {
 301         if via := req.Header.Get("Via"); strings.Index(via, " "+viaAlias) >= 0 {
 302                 log.Printf("proxy loop detected (request has Via: %q): perhaps keepproxy is misidentified by gateway config as an external client, or its keep_services record does not have service_type=proxy?", via)
 303                 http.Error(resp, errLoopDetected.Error(), http.StatusInternalServerError)
 304                 return errLoopDetected
 305         }
 306         return nil
 307 }
 308
 309 func SetCorsHeaders(resp http.ResponseWriter) {
 310         resp.Header().Set("Access-Control-Allow-Methods", "GET, HEAD, POST, PUT, OPTIONS")
 311         resp.Header().Set("Access-Control-Allow-Origin", "*")
 312         resp.Header().Set("Access-Control-Allow-Headers", "Authorization, Content-Length, Content-Type, X-Keep-Desired-Replicas")
 313         resp.Header().Set("Access-Control-Max-Age", "86486400")
 314 }
 315
 316 type InvalidPathHandler struct{}
 317
 318 func (InvalidPathHandler) ServeHTTP(resp http.ResponseWriter, req *http.Request) {
 319         log.Printf("%s: %s %s unroutable", GetRemoteAddress(req), req.Method, req.URL.Path)
 320         http.Error(resp, "Bad request", http.StatusBadRequest)
 321 }
 322
 323 func (h *proxyHandler) Options(resp http.ResponseWriter, req *http.Request) {
 324         log.Printf("%s: %s %s", GetRemoteAddress(req), req.Method, req.URL.Path)
 325         SetCorsHeaders(resp)
 326 }
 327
 328 var BadAuthorizationHeader = errors.New("Missing or invalid Authorization header")
 329 var ContentLengthMismatch = errors.New("Actual length != expected content length")
 330 var MethodNotSupported = errors.New("Method not supported")
 331
 332 var removeHint, _ = regexp.Compile("\\+K@[a-z0-9]{5}(\\+|$)")
 333
 334 func (h *proxyHandler) Get(resp http.ResponseWriter, req *http.Request) {
 335         if err := h.checkLoop(resp, req); err != nil {
 336                 return
 337         }
 338         SetCorsHeaders(resp)
 339         resp.Header().Set("Via", req.Proto+" "+viaAlias)
 340
 341         locator := mux.Vars(req)["locator"]
 342         var err error
 343         var status int
 344         var expectLength, responseLength int64
 345         var proxiedURI = "-"
 346
 347         defer func() {
 348                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, responseLength, proxiedURI, err)
 349                 if status != http.StatusOK {
 350                         http.Error(resp, err.Error(), status)
 351                 }
 352         }()
 353
 354         kc := h.makeKeepClient(req)
 355
 356         var pass bool
 357         var tok string
 358         if pass, tok = CheckAuthorizationHeader(kc, h.ApiTokenCache, req); !pass {
 359                 status, err = http.StatusForbidden, BadAuthorizationHeader
 360                 return
 361         }
 362
 363         // Copy ArvadosClient struct and use the client's API token
 364         arvclient := *kc.Arvados
 365         arvclient.ApiToken = tok
 366         kc.Arvados = &arvclient
 367
 368         var reader io.ReadCloser
 369
 370         locator = removeHint.ReplaceAllString(locator, "$1")
 371
 372         switch req.Method {
 373         case "HEAD":
 374                 expectLength, proxiedURI, err = kc.Ask(locator)
 375         case "GET":
 376                 reader, expectLength, proxiedURI, err = kc.Get(locator)
 377                 if reader != nil {
 378                         defer reader.Close()
 379                 }
 380         default:
 381                 status, err = http.StatusNotImplemented, MethodNotSupported
 382                 return
 383         }
 384
 385         if expectLength == -1 {
 386                 log.Println("Warning:", GetRemoteAddress(req), req.Method, proxiedURI, "Content-Length not provided")
 387         }
 388
 389         switch respErr := err.(type) {
 390         case nil:
 391                 status = http.StatusOK
 392                 resp.Header().Set("Content-Length", fmt.Sprint(expectLength))
 393                 switch req.Method {
 394                 case "HEAD":
 395                         responseLength = 0
 396                 case "GET":
 397                         responseLength, err = io.Copy(resp, reader)
 398                         if err == nil && expectLength > -1 && responseLength != expectLength {
 399                                 err = ContentLengthMismatch
 400                         }
 401                 }
 402         case keepclient.Error:
 403                 if respErr == keepclient.BlockNotFound {
 404                         status = http.StatusNotFound
 405                 } else if respErr.Temporary() {
 406                         status = http.StatusBadGateway
 407                 } else {
 408                         status = 422
 409                 }
 410         default:
 411                 status = http.StatusInternalServerError
 412         }
 413 }
 414
 415 var LengthRequiredError = errors.New(http.StatusText(http.StatusLengthRequired))
 416 var LengthMismatchError = errors.New("Locator size hint does not match Content-Length header")
 417
 418 func (h *proxyHandler) Put(resp http.ResponseWriter, req *http.Request) {
 419         if err := h.checkLoop(resp, req); err != nil {
 420                 return
 421         }
 422         SetCorsHeaders(resp)
 423         resp.Header().Set("Via", "HTTP/1.1 "+viaAlias)
 424
 425         kc := h.makeKeepClient(req)
 426
 427         var err error
 428         var expectLength int64
 429         var status = http.StatusInternalServerError
 430         var wroteReplicas int
 431         var locatorOut string = "-"
 432
 433         defer func() {
 434                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, kc.Want_replicas, wroteReplicas, locatorOut, err)
 435                 if status != http.StatusOK {
 436                         http.Error(resp, err.Error(), status)
 437                 }
 438         }()
 439
 440         locatorIn := mux.Vars(req)["locator"]
 441
 442         _, err = fmt.Sscanf(req.Header.Get("Content-Length"), "%d", &expectLength)
 443         if err != nil || expectLength < 0 {
 444                 err = LengthRequiredError
 445                 status = http.StatusLengthRequired
 446                 return
 447         }
 448
 449         if locatorIn != "" {
 450                 var loc *keepclient.Locator
 451                 if loc, err = keepclient.MakeLocator(locatorIn); err != nil {
 452                         status = http.StatusBadRequest
 453                         return
 454                 } else if loc.Size > 0 && int64(loc.Size) != expectLength {
 455                         err = LengthMismatchError
 456                         status = http.StatusBadRequest
 457                         return
 458                 }
 459         }
 460
 461         var pass bool
 462         var tok string
 463         if pass, tok = CheckAuthorizationHeader(kc, h.ApiTokenCache, req); !pass {
 464                 err = BadAuthorizationHeader
 465                 status = http.StatusForbidden
 466                 return
 467         }
 468
 469         // Copy ArvadosClient struct and use the client's API token
 470         arvclient := *kc.Arvados
 471         arvclient.ApiToken = tok
 472         kc.Arvados = &arvclient
 473
 474         // Check if the client specified the number of replicas
 475         if req.Header.Get("X-Keep-Desired-Replicas") != "" {
 476                 var r int
 477                 _, err := fmt.Sscanf(req.Header.Get(keepclient.X_Keep_Desired_Replicas), "%d", &r)
 478                 if err == nil {
 479                         kc.Want_replicas = r
 480                 }
 481         }
 482
 483         // Now try to put the block through
 484         if locatorIn == "" {
 485                 if bytes, err := ioutil.ReadAll(req.Body); err != nil {
 486                         err = errors.New(fmt.Sprintf("Error reading request body: %s", err))
 487                         status = http.StatusInternalServerError
 488                         return
 489                 } else {
 490                         locatorOut, wroteReplicas, err = kc.PutB(bytes)
 491                 }
 492         } else {
 493                 locatorOut, wroteReplicas, err = kc.PutHR(locatorIn, req.Body, expectLength)
 494         }
 495
 496         // Tell the client how many successful PUTs we accomplished
 497         resp.Header().Set(keepclient.X_Keep_Replicas_Stored, fmt.Sprintf("%d", wroteReplicas))
 498
 499         switch err.(type) {
 500         case nil:
 501                 status = http.StatusOK
 502                 _, err = io.WriteString(resp, locatorOut)
 503
 504         case keepclient.OversizeBlockError:
 505                 // Too much data
 506                 status = http.StatusRequestEntityTooLarge
 507
 508         case keepclient.InsufficientReplicasError:
 509                 if wroteReplicas > 0 {
 510                         // At least one write is considered success.  The
 511                         // client can decide if getting less than the number of
 512                         // replications it asked for is a fatal error.
 513                         status = http.StatusOK
 514                         _, err = io.WriteString(resp, locatorOut)
 515                 } else {
 516                         status = http.StatusServiceUnavailable
 517                 }
 518
 519         default:
 520                 status = http.StatusBadGateway
 521         }
 522 }
 523
 524 // ServeHTTP implementation for IndexHandler
 525 // Supports only GET requests for /index/{prefix:[0-9a-f]{0,32}}
 526 // For each keep server found in LocalRoots:
 527 //   Invokes GetIndex using keepclient
 528 //   Expects "complete" response (terminating with blank new line)
 529 //   Aborts on any errors
 530 // Concatenates responses from all those keep servers and returns
 531 func (h *proxyHandler) Index(resp http.ResponseWriter, req *http.Request) {
 532         SetCorsHeaders(resp)
 533
 534         prefix := mux.Vars(req)["prefix"]
 535         var err error
 536         var status int
 537
 538         defer func() {
 539                 if status != http.StatusOK {
 540                         http.Error(resp, err.Error(), status)
 541                 }
 542         }()
 543
 544         kc := h.makeKeepClient(req)
 545         ok, token := CheckAuthorizationHeader(kc, h.ApiTokenCache, req)
 546         if !ok {
 547                 status, err = http.StatusForbidden, BadAuthorizationHeader
 548                 return
 549         }
 550
 551         // Copy ArvadosClient struct and use the client's API token
 552         arvclient := *kc.Arvados
 553         arvclient.ApiToken = token
 554         kc.Arvados = &arvclient
 555
 556         // Only GET method is supported
 557         if req.Method != "GET" {
 558                 status, err = http.StatusNotImplemented, MethodNotSupported
 559                 return
 560         }
 561
 562         // Get index from all LocalRoots and write to resp
 563         var reader io.Reader
 564         for uuid := range kc.LocalRoots() {
 565                 reader, err = kc.GetIndex(uuid, prefix)
 566                 if err != nil {
 567                         status = http.StatusBadGateway
 568                         return
 569                 }
 570
 571                 _, err = io.Copy(resp, reader)
 572                 if err != nil {
 573                         status = http.StatusBadGateway
 574                         return
 575                 }
 576         }
 577
 578         // Got index from all the keep servers and wrote to resp
 579         status = http.StatusOK
 580         resp.Write([]byte("\n"))
 581 }
 582
 583 func (h *proxyHandler) makeKeepClient(req *http.Request) *keepclient.KeepClient {
 584         kc := *h.KeepClient
 585         kc.HTTPClient = &proxyClient{
 586                 client: &http.Client{
 587                         Timeout:   h.timeout,
 588                         Transport: h.transport,
 589                 },
 590                 proto: req.Proto,
 591         }
 592         return &kc
 593 }