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         kc.Client.(*http.Client).Timeout = time.Duration(cfg.Timeout)
 137         go kc.RefreshServices(5*time.Minute, 3*time.Second)
 138
 139         listener, err = net.Listen("tcp", cfg.Listen)
 140         if err != nil {
 141                 log.Fatalf("listen(%s): %s", cfg.Listen, err)
 142         }
 143         if _, err := daemon.SdNotify(false, "READY=1"); err != nil {
 144                 log.Printf("Error notifying init daemon: %v", err)
 145         }
 146         log.Println("Listening at", listener.Addr())
 147
 148         // Shut down the server gracefully (by closing the listener)
 149         // if SIGTERM is received.
 150         term := make(chan os.Signal, 1)
 151         go func(sig <-chan os.Signal) {
 152                 s := <-sig
 153                 log.Println("caught signal:", s)
 154                 listener.Close()
 155         }(term)
 156         signal.Notify(term, syscall.SIGTERM)
 157         signal.Notify(term, syscall.SIGINT)
 158
 159         // Start serving requests.
 160         router = MakeRESTRouter(!cfg.DisableGet, !cfg.DisablePut, kc)
 161         http.Serve(listener, router)
 162
 163         log.Println("shutting down")
 164 }
 165
 166 type ApiTokenCache struct {
 167         tokens     map[string]int64
 168         lock       sync.Mutex
 169         expireTime int64
 170 }
 171
 172 // Cache the token and set an expire time.  If we already have an expire time
 173 // on the token, it is not updated.
 174 func (this *ApiTokenCache) RememberToken(token string) {
 175         this.lock.Lock()
 176         defer this.lock.Unlock()
 177
 178         now := time.Now().Unix()
 179         if this.tokens[token] == 0 {
 180                 this.tokens[token] = now + this.expireTime
 181         }
 182 }
 183
 184 // Check if the cached token is known and still believed to be valid.
 185 func (this *ApiTokenCache) RecallToken(token string) bool {
 186         this.lock.Lock()
 187         defer this.lock.Unlock()
 188
 189         now := time.Now().Unix()
 190         if this.tokens[token] == 0 {
 191                 // Unknown token
 192                 return false
 193         } else if now < this.tokens[token] {
 194                 // Token is known and still valid
 195                 return true
 196         } else {
 197                 // Token is expired
 198                 this.tokens[token] = 0
 199                 return false
 200         }
 201 }
 202
 203 func GetRemoteAddress(req *http.Request) string {
 204         if xff := req.Header.Get("X-Forwarded-For"); xff != "" {
 205                 return xff + "," + req.RemoteAddr
 206         }
 207         return req.RemoteAddr
 208 }
 209
 210 func CheckAuthorizationHeader(kc *keepclient.KeepClient, cache *ApiTokenCache, req *http.Request) (pass bool, tok string) {
 211         var auth string
 212         if auth = req.Header.Get("Authorization"); auth == "" {
 213                 return false, ""
 214         }
 215
 216         _, err := fmt.Sscanf(auth, "OAuth2 %s", &tok)
 217         if err != nil {
 218                 // Scanning error
 219                 return false, ""
 220         }
 221
 222         if cache.RecallToken(tok) {
 223                 // Valid in the cache, short circuit
 224                 return true, tok
 225         }
 226
 227         arv := *kc.Arvados
 228         arv.ApiToken = tok
 229         if err := arv.Call("HEAD", "users", "", "current", nil, nil); err != nil {
 230                 log.Printf("%s: CheckAuthorizationHeader error: %v", GetRemoteAddress(req), err)
 231                 return false, ""
 232         }
 233
 234         // Success!  Update cache
 235         cache.RememberToken(tok)
 236
 237         return true, tok
 238 }
 239
 240 type proxyHandler struct {
 241         http.Handler
 242         *keepclient.KeepClient
 243         *ApiTokenCache
 244 }
 245
 246 // MakeRESTRouter returns an http.Handler that passes GET and PUT
 247 // requests to the appropriate handlers.
 248 func MakeRESTRouter(enable_get bool, enable_put bool, kc *keepclient.KeepClient) http.Handler {
 249         rest := mux.NewRouter()
 250         h := &proxyHandler{
 251                 Handler:    rest,
 252                 KeepClient: kc,
 253                 ApiTokenCache: &ApiTokenCache{
 254                         tokens:     make(map[string]int64),
 255                         expireTime: 300,
 256                 },
 257         }
 258
 259         if enable_get {
 260                 rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Get).Methods("GET", "HEAD")
 261                 rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Get).Methods("GET", "HEAD")
 262
 263                 // List all blocks
 264                 rest.HandleFunc(`/index`, h.Index).Methods("GET")
 265
 266                 // List blocks whose hash has the given prefix
 267                 rest.HandleFunc(`/index/{prefix:[0-9a-f]{0,32}}`, h.Index).Methods("GET")
 268         }
 269
 270         if enable_put {
 271                 rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Put).Methods("PUT")
 272                 rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Put).Methods("PUT")
 273                 rest.HandleFunc(`/`, h.Put).Methods("POST")
 274                 rest.HandleFunc(`/{any}`, h.Options).Methods("OPTIONS")
 275                 rest.HandleFunc(`/`, h.Options).Methods("OPTIONS")
 276         }
 277
 278         rest.NotFoundHandler = InvalidPathHandler{}
 279         return h
 280 }
 281
 282 var errLoopDetected = errors.New("loop detected")
 283
 284 func (*proxyHandler) checkLoop(resp http.ResponseWriter, req *http.Request) error {
 285         if via := req.Header.Get("Via"); strings.Index(via, " "+viaAlias) >= 0 {
 286                 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)
 287                 http.Error(resp, errLoopDetected.Error(), http.StatusInternalServerError)
 288                 return errLoopDetected
 289         }
 290         return nil
 291 }
 292
 293 func SetCorsHeaders(resp http.ResponseWriter) {
 294         resp.Header().Set("Access-Control-Allow-Methods", "GET, HEAD, POST, PUT, OPTIONS")
 295         resp.Header().Set("Access-Control-Allow-Origin", "*")
 296         resp.Header().Set("Access-Control-Allow-Headers", "Authorization, Content-Length, Content-Type, X-Keep-Desired-Replicas")
 297         resp.Header().Set("Access-Control-Max-Age", "86486400")
 298 }
 299
 300 type InvalidPathHandler struct{}
 301
 302 func (InvalidPathHandler) ServeHTTP(resp http.ResponseWriter, req *http.Request) {
 303         log.Printf("%s: %s %s unroutable", GetRemoteAddress(req), req.Method, req.URL.Path)
 304         http.Error(resp, "Bad request", http.StatusBadRequest)
 305 }
 306
 307 func (h *proxyHandler) Options(resp http.ResponseWriter, req *http.Request) {
 308         log.Printf("%s: %s %s", GetRemoteAddress(req), req.Method, req.URL.Path)
 309         SetCorsHeaders(resp)
 310 }
 311
 312 var BadAuthorizationHeader = errors.New("Missing or invalid Authorization header")
 313 var ContentLengthMismatch = errors.New("Actual length != expected content length")
 314 var MethodNotSupported = errors.New("Method not supported")
 315
 316 var removeHint, _ = regexp.Compile("\\+K@[a-z0-9]{5}(\\+|$)")
 317
 318 func (h *proxyHandler) Get(resp http.ResponseWriter, req *http.Request) {
 319         if err := h.checkLoop(resp, req); err != nil {
 320                 return
 321         }
 322         SetCorsHeaders(resp)
 323         resp.Header().Set("Via", req.Proto+" "+viaAlias)
 324
 325         locator := mux.Vars(req)["locator"]
 326         var err error
 327         var status int
 328         var expectLength, responseLength int64
 329         var proxiedURI = "-"
 330
 331         defer func() {
 332                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, responseLength, proxiedURI, err)
 333                 if status != http.StatusOK {
 334                         http.Error(resp, err.Error(), status)
 335                 }
 336         }()
 337
 338         kc := *h.KeepClient
 339         kc.Client = &proxyClient{client: kc.Client, proto: req.Proto}
 340
 341         var pass bool
 342         var tok string
 343         if pass, tok = CheckAuthorizationHeader(&kc, h.ApiTokenCache, req); !pass {
 344                 status, err = http.StatusForbidden, BadAuthorizationHeader
 345                 return
 346         }
 347
 348         // Copy ArvadosClient struct and use the client's API token
 349         arvclient := *kc.Arvados
 350         arvclient.ApiToken = tok
 351         kc.Arvados = &arvclient
 352
 353         var reader io.ReadCloser
 354
 355         locator = removeHint.ReplaceAllString(locator, "$1")
 356
 357         switch req.Method {
 358         case "HEAD":
 359                 expectLength, proxiedURI, err = kc.Ask(locator)
 360         case "GET":
 361                 reader, expectLength, proxiedURI, err = kc.Get(locator)
 362                 if reader != nil {
 363                         defer reader.Close()
 364                 }
 365         default:
 366                 status, err = http.StatusNotImplemented, MethodNotSupported
 367                 return
 368         }
 369
 370         if expectLength == -1 {
 371                 log.Println("Warning:", GetRemoteAddress(req), req.Method, proxiedURI, "Content-Length not provided")
 372         }
 373
 374         switch respErr := err.(type) {
 375         case nil:
 376                 status = http.StatusOK
 377                 resp.Header().Set("Content-Length", fmt.Sprint(expectLength))
 378                 switch req.Method {
 379                 case "HEAD":
 380                         responseLength = 0
 381                 case "GET":
 382                         responseLength, err = io.Copy(resp, reader)
 383                         if err == nil && expectLength > -1 && responseLength != expectLength {
 384                                 err = ContentLengthMismatch
 385                         }
 386                 }
 387         case keepclient.Error:
 388                 if respErr == keepclient.BlockNotFound {
 389                         status = http.StatusNotFound
 390                 } else if respErr.Temporary() {
 391                         status = http.StatusBadGateway
 392                 } else {
 393                         status = 422
 394                 }
 395         default:
 396                 status = http.StatusInternalServerError
 397         }
 398 }
 399
 400 var LengthRequiredError = errors.New(http.StatusText(http.StatusLengthRequired))
 401 var LengthMismatchError = errors.New("Locator size hint does not match Content-Length header")
 402
 403 func (h *proxyHandler) Put(resp http.ResponseWriter, req *http.Request) {
 404         if err := h.checkLoop(resp, req); err != nil {
 405                 return
 406         }
 407         SetCorsHeaders(resp)
 408         resp.Header().Set("Via", "HTTP/1.1 "+viaAlias)
 409
 410         kc := *h.KeepClient
 411         kc.Client = &proxyClient{client: kc.Client, proto: req.Proto}
 412
 413         var err error
 414         var expectLength int64
 415         var status = http.StatusInternalServerError
 416         var wroteReplicas int
 417         var locatorOut string = "-"
 418
 419         defer func() {
 420                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, kc.Want_replicas, wroteReplicas, locatorOut, err)
 421                 if status != http.StatusOK {
 422                         http.Error(resp, err.Error(), status)
 423                 }
 424         }()
 425
 426         locatorIn := mux.Vars(req)["locator"]
 427
 428         _, err = fmt.Sscanf(req.Header.Get("Content-Length"), "%d", &expectLength)
 429         if err != nil || expectLength < 0 {
 430                 err = LengthRequiredError
 431                 status = http.StatusLengthRequired
 432                 return
 433         }
 434
 435         if locatorIn != "" {
 436                 var loc *keepclient.Locator
 437                 if loc, err = keepclient.MakeLocator(locatorIn); err != nil {
 438                         status = http.StatusBadRequest
 439                         return
 440                 } else if loc.Size > 0 && int64(loc.Size) != expectLength {
 441                         err = LengthMismatchError
 442                         status = http.StatusBadRequest
 443                         return
 444                 }
 445         }
 446
 447         var pass bool
 448         var tok string
 449         if pass, tok = CheckAuthorizationHeader(&kc, h.ApiTokenCache, req); !pass {
 450                 err = BadAuthorizationHeader
 451                 status = http.StatusForbidden
 452                 return
 453         }
 454
 455         // Copy ArvadosClient struct and use the client's API token
 456         arvclient := *kc.Arvados
 457         arvclient.ApiToken = tok
 458         kc.Arvados = &arvclient
 459
 460         // Check if the client specified the number of replicas
 461         if req.Header.Get("X-Keep-Desired-Replicas") != "" {
 462                 var r int
 463                 _, err := fmt.Sscanf(req.Header.Get(keepclient.X_Keep_Desired_Replicas), "%d", &r)
 464                 if err == nil {
 465                         kc.Want_replicas = r
 466                 }
 467         }
 468
 469         // Now try to put the block through
 470         if locatorIn == "" {
 471                 if bytes, err := ioutil.ReadAll(req.Body); err != nil {
 472                         err = errors.New(fmt.Sprintf("Error reading request body: %s", err))
 473                         status = http.StatusInternalServerError
 474                         return
 475                 } else {
 476                         locatorOut, wroteReplicas, err = kc.PutB(bytes)
 477                 }
 478         } else {
 479                 locatorOut, wroteReplicas, err = kc.PutHR(locatorIn, req.Body, expectLength)
 480         }
 481
 482         // Tell the client how many successful PUTs we accomplished
 483         resp.Header().Set(keepclient.X_Keep_Replicas_Stored, fmt.Sprintf("%d", wroteReplicas))
 484
 485         switch err.(type) {
 486         case nil:
 487                 status = http.StatusOK
 488                 _, err = io.WriteString(resp, locatorOut)
 489
 490         case keepclient.OversizeBlockError:
 491                 // Too much data
 492                 status = http.StatusRequestEntityTooLarge
 493
 494         case keepclient.InsufficientReplicasError:
 495                 if wroteReplicas > 0 {
 496                         // At least one write is considered success.  The
 497                         // client can decide if getting less than the number of
 498                         // replications it asked for is a fatal error.
 499                         status = http.StatusOK
 500                         _, err = io.WriteString(resp, locatorOut)
 501                 } else {
 502                         status = http.StatusServiceUnavailable
 503                 }
 504
 505         default:
 506                 status = http.StatusBadGateway
 507         }
 508 }
 509
 510 // ServeHTTP implementation for IndexHandler
 511 // Supports only GET requests for /index/{prefix:[0-9a-f]{0,32}}
 512 // For each keep server found in LocalRoots:
 513 //   Invokes GetIndex using keepclient
 514 //   Expects "complete" response (terminating with blank new line)
 515 //   Aborts on any errors
 516 // Concatenates responses from all those keep servers and returns
 517 func (h *proxyHandler) Index(resp http.ResponseWriter, req *http.Request) {
 518         SetCorsHeaders(resp)
 519
 520         prefix := mux.Vars(req)["prefix"]
 521         var err error
 522         var status int
 523
 524         defer func() {
 525                 if status != http.StatusOK {
 526                         http.Error(resp, err.Error(), status)
 527                 }
 528         }()
 529
 530         kc := *h.KeepClient
 531
 532         ok, token := CheckAuthorizationHeader(&kc, h.ApiTokenCache, req)
 533         if !ok {
 534                 status, err = http.StatusForbidden, BadAuthorizationHeader
 535                 return
 536         }
 537
 538         // Copy ArvadosClient struct and use the client's API token
 539         arvclient := *kc.Arvados
 540         arvclient.ApiToken = token
 541         kc.Arvados = &arvclient
 542
 543         // Only GET method is supported
 544         if req.Method != "GET" {
 545                 status, err = http.StatusNotImplemented, MethodNotSupported
 546                 return
 547         }
 548
 549         // Get index from all LocalRoots and write to resp
 550         var reader io.Reader
 551         for uuid := range kc.LocalRoots() {
 552                 reader, err = kc.GetIndex(uuid, prefix)
 553                 if err != nil {
 554                         status = http.StatusBadGateway
 555                         return
 556                 }
 557
 558                 _, err = io.Copy(resp, reader)
 559                 if err != nil {
 560                         status = http.StatusBadGateway
 561                         return
 562                 }
 563         }
 564
 565         // Got index from all the keep servers and wrote to resp
 566         status = http.StatusOK
 567         resp.Write([]byte("\n"))
 568 }