services/keepproxy/keepproxy.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package main
   6
   7 import (
   8         "errors"
   9         "flag"
  10         "fmt"
  11         "io"
  12         "io/ioutil"
  13         "net"
  14         "net/http"
  15         "os"
  16         "os/signal"
  17         "regexp"
  18         "strings"
  19         "sync"
  20         "syscall"
  21         "time"
  22
  23         "git.arvados.org/arvados.git/lib/config"
  24         "git.arvados.org/arvados.git/sdk/go/arvados"
  25         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  26         "git.arvados.org/arvados.git/sdk/go/health"
  27         "git.arvados.org/arvados.git/sdk/go/httpserver"
  28         "git.arvados.org/arvados.git/sdk/go/keepclient"
  29         "github.com/coreos/go-systemd/daemon"
  30         "github.com/ghodss/yaml"
  31         "github.com/gorilla/mux"
  32         log "github.com/sirupsen/logrus"
  33 )
  34
  35 var version = "dev"
  36
  37 var (
  38         listener net.Listener
  39         router   http.Handler
  40 )
  41
  42 const rfc3339NanoFixed = "2006-01-02T15:04:05.000000000Z07:00"
  43
  44 func configure(logger log.FieldLogger, args []string) (*arvados.Cluster, error) {
  45         flags := flag.NewFlagSet(args[0], flag.ExitOnError)
  46
  47         dumpConfig := flags.Bool("dump-config", false, "write current configuration to stdout and exit")
  48         getVersion := flags.Bool("version", false, "Print version information and exit.")
  49
  50         loader := config.NewLoader(os.Stdin, logger)
  51         loader.SetupFlags(flags)
  52
  53         args = loader.MungeLegacyConfigArgs(logger, args[1:], "-legacy-keepproxy-config")
  54         flags.Parse(args)
  55
  56         // Print version information if requested
  57         if *getVersion {
  58                 fmt.Printf("keepproxy %s\n", version)
  59                 return nil, nil
  60         }
  61
  62         cfg, err := loader.Load()
  63         if err != nil {
  64                 return nil, err
  65         }
  66         cluster, err := cfg.GetCluster("")
  67         if err != nil {
  68                 return nil, err
  69         }
  70
  71         if *dumpConfig {
  72                 out, err := yaml.Marshal(cfg)
  73                 if err != nil {
  74                         return nil, err
  75                 }
  76                 if _, err := os.Stdout.Write(out); err != nil {
  77                         return nil, err
  78                 }
  79                 return nil, nil
  80         }
  81         return cluster, nil
  82 }
  83
  84 func main() {
  85         logger := log.New()
  86         logger.Formatter = &log.JSONFormatter{
  87                 TimestampFormat: rfc3339NanoFixed,
  88         }
  89
  90         cluster, err := configure(logger, os.Args)
  91         if err != nil {
  92                 log.Fatal(err)
  93         }
  94         if cluster == nil {
  95                 return
  96         }
  97
  98         log.Printf("keepproxy %s started", version)
  99
 100         if err := run(logger, cluster); err != nil {
 101                 log.Fatal(err)
 102         }
 103
 104         log.Println("shutting down")
 105 }
 106
 107 func run(logger log.FieldLogger, cluster *arvados.Cluster) error {
 108         client, err := arvados.NewClientFromConfig(cluster)
 109         if err != nil {
 110                 return err
 111         }
 112         client.AuthToken = cluster.SystemRootToken
 113
 114         arv, err := arvadosclient.New(client)
 115         if err != nil {
 116                 return fmt.Errorf("Error setting up arvados client %v", err)
 117         }
 118
 119         // If a config file is available, use the keepstores defined there
 120         // instead of the legacy autodiscover mechanism via the API server
 121         for k := range cluster.Services.Keepstore.InternalURLs {
 122                 arv.KeepServiceURIs = append(arv.KeepServiceURIs, strings.TrimRight(k.String(), "/"))
 123         }
 124
 125         if cluster.SystemLogs.LogLevel == "debug" {
 126                 keepclient.DebugPrintf = log.Printf
 127         }
 128         kc, err := keepclient.MakeKeepClient(arv)
 129         if err != nil {
 130                 return fmt.Errorf("Error setting up keep client %v", err)
 131         }
 132         keepclient.RefreshServiceDiscoveryOnSIGHUP()
 133
 134         if cluster.Collections.DefaultReplication > 0 {
 135                 kc.Want_replicas = cluster.Collections.DefaultReplication
 136         }
 137
 138         var listen arvados.URL
 139         for listen = range cluster.Services.Keepproxy.InternalURLs {
 140                 break
 141         }
 142
 143         var lErr error
 144         listener, lErr = net.Listen("tcp", listen.Host)
 145         if lErr != nil {
 146                 return fmt.Errorf("listen(%s): %v", listen.Host, lErr)
 147         }
 148
 149         if _, err := daemon.SdNotify(false, "READY=1"); err != nil {
 150                 log.Printf("Error notifying init daemon: %v", err)
 151         }
 152         log.Println("listening at", listener.Addr())
 153
 154         // Shut down the server gracefully (by closing the listener)
 155         // if SIGTERM is received.
 156         term := make(chan os.Signal, 1)
 157         go func(sig <-chan os.Signal) {
 158                 s := <-sig
 159                 log.Println("caught signal:", s)
 160                 listener.Close()
 161         }(term)
 162         signal.Notify(term, syscall.SIGTERM)
 163         signal.Notify(term, syscall.SIGINT)
 164
 165         // Start serving requests.
 166         router = MakeRESTRouter(kc, time.Duration(keepclient.DefaultProxyRequestTimeout), cluster.ManagementToken)
 167         return http.Serve(listener, httpserver.AddRequestIDs(httpserver.LogRequests(router)))
 168 }
 169
 170 type ApiTokenCache struct {
 171         tokens     map[string]int64
 172         lock       sync.Mutex
 173         expireTime int64
 174 }
 175
 176 // Cache the token and set an expire time.  If we already have an expire time
 177 // on the token, it is not updated.
 178 func (this *ApiTokenCache) RememberToken(token string) {
 179         this.lock.Lock()
 180         defer this.lock.Unlock()
 181
 182         now := time.Now().Unix()
 183         if this.tokens[token] == 0 {
 184                 this.tokens[token] = now + this.expireTime
 185         }
 186 }
 187
 188 // Check if the cached token is known and still believed to be valid.
 189 func (this *ApiTokenCache) RecallToken(token string) bool {
 190         this.lock.Lock()
 191         defer this.lock.Unlock()
 192
 193         now := time.Now().Unix()
 194         if this.tokens[token] == 0 {
 195                 // Unknown token
 196                 return false
 197         } else if now < this.tokens[token] {
 198                 // Token is known and still valid
 199                 return true
 200         } else {
 201                 // Token is expired
 202                 this.tokens[token] = 0
 203                 return false
 204         }
 205 }
 206
 207 func GetRemoteAddress(req *http.Request) string {
 208         if xff := req.Header.Get("X-Forwarded-For"); xff != "" {
 209                 return xff + "," + req.RemoteAddr
 210         }
 211         return req.RemoteAddr
 212 }
 213
 214 func CheckAuthorizationHeader(kc *keepclient.KeepClient, cache *ApiTokenCache, req *http.Request) (pass bool, tok string) {
 215         parts := strings.SplitN(req.Header.Get("Authorization"), " ", 2)
 216         if len(parts) < 2 || !(parts[0] == "OAuth2" || parts[0] == "Bearer") || len(parts[1]) == 0 {
 217                 return false, ""
 218         }
 219         tok = parts[1]
 220
 221         // Tokens are validated differently depending on what kind of
 222         // operation is being performed. For example, tokens in
 223         // collection-sharing links permit GET requests, but not
 224         // PUT requests.
 225         var op string
 226         if req.Method == "GET" || req.Method == "HEAD" {
 227                 op = "read"
 228         } else {
 229                 op = "write"
 230         }
 231
 232         if cache.RecallToken(op + ":" + tok) {
 233                 // Valid in the cache, short circuit
 234                 return true, tok
 235         }
 236
 237         var err error
 238         arv := *kc.Arvados
 239         arv.ApiToken = tok
 240         arv.RequestID = req.Header.Get("X-Request-Id")
 241         if op == "read" {
 242                 err = arv.Call("HEAD", "keep_services", "", "accessible", nil, nil)
 243         } else {
 244                 err = arv.Call("HEAD", "users", "", "current", nil, nil)
 245         }
 246         if err != nil {
 247                 log.Printf("%s: CheckAuthorizationHeader error: %v", GetRemoteAddress(req), err)
 248                 return false, ""
 249         }
 250
 251         // Success!  Update cache
 252         cache.RememberToken(op + ":" + tok)
 253
 254         return true, tok
 255 }
 256
 257 // We need to make a private copy of the default http transport early
 258 // in initialization, then make copies of our private copy later. It
 259 // won't be safe to copy http.DefaultTransport itself later, because
 260 // its private mutexes might have already been used. (Without this,
 261 // the test suite sometimes panics "concurrent map writes" in
 262 // net/http.(*Transport).removeIdleConnLocked().)
 263 var defaultTransport = *(http.DefaultTransport.(*http.Transport))
 264
 265 type proxyHandler struct {
 266         http.Handler
 267         *keepclient.KeepClient
 268         *ApiTokenCache
 269         timeout   time.Duration
 270         transport *http.Transport
 271 }
 272
 273 // MakeRESTRouter returns an http.Handler that passes GET and PUT
 274 // requests to the appropriate handlers.
 275 func MakeRESTRouter(kc *keepclient.KeepClient, timeout time.Duration, mgmtToken string) http.Handler {
 276         rest := mux.NewRouter()
 277
 278         transport := defaultTransport
 279         transport.DialContext = (&net.Dialer{
 280                 Timeout:   keepclient.DefaultConnectTimeout,
 281                 KeepAlive: keepclient.DefaultKeepAlive,
 282                 DualStack: true,
 283         }).DialContext
 284         transport.TLSClientConfig = arvadosclient.MakeTLSConfig(kc.Arvados.ApiInsecure)
 285         transport.TLSHandshakeTimeout = keepclient.DefaultTLSHandshakeTimeout
 286
 287         h := &proxyHandler{
 288                 Handler:    rest,
 289                 KeepClient: kc,
 290                 timeout:    timeout,
 291                 transport:  &transport,
 292                 ApiTokenCache: &ApiTokenCache{
 293                         tokens:     make(map[string]int64),
 294                         expireTime: 300,
 295                 },
 296         }
 297
 298         rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Get).Methods("GET", "HEAD")
 299         rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Get).Methods("GET", "HEAD")
 300
 301         // List all blocks
 302         rest.HandleFunc(`/index`, h.Index).Methods("GET")
 303
 304         // List blocks whose hash has the given prefix
 305         rest.HandleFunc(`/index/{prefix:[0-9a-f]{0,32}}`, h.Index).Methods("GET")
 306
 307         rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Put).Methods("PUT")
 308         rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Put).Methods("PUT")
 309         rest.HandleFunc(`/`, h.Put).Methods("POST")
 310         rest.HandleFunc(`/{any}`, h.Options).Methods("OPTIONS")
 311         rest.HandleFunc(`/`, h.Options).Methods("OPTIONS")
 312
 313         rest.Handle("/_health/{check}", &health.Handler{
 314                 Token:  mgmtToken,
 315                 Prefix: "/_health/",
 316         }).Methods("GET")
 317
 318         rest.NotFoundHandler = InvalidPathHandler{}
 319         return h
 320 }
 321
 322 var errLoopDetected = errors.New("loop detected")
 323
 324 func (*proxyHandler) checkLoop(resp http.ResponseWriter, req *http.Request) error {
 325         if via := req.Header.Get("Via"); strings.Index(via, " "+viaAlias) >= 0 {
 326                 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)
 327                 http.Error(resp, errLoopDetected.Error(), http.StatusInternalServerError)
 328                 return errLoopDetected
 329         }
 330         return nil
 331 }
 332
 333 func SetCorsHeaders(resp http.ResponseWriter) {
 334         resp.Header().Set("Access-Control-Allow-Methods", "GET, HEAD, POST, PUT, OPTIONS")
 335         resp.Header().Set("Access-Control-Allow-Origin", "*")
 336         resp.Header().Set("Access-Control-Allow-Headers", "Authorization, Content-Length, Content-Type, X-Keep-Desired-Replicas")
 337         resp.Header().Set("Access-Control-Max-Age", "86486400")
 338 }
 339
 340 type InvalidPathHandler struct{}
 341
 342 func (InvalidPathHandler) ServeHTTP(resp http.ResponseWriter, req *http.Request) {
 343         log.Printf("%s: %s %s unroutable", GetRemoteAddress(req), req.Method, req.URL.Path)
 344         http.Error(resp, "Bad request", http.StatusBadRequest)
 345 }
 346
 347 func (h *proxyHandler) Options(resp http.ResponseWriter, req *http.Request) {
 348         log.Printf("%s: %s %s", GetRemoteAddress(req), req.Method, req.URL.Path)
 349         SetCorsHeaders(resp)
 350 }
 351
 352 var BadAuthorizationHeader = errors.New("Missing or invalid Authorization header")
 353 var ContentLengthMismatch = errors.New("Actual length != expected content length")
 354 var MethodNotSupported = errors.New("Method not supported")
 355
 356 var removeHint, _ = regexp.Compile("\\+K@[a-z0-9]{5}(\\+|$)")
 357
 358 func (h *proxyHandler) Get(resp http.ResponseWriter, req *http.Request) {
 359         if err := h.checkLoop(resp, req); err != nil {
 360                 return
 361         }
 362         SetCorsHeaders(resp)
 363         resp.Header().Set("Via", req.Proto+" "+viaAlias)
 364
 365         locator := mux.Vars(req)["locator"]
 366         var err error
 367         var status int
 368         var expectLength, responseLength int64
 369         var proxiedURI = "-"
 370
 371         defer func() {
 372                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, responseLength, proxiedURI, err)
 373                 if status != http.StatusOK {
 374                         http.Error(resp, err.Error(), status)
 375                 }
 376         }()
 377
 378         kc := h.makeKeepClient(req)
 379
 380         var pass bool
 381         var tok string
 382         if pass, tok = CheckAuthorizationHeader(kc, h.ApiTokenCache, req); !pass {
 383                 status, err = http.StatusForbidden, BadAuthorizationHeader
 384                 return
 385         }
 386
 387         // Copy ArvadosClient struct and use the client's API token
 388         arvclient := *kc.Arvados
 389         arvclient.ApiToken = tok
 390         kc.Arvados = &arvclient
 391
 392         var reader io.ReadCloser
 393
 394         locator = removeHint.ReplaceAllString(locator, "$1")
 395
 396         switch req.Method {
 397         case "HEAD":
 398                 expectLength, proxiedURI, err = kc.Ask(locator)
 399         case "GET":
 400                 reader, expectLength, proxiedURI, err = kc.Get(locator)
 401                 if reader != nil {
 402                         defer reader.Close()
 403                 }
 404         default:
 405                 status, err = http.StatusNotImplemented, MethodNotSupported
 406                 return
 407         }
 408
 409         if expectLength == -1 {
 410                 log.Println("Warning:", GetRemoteAddress(req), req.Method, proxiedURI, "Content-Length not provided")
 411         }
 412
 413         switch respErr := err.(type) {
 414         case nil:
 415                 status = http.StatusOK
 416                 resp.Header().Set("Content-Length", fmt.Sprint(expectLength))
 417                 switch req.Method {
 418                 case "HEAD":
 419                         responseLength = 0
 420                 case "GET":
 421                         responseLength, err = io.Copy(resp, reader)
 422                         if err == nil && expectLength > -1 && responseLength != expectLength {
 423                                 err = ContentLengthMismatch
 424                         }
 425                 }
 426         case keepclient.Error:
 427                 if respErr == keepclient.BlockNotFound {
 428                         status = http.StatusNotFound
 429                 } else if respErr.Temporary() {
 430                         status = http.StatusBadGateway
 431                 } else {
 432                         status = 422
 433                 }
 434         default:
 435                 status = http.StatusInternalServerError
 436         }
 437 }
 438
 439 var LengthRequiredError = errors.New(http.StatusText(http.StatusLengthRequired))
 440 var LengthMismatchError = errors.New("Locator size hint does not match Content-Length header")
 441
 442 func (h *proxyHandler) Put(resp http.ResponseWriter, req *http.Request) {
 443         if err := h.checkLoop(resp, req); err != nil {
 444                 return
 445         }
 446         SetCorsHeaders(resp)
 447         resp.Header().Set("Via", "HTTP/1.1 "+viaAlias)
 448
 449         kc := h.makeKeepClient(req)
 450
 451         var err error
 452         var expectLength int64
 453         var status = http.StatusInternalServerError
 454         var wroteReplicas int
 455         var locatorOut string = "-"
 456
 457         defer func() {
 458                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, kc.Want_replicas, wroteReplicas, locatorOut, err)
 459                 if status != http.StatusOK {
 460                         http.Error(resp, err.Error(), status)
 461                 }
 462         }()
 463
 464         locatorIn := mux.Vars(req)["locator"]
 465
 466         // Check if the client specified storage classes
 467         if req.Header.Get("X-Keep-Storage-Classes") != "" {
 468                 var scl []string
 469                 for _, sc := range strings.Split(req.Header.Get("X-Keep-Storage-Classes"), ",") {
 470                         scl = append(scl, strings.Trim(sc, " "))
 471                 }
 472                 kc.StorageClasses = scl
 473         }
 474
 475         _, err = fmt.Sscanf(req.Header.Get("Content-Length"), "%d", &expectLength)
 476         if err != nil || expectLength < 0 {
 477                 err = LengthRequiredError
 478                 status = http.StatusLengthRequired
 479                 return
 480         }
 481
 482         if locatorIn != "" {
 483                 var loc *keepclient.Locator
 484                 if loc, err = keepclient.MakeLocator(locatorIn); err != nil {
 485                         status = http.StatusBadRequest
 486                         return
 487                 } else if loc.Size > 0 && int64(loc.Size) != expectLength {
 488                         err = LengthMismatchError
 489                         status = http.StatusBadRequest
 490                         return
 491                 }
 492         }
 493
 494         var pass bool
 495         var tok string
 496         if pass, tok = CheckAuthorizationHeader(kc, h.ApiTokenCache, req); !pass {
 497                 err = BadAuthorizationHeader
 498                 status = http.StatusForbidden
 499                 return
 500         }
 501
 502         // Copy ArvadosClient struct and use the client's API token
 503         arvclient := *kc.Arvados
 504         arvclient.ApiToken = tok
 505         kc.Arvados = &arvclient
 506
 507         // Check if the client specified the number of replicas
 508         if req.Header.Get("X-Keep-Desired-Replicas") != "" {
 509                 var r int
 510                 _, err := fmt.Sscanf(req.Header.Get(keepclient.X_Keep_Desired_Replicas), "%d", &r)
 511                 if err == nil {
 512                         kc.Want_replicas = r
 513                 }
 514         }
 515
 516         // Now try to put the block through
 517         if locatorIn == "" {
 518                 bytes, err2 := ioutil.ReadAll(req.Body)
 519                 if err2 != nil {
 520                         err = fmt.Errorf("Error reading request body: %s", err2)
 521                         status = http.StatusInternalServerError
 522                         return
 523                 }
 524                 locatorOut, wroteReplicas, err = kc.PutB(bytes)
 525         } else {
 526                 locatorOut, wroteReplicas, err = kc.PutHR(locatorIn, req.Body, expectLength)
 527         }
 528
 529         // Tell the client how many successful PUTs we accomplished
 530         resp.Header().Set(keepclient.X_Keep_Replicas_Stored, fmt.Sprintf("%d", wroteReplicas))
 531
 532         switch err.(type) {
 533         case nil:
 534                 status = http.StatusOK
 535                 _, err = io.WriteString(resp, locatorOut)
 536
 537         case keepclient.OversizeBlockError:
 538                 // Too much data
 539                 status = http.StatusRequestEntityTooLarge
 540
 541         case keepclient.InsufficientReplicasError:
 542                 if wroteReplicas > 0 {
 543                         // At least one write is considered success.  The
 544                         // client can decide if getting less than the number of
 545                         // replications it asked for is a fatal error.
 546                         status = http.StatusOK
 547                         _, err = io.WriteString(resp, locatorOut)
 548                 } else {
 549                         status = http.StatusServiceUnavailable
 550                 }
 551
 552         default:
 553                 status = http.StatusBadGateway
 554         }
 555 }
 556
 557 // ServeHTTP implementation for IndexHandler
 558 // Supports only GET requests for /index/{prefix:[0-9a-f]{0,32}}
 559 // For each keep server found in LocalRoots:
 560 //   Invokes GetIndex using keepclient
 561 //   Expects "complete" response (terminating with blank new line)
 562 //   Aborts on any errors
 563 // Concatenates responses from all those keep servers and returns
 564 func (h *proxyHandler) Index(resp http.ResponseWriter, req *http.Request) {
 565         SetCorsHeaders(resp)
 566
 567         prefix := mux.Vars(req)["prefix"]
 568         var err error
 569         var status int
 570
 571         defer func() {
 572                 if status != http.StatusOK {
 573                         http.Error(resp, err.Error(), status)
 574                 }
 575         }()
 576
 577         kc := h.makeKeepClient(req)
 578         ok, token := CheckAuthorizationHeader(kc, h.ApiTokenCache, req)
 579         if !ok {
 580                 status, err = http.StatusForbidden, BadAuthorizationHeader
 581                 return
 582         }
 583
 584         // Copy ArvadosClient struct and use the client's API token
 585         arvclient := *kc.Arvados
 586         arvclient.ApiToken = token
 587         kc.Arvados = &arvclient
 588
 589         // Only GET method is supported
 590         if req.Method != "GET" {
 591                 status, err = http.StatusNotImplemented, MethodNotSupported
 592                 return
 593         }
 594
 595         // Get index from all LocalRoots and write to resp
 596         var reader io.Reader
 597         for uuid := range kc.LocalRoots() {
 598                 reader, err = kc.GetIndex(uuid, prefix)
 599                 if err != nil {
 600                         status = http.StatusBadGateway
 601                         return
 602                 }
 603
 604                 _, err = io.Copy(resp, reader)
 605                 if err != nil {
 606                         status = http.StatusBadGateway
 607                         return
 608                 }
 609         }
 610
 611         // Got index from all the keep servers and wrote to resp
 612         status = http.StatusOK
 613         resp.Write([]byte("\n"))
 614 }
 615
 616 func (h *proxyHandler) makeKeepClient(req *http.Request) *keepclient.KeepClient {
 617         kc := *h.KeepClient
 618         kc.RequestID = req.Header.Get("X-Request-Id")
 619         kc.HTTPClient = &proxyClient{
 620                 client: &http.Client{
 621                         Timeout:   h.timeout,
 622                         Transport: h.transport,
 623                 },
 624                 proto: req.Proto,
 625         }
 626         return &kc
 627 }