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 // RememberToken caches the token and set an expire time.  If we already have
 177 // an expire time on the token, it is not updated.
 178 func (cache *APITokenCache) RememberToken(token string) {
 179         cache.lock.Lock()
 180         defer cache.lock.Unlock()
 181
 182         now := time.Now().Unix()
 183         if cache.tokens[token] == 0 {
 184                 cache.tokens[token] = now + cache.expireTime
 185         }
 186 }
 187
 188 // RecallToken checks if the cached token is known and still believed to be
 189 // valid.
 190 func (cache *APITokenCache) RecallToken(token string) bool {
 191         cache.lock.Lock()
 192         defer cache.lock.Unlock()
 193
 194         now := time.Now().Unix()
 195         if cache.tokens[token] == 0 {
 196                 // Unknown token
 197                 return false
 198         } else if now < cache.tokens[token] {
 199                 // Token is known and still valid
 200                 return true
 201         } else {
 202                 // Token is expired
 203                 cache.tokens[token] = 0
 204                 return false
 205         }
 206 }
 207
 208 // GetRemoteAddress returns a string with the remote address for the request.
 209 // If the X-Forwarded-For header is set and has a non-zero length, it returns a
 210 // string made from a comma separated list of all the remote addresses,
 211 // starting with the one(s) from the X-Forwarded-For header.
 212 func GetRemoteAddress(req *http.Request) string {
 213         if xff := req.Header.Get("X-Forwarded-For"); xff != "" {
 214                 return xff + "," + req.RemoteAddr
 215         }
 216         return req.RemoteAddr
 217 }
 218
 219 func CheckAuthorizationHeader(kc *keepclient.KeepClient, cache *APITokenCache, req *http.Request) (pass bool, tok string) {
 220         parts := strings.SplitN(req.Header.Get("Authorization"), " ", 2)
 221         if len(parts) < 2 || !(parts[0] == "OAuth2" || parts[0] == "Bearer") || len(parts[1]) == 0 {
 222                 return false, ""
 223         }
 224         tok = parts[1]
 225
 226         // Tokens are validated differently depending on what kind of
 227         // operation is being performed. For example, tokens in
 228         // collection-sharing links permit GET requests, but not
 229         // PUT requests.
 230         var op string
 231         if req.Method == "GET" || req.Method == "HEAD" {
 232                 op = "read"
 233         } else {
 234                 op = "write"
 235         }
 236
 237         if cache.RecallToken(op + ":" + tok) {
 238                 // Valid in the cache, short circuit
 239                 return true, tok
 240         }
 241
 242         var err error
 243         arv := *kc.Arvados
 244         arv.ApiToken = tok
 245         arv.RequestID = req.Header.Get("X-Request-Id")
 246         if op == "read" {
 247                 err = arv.Call("HEAD", "keep_services", "", "accessible", nil, nil)
 248         } else {
 249                 err = arv.Call("HEAD", "users", "", "current", nil, nil)
 250         }
 251         if err != nil {
 252                 log.Printf("%s: CheckAuthorizationHeader error: %v", GetRemoteAddress(req), err)
 253                 return false, ""
 254         }
 255
 256         // Success!  Update cache
 257         cache.RememberToken(op + ":" + tok)
 258
 259         return true, tok
 260 }
 261
 262 // We need to make a private copy of the default http transport early
 263 // in initialization, then make copies of our private copy later. It
 264 // won't be safe to copy http.DefaultTransport itself later, because
 265 // its private mutexes might have already been used. (Without this,
 266 // the test suite sometimes panics "concurrent map writes" in
 267 // net/http.(*Transport).removeIdleConnLocked().)
 268 var defaultTransport = *(http.DefaultTransport.(*http.Transport))
 269
 270 type proxyHandler struct {
 271         http.Handler
 272         *keepclient.KeepClient
 273         *APITokenCache
 274         timeout   time.Duration
 275         transport *http.Transport
 276 }
 277
 278 // MakeRESTRouter returns an http.Handler that passes GET and PUT
 279 // requests to the appropriate handlers.
 280 func MakeRESTRouter(kc *keepclient.KeepClient, timeout time.Duration, mgmtToken string) http.Handler {
 281         rest := mux.NewRouter()
 282
 283         transport := defaultTransport
 284         transport.DialContext = (&net.Dialer{
 285                 Timeout:   keepclient.DefaultConnectTimeout,
 286                 KeepAlive: keepclient.DefaultKeepAlive,
 287                 DualStack: true,
 288         }).DialContext
 289         transport.TLSClientConfig = arvadosclient.MakeTLSConfig(kc.Arvados.ApiInsecure)
 290         transport.TLSHandshakeTimeout = keepclient.DefaultTLSHandshakeTimeout
 291
 292         h := &proxyHandler{
 293                 Handler:    rest,
 294                 KeepClient: kc,
 295                 timeout:    timeout,
 296                 transport:  &transport,
 297                 APITokenCache: &APITokenCache{
 298                         tokens:     make(map[string]int64),
 299                         expireTime: 300,
 300                 },
 301         }
 302
 303         rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Get).Methods("GET", "HEAD")
 304         rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Get).Methods("GET", "HEAD")
 305
 306         // List all blocks
 307         rest.HandleFunc(`/index`, h.Index).Methods("GET")
 308
 309         // List blocks whose hash has the given prefix
 310         rest.HandleFunc(`/index/{prefix:[0-9a-f]{0,32}}`, h.Index).Methods("GET")
 311
 312         rest.HandleFunc(`/{locator:[0-9a-f]{32}\+.*}`, h.Put).Methods("PUT")
 313         rest.HandleFunc(`/{locator:[0-9a-f]{32}}`, h.Put).Methods("PUT")
 314         rest.HandleFunc(`/`, h.Put).Methods("POST")
 315         rest.HandleFunc(`/{any}`, h.Options).Methods("OPTIONS")
 316         rest.HandleFunc(`/`, h.Options).Methods("OPTIONS")
 317
 318         rest.Handle("/_health/{check}", &health.Handler{
 319                 Token:  mgmtToken,
 320                 Prefix: "/_health/",
 321         }).Methods("GET")
 322
 323         rest.NotFoundHandler = InvalidPathHandler{}
 324         return h
 325 }
 326
 327 var errLoopDetected = errors.New("loop detected")
 328
 329 func (*proxyHandler) checkLoop(resp http.ResponseWriter, req *http.Request) error {
 330         if via := req.Header.Get("Via"); strings.Index(via, " "+viaAlias) >= 0 {
 331                 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)
 332                 http.Error(resp, errLoopDetected.Error(), http.StatusInternalServerError)
 333                 return errLoopDetected
 334         }
 335         return nil
 336 }
 337
 338 func SetCorsHeaders(resp http.ResponseWriter) {
 339         resp.Header().Set("Access-Control-Allow-Methods", "GET, HEAD, POST, PUT, OPTIONS")
 340         resp.Header().Set("Access-Control-Allow-Origin", "*")
 341         resp.Header().Set("Access-Control-Allow-Headers", "Authorization, Content-Length, Content-Type, X-Keep-Desired-Replicas")
 342         resp.Header().Set("Access-Control-Max-Age", "86486400")
 343 }
 344
 345 type InvalidPathHandler struct{}
 346
 347 func (InvalidPathHandler) ServeHTTP(resp http.ResponseWriter, req *http.Request) {
 348         log.Printf("%s: %s %s unroutable", GetRemoteAddress(req), req.Method, req.URL.Path)
 349         http.Error(resp, "Bad request", http.StatusBadRequest)
 350 }
 351
 352 func (h *proxyHandler) Options(resp http.ResponseWriter, req *http.Request) {
 353         log.Printf("%s: %s %s", GetRemoteAddress(req), req.Method, req.URL.Path)
 354         SetCorsHeaders(resp)
 355 }
 356
 357 var errBadAuthorizationHeader = errors.New("Missing or invalid Authorization header")
 358 var errContentLengthMismatch = errors.New("Actual length != expected content length")
 359 var errMethodNotSupported = errors.New("Method not supported")
 360
 361 var removeHint, _ = regexp.Compile("\\+K@[a-z0-9]{5}(\\+|$)")
 362
 363 func (h *proxyHandler) Get(resp http.ResponseWriter, req *http.Request) {
 364         if err := h.checkLoop(resp, req); err != nil {
 365                 return
 366         }
 367         SetCorsHeaders(resp)
 368         resp.Header().Set("Via", req.Proto+" "+viaAlias)
 369
 370         locator := mux.Vars(req)["locator"]
 371         var err error
 372         var status int
 373         var expectLength, responseLength int64
 374         var proxiedURI = "-"
 375
 376         defer func() {
 377                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, responseLength, proxiedURI, err)
 378                 if status != http.StatusOK {
 379                         http.Error(resp, err.Error(), status)
 380                 }
 381         }()
 382
 383         kc := h.makeKeepClient(req)
 384
 385         var pass bool
 386         var tok string
 387         if pass, tok = CheckAuthorizationHeader(kc, h.APITokenCache, req); !pass {
 388                 status, err = http.StatusForbidden, errBadAuthorizationHeader
 389                 return
 390         }
 391
 392         // Copy ArvadosClient struct and use the client's API token
 393         arvclient := *kc.Arvados
 394         arvclient.ApiToken = tok
 395         kc.Arvados = &arvclient
 396
 397         var reader io.ReadCloser
 398
 399         locator = removeHint.ReplaceAllString(locator, "$1")
 400
 401         switch req.Method {
 402         case "HEAD":
 403                 expectLength, proxiedURI, err = kc.Ask(locator)
 404         case "GET":
 405                 reader, expectLength, proxiedURI, err = kc.Get(locator)
 406                 if reader != nil {
 407                         defer reader.Close()
 408                 }
 409         default:
 410                 status, err = http.StatusNotImplemented, errMethodNotSupported
 411                 return
 412         }
 413
 414         if expectLength == -1 {
 415                 log.Println("Warning:", GetRemoteAddress(req), req.Method, proxiedURI, "Content-Length not provided")
 416         }
 417
 418         switch respErr := err.(type) {
 419         case nil:
 420                 status = http.StatusOK
 421                 resp.Header().Set("Content-Length", fmt.Sprint(expectLength))
 422                 switch req.Method {
 423                 case "HEAD":
 424                         responseLength = 0
 425                 case "GET":
 426                         responseLength, err = io.Copy(resp, reader)
 427                         if err == nil && expectLength > -1 && responseLength != expectLength {
 428                                 err = errContentLengthMismatch
 429                         }
 430                 }
 431         case keepclient.Error:
 432                 if respErr == keepclient.BlockNotFound {
 433                         status = http.StatusNotFound
 434                 } else if respErr.Temporary() {
 435                         status = http.StatusBadGateway
 436                 } else {
 437                         status = 422
 438                 }
 439         default:
 440                 status = http.StatusInternalServerError
 441         }
 442 }
 443
 444 var errLengthRequired = errors.New(http.StatusText(http.StatusLengthRequired))
 445 var errLengthMismatch = errors.New("Locator size hint does not match Content-Length header")
 446
 447 func (h *proxyHandler) Put(resp http.ResponseWriter, req *http.Request) {
 448         if err := h.checkLoop(resp, req); err != nil {
 449                 return
 450         }
 451         SetCorsHeaders(resp)
 452         resp.Header().Set("Via", "HTTP/1.1 "+viaAlias)
 453
 454         kc := h.makeKeepClient(req)
 455
 456         var err error
 457         var expectLength int64
 458         var status = http.StatusInternalServerError
 459         var wroteReplicas int
 460         var locatorOut string = "-"
 461
 462         defer func() {
 463                 log.Println(GetRemoteAddress(req), req.Method, req.URL.Path, status, expectLength, kc.Want_replicas, wroteReplicas, locatorOut, err)
 464                 if status != http.StatusOK {
 465                         http.Error(resp, err.Error(), status)
 466                 }
 467         }()
 468
 469         locatorIn := mux.Vars(req)["locator"]
 470
 471         // Check if the client specified storage classes
 472         if req.Header.Get("X-Keep-Storage-Classes") != "" {
 473                 var scl []string
 474                 for _, sc := range strings.Split(req.Header.Get("X-Keep-Storage-Classes"), ",") {
 475                         scl = append(scl, strings.Trim(sc, " "))
 476                 }
 477                 kc.StorageClasses = scl
 478         }
 479
 480         _, err = fmt.Sscanf(req.Header.Get("Content-Length"), "%d", &expectLength)
 481         if err != nil || expectLength < 0 {
 482                 err = errLengthRequired
 483                 status = http.StatusLengthRequired
 484                 return
 485         }
 486
 487         if locatorIn != "" {
 488                 var loc *keepclient.Locator
 489                 if loc, err = keepclient.MakeLocator(locatorIn); err != nil {
 490                         status = http.StatusBadRequest
 491                         return
 492                 } else if loc.Size > 0 && int64(loc.Size) != expectLength {
 493                         err = errLengthMismatch
 494                         status = http.StatusBadRequest
 495                         return
 496                 }
 497         }
 498
 499         var pass bool
 500         var tok string
 501         if pass, tok = CheckAuthorizationHeader(kc, h.APITokenCache, req); !pass {
 502                 err = errBadAuthorizationHeader
 503                 status = http.StatusForbidden
 504                 return
 505         }
 506
 507         // Copy ArvadosClient struct and use the client's API token
 508         arvclient := *kc.Arvados
 509         arvclient.ApiToken = tok
 510         kc.Arvados = &arvclient
 511
 512         // Check if the client specified the number of replicas
 513         if req.Header.Get("X-Keep-Desired-Replicas") != "" {
 514                 var r int
 515                 _, err := fmt.Sscanf(req.Header.Get(keepclient.XKeepDesiredReplicas), "%d", &r)
 516                 if err == nil {
 517                         kc.Want_replicas = r
 518                 }
 519         }
 520
 521         // Now try to put the block through
 522         if locatorIn == "" {
 523                 bytes, err2 := ioutil.ReadAll(req.Body)
 524                 if err2 != nil {
 525                         err = fmt.Errorf("Error reading request body: %s", err2)
 526                         status = http.StatusInternalServerError
 527                         return
 528                 }
 529                 locatorOut, wroteReplicas, err = kc.PutB(bytes)
 530         } else {
 531                 locatorOut, wroteReplicas, err = kc.PutHR(locatorIn, req.Body, expectLength)
 532         }
 533
 534         // Tell the client how many successful PUTs we accomplished
 535         resp.Header().Set(keepclient.XKeepReplicasStored, fmt.Sprintf("%d", wroteReplicas))
 536
 537         switch err.(type) {
 538         case nil:
 539                 status = http.StatusOK
 540                 _, err = io.WriteString(resp, locatorOut)
 541
 542         case keepclient.OversizeBlockError:
 543                 // Too much data
 544                 status = http.StatusRequestEntityTooLarge
 545
 546         case keepclient.InsufficientReplicasError:
 547                 if wroteReplicas > 0 {
 548                         // At least one write is considered success.  The
 549                         // client can decide if getting less than the number of
 550                         // replications it asked for is a fatal error.
 551                         status = http.StatusOK
 552                         _, err = io.WriteString(resp, locatorOut)
 553                 } else {
 554                         status = http.StatusServiceUnavailable
 555                 }
 556
 557         default:
 558                 status = http.StatusBadGateway
 559         }
 560 }
 561
 562 // ServeHTTP implementation for IndexHandler
 563 // Supports only GET requests for /index/{prefix:[0-9a-f]{0,32}}
 564 // For each keep server found in LocalRoots:
 565 //   Invokes GetIndex using keepclient
 566 //   Expects "complete" response (terminating with blank new line)
 567 //   Aborts on any errors
 568 // Concatenates responses from all those keep servers and returns
 569 func (h *proxyHandler) Index(resp http.ResponseWriter, req *http.Request) {
 570         SetCorsHeaders(resp)
 571
 572         prefix := mux.Vars(req)["prefix"]
 573         var err error
 574         var status int
 575
 576         defer func() {
 577                 if status != http.StatusOK {
 578                         http.Error(resp, err.Error(), status)
 579                 }
 580         }()
 581
 582         kc := h.makeKeepClient(req)
 583         ok, token := CheckAuthorizationHeader(kc, h.APITokenCache, req)
 584         if !ok {
 585                 status, err = http.StatusForbidden, errBadAuthorizationHeader
 586                 return
 587         }
 588
 589         // Copy ArvadosClient struct and use the client's API token
 590         arvclient := *kc.Arvados
 591         arvclient.ApiToken = token
 592         kc.Arvados = &arvclient
 593
 594         // Only GET method is supported
 595         if req.Method != "GET" {
 596                 status, err = http.StatusNotImplemented, errMethodNotSupported
 597                 return
 598         }
 599
 600         // Get index from all LocalRoots and write to resp
 601         var reader io.Reader
 602         for uuid := range kc.LocalRoots() {
 603                 reader, err = kc.GetIndex(uuid, prefix)
 604                 if err != nil {
 605                         status = http.StatusBadGateway
 606                         return
 607                 }
 608
 609                 _, err = io.Copy(resp, reader)
 610                 if err != nil {
 611                         status = http.StatusBadGateway
 612                         return
 613                 }
 614         }
 615
 616         // Got index from all the keep servers and wrote to resp
 617         status = http.StatusOK
 618         resp.Write([]byte("\n"))
 619 }
 620
 621 func (h *proxyHandler) makeKeepClient(req *http.Request) *keepclient.KeepClient {
 622         kc := *h.KeepClient
 623         kc.RequestID = req.Header.Get("X-Request-Id")
 624         kc.HTTPClient = &proxyClient{
 625                 client: &http.Client{
 626                         Timeout:   h.timeout,
 627                         Transport: h.transport,
 628                 },
 629                 proto: req.Proto,
 630         }
 631         return &kc
 632 }