lib/controller/federation.go

   1 // Copyright (C) The Arvados Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package controller
   6
   7 import (
   8         "bytes"
   9         "database/sql"
  10         "encoding/json"
  11         "fmt"
  12         "io"
  13         "io/ioutil"
  14         "mime"
  15         "net/http"
  16         "net/url"
  17         "regexp"
  18         "strings"
  19
  20         "git.arvados.org/arvados.git/sdk/go/arvados"
  21         "git.arvados.org/arvados.git/sdk/go/auth"
  22         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  23         "github.com/jmcvetta/randutil"
  24 )
  25
  26 var pathPattern = `^/arvados/v1/%s(/([0-9a-z]{5})-%s-[0-9a-z]{15})?(.*)$`
  27 var wfRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "workflows", "7fd4e"))
  28 var containersRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "containers", "dz642"))
  29 var containerRequestsRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "container_requests", "xvhdp"))
  30 var collectionsRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "collections", "4zz18"))
  31 var collectionsByPDHRe = regexp.MustCompile(`^/arvados/v1/collections/([0-9a-fA-F]{32}\+[0-9]+)+$`)
  32 var linksRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "links", "o0j2j"))
  33
  34 func (h *Handler) remoteClusterRequest(remoteID string, req *http.Request) (*http.Response, error) {
  35         remote, ok := h.Cluster.RemoteClusters[remoteID]
  36         if !ok {
  37                 return nil, HTTPError{fmt.Sprintf("no proxy available for cluster %v", remoteID), http.StatusNotFound}
  38         }
  39         scheme := remote.Scheme
  40         if scheme == "" {
  41                 scheme = "https"
  42         }
  43         saltedReq, err := h.saltAuthToken(req, remoteID)
  44         if err != nil {
  45                 return nil, err
  46         }
  47         urlOut := &url.URL{
  48                 Scheme:   scheme,
  49                 Host:     remote.Host,
  50                 Path:     saltedReq.URL.Path,
  51                 RawPath:  saltedReq.URL.RawPath,
  52                 RawQuery: saltedReq.URL.RawQuery,
  53         }
  54         client := h.secureClient
  55         if remote.Insecure {
  56                 client = h.insecureClient
  57         }
  58         return h.proxy.Do(saltedReq, urlOut, client)
  59 }
  60
  61 // Buffer request body, parse form parameters in request, and then
  62 // replace original body with the buffer so it can be re-read by
  63 // downstream proxy steps.
  64 func loadParamsFromForm(req *http.Request) error {
  65         var postBody *bytes.Buffer
  66         if ct := req.Header.Get("Content-Type"); ct == "" {
  67                 // Assume application/octet-stream, i.e., no form to parse.
  68         } else if ct, _, err := mime.ParseMediaType(ct); err != nil {
  69                 return err
  70         } else if ct == "application/x-www-form-urlencoded" && req.Body != nil {
  71                 var cl int64
  72                 if req.ContentLength > 0 {
  73                         cl = req.ContentLength
  74                 }
  75                 postBody = bytes.NewBuffer(make([]byte, 0, cl))
  76                 originalBody := req.Body
  77                 defer originalBody.Close()
  78                 req.Body = ioutil.NopCloser(io.TeeReader(req.Body, postBody))
  79         }
  80
  81         err := req.ParseForm()
  82         if err != nil {
  83                 return err
  84         }
  85
  86         if req.Body != nil && postBody != nil {
  87                 req.Body = ioutil.NopCloser(postBody)
  88         }
  89         return nil
  90 }
  91
  92 func (h *Handler) setupProxyRemoteCluster(next http.Handler) http.Handler {
  93         mux := http.NewServeMux()
  94
  95         wfHandler := &genericFederatedRequestHandler{next, h, wfRe, nil}
  96         containersHandler := &genericFederatedRequestHandler{next, h, containersRe, nil}
  97         containerRequestsHandler := &genericFederatedRequestHandler{next, h, containerRequestsRe,
  98                 []federatedRequestDelegate{remoteContainerRequestCreate}}
  99         collectionsRequestsHandler := &genericFederatedRequestHandler{next, h, collectionsRe,
 100                 []federatedRequestDelegate{fetchRemoteCollectionByUUID, fetchRemoteCollectionByPDH}}
 101         linksRequestsHandler := &genericFederatedRequestHandler{next, h, linksRe, nil}
 102
 103         mux.Handle("/arvados/v1/workflows", wfHandler)
 104         mux.Handle("/arvados/v1/workflows/", wfHandler)
 105         mux.Handle("/arvados/v1/containers", containersHandler)
 106         mux.Handle("/arvados/v1/containers/", containersHandler)
 107         mux.Handle("/arvados/v1/container_requests", containerRequestsHandler)
 108         mux.Handle("/arvados/v1/container_requests/", containerRequestsHandler)
 109         mux.Handle("/arvados/v1/collections", collectionsRequestsHandler)
 110         mux.Handle("/arvados/v1/collections/", collectionsRequestsHandler)
 111         mux.Handle("/arvados/v1/links", linksRequestsHandler)
 112         mux.Handle("/arvados/v1/links/", linksRequestsHandler)
 113         mux.Handle("/", next)
 114
 115         return http.HandlerFunc(func(w http.ResponseWriter, req *http.Request) {
 116                 parts := strings.Split(req.Header.Get("Authorization"), "/")
 117                 alreadySalted := (len(parts) == 3 && parts[0] == "Bearer v2" && len(parts[2]) == 40)
 118
 119                 if alreadySalted ||
 120                         strings.Index(req.Header.Get("Via"), "arvados-controller") != -1 {
 121                         // The token is already salted, or this is a
 122                         // request from another instance of
 123                         // arvados-controller.  In either case, we
 124                         // don't want to proxy this query, so just
 125                         // continue down the instance handler stack.
 126                         next.ServeHTTP(w, req)
 127                         return
 128                 }
 129
 130                 mux.ServeHTTP(w, req)
 131         })
 132
 133         return mux
 134 }
 135
 136 type CurrentUser struct {
 137         Authorization arvados.APIClientAuthorization
 138         UUID          string
 139 }
 140
 141 // validateAPItoken extracts the token from the provided http request,
 142 // checks it again api_client_authorizations table in the database,
 143 // and fills in the token scope and user UUID.  Does not handle remote
 144 // tokens unless they are already in the database and not expired.
 145 //
 146 // Return values are:
 147 //
 148 // nil, false, non-nil -- if there was an internal error
 149 //
 150 // nil, false, nil -- if the token is invalid
 151 //
 152 // non-nil, true, nil -- if the token is valid
 153 func (h *Handler) validateAPItoken(req *http.Request, token string) (*CurrentUser, bool, error) {
 154         user := CurrentUser{Authorization: arvados.APIClientAuthorization{APIToken: token}}
 155         db, err := h.db(req.Context())
 156         if err != nil {
 157                 ctxlog.FromContext(req.Context()).WithError(err).Debugf("validateAPItoken(%s): database error", token)
 158                 return nil, false, err
 159         }
 160
 161         var uuid string
 162         if strings.HasPrefix(token, "v2/") {
 163                 sp := strings.Split(token, "/")
 164                 uuid = sp[1]
 165                 token = sp[2]
 166         }
 167         user.Authorization.APIToken = token
 168         var scopes string
 169         err = db.QueryRowContext(req.Context(), `SELECT api_client_authorizations.uuid, api_client_authorizations.scopes, users.uuid FROM api_client_authorizations JOIN users on api_client_authorizations.user_id=users.id WHERE api_token=$1 AND (expires_at IS NULL OR expires_at > current_timestamp AT TIME ZONE 'UTC') LIMIT 1`, token).Scan(&user.Authorization.UUID, &scopes, &user.UUID)
 170         if err == sql.ErrNoRows {
 171                 ctxlog.FromContext(req.Context()).Debugf("validateAPItoken(%s): not found in database", token)
 172                 return nil, false, nil
 173         } else if err != nil {
 174                 ctxlog.FromContext(req.Context()).WithError(err).Debugf("validateAPItoken(%s): database error", token)
 175                 return nil, false, err
 176         }
 177         if uuid != "" && user.Authorization.UUID != uuid {
 178                 // secret part matches, but UUID doesn't -- somewhat surprising
 179                 ctxlog.FromContext(req.Context()).Debugf("validateAPItoken(%s): secret part found, but with different UUID: %s", token, user.Authorization.UUID)
 180                 return nil, false, nil
 181         }
 182         err = json.Unmarshal([]byte(scopes), &user.Authorization.Scopes)
 183         if err != nil {
 184                 ctxlog.FromContext(req.Context()).WithError(err).Debugf("validateAPItoken(%s): error parsing scopes from db", token)
 185                 return nil, false, err
 186         }
 187         ctxlog.FromContext(req.Context()).Debugf("validateAPItoken(%s): ok", token)
 188         return &user, true, nil
 189 }
 190
 191 func (h *Handler) createAPItoken(req *http.Request, userUUID string, scopes []string) (*arvados.APIClientAuthorization, error) {
 192         db, err := h.db(req.Context())
 193         if err != nil {
 194                 return nil, err
 195         }
 196         rd, err := randutil.String(15, "abcdefghijklmnopqrstuvwxyz0123456789")
 197         if err != nil {
 198                 return nil, err
 199         }
 200         uuid := fmt.Sprintf("%v-gj3su-%v", h.Cluster.ClusterID, rd)
 201         token, err := randutil.String(50, "abcdefghijklmnopqrstuvwxyz0123456789")
 202         if err != nil {
 203                 return nil, err
 204         }
 205         if len(scopes) == 0 {
 206                 scopes = append(scopes, "all")
 207         }
 208         scopesjson, err := json.Marshal(scopes)
 209         if err != nil {
 210                 return nil, err
 211         }
 212         _, err = db.ExecContext(req.Context(),
 213                 `INSERT INTO api_client_authorizations
 214 (uuid, api_token, expires_at, scopes,
 215 user_id,
 216 api_client_id, created_at, updated_at)
 217 VALUES ($1, $2, CURRENT_TIMESTAMP AT TIME ZONE 'UTC' + INTERVAL '2 weeks', $3,
 218 (SELECT id FROM users WHERE users.uuid=$4 LIMIT 1),
 219 0, CURRENT_TIMESTAMP AT TIME ZONE 'UTC', CURRENT_TIMESTAMP AT TIME ZONE 'UTC')`,
 220                 uuid, token, string(scopesjson), userUUID)
 221
 222         if err != nil {
 223                 return nil, err
 224         }
 225
 226         return &arvados.APIClientAuthorization{
 227                 UUID:      uuid,
 228                 APIToken:  token,
 229                 ExpiresAt: "",
 230                 Scopes:    scopes}, nil
 231 }
 232
 233 // Extract the auth token supplied in req, and replace it with a
 234 // salted token for the remote cluster.
 235 func (h *Handler) saltAuthToken(req *http.Request, remote string) (updatedReq *http.Request, err error) {
 236         updatedReq = (&http.Request{
 237                 Method:        req.Method,
 238                 URL:           req.URL,
 239                 Header:        req.Header,
 240                 Body:          req.Body,
 241                 ContentLength: req.ContentLength,
 242                 Host:          req.Host,
 243         }).WithContext(req.Context())
 244
 245         creds := auth.NewCredentials()
 246         creds.LoadTokensFromHTTPRequest(updatedReq)
 247         if len(creds.Tokens) == 0 && updatedReq.Header.Get("Content-Type") == "application/x-www-form-encoded" {
 248                 // Override ParseForm's 10MiB limit by ensuring
 249                 // req.Body is a *http.maxBytesReader.
 250                 updatedReq.Body = http.MaxBytesReader(nil, updatedReq.Body, 1<<28) // 256MiB. TODO: use MaxRequestSize from discovery doc or config.
 251                 if err := creds.LoadTokensFromHTTPRequestBody(updatedReq); err != nil {
 252                         return nil, err
 253                 }
 254                 // Replace req.Body with a buffer that re-encodes the
 255                 // form without api_token, in case we end up
 256                 // forwarding the request.
 257                 if updatedReq.PostForm != nil {
 258                         updatedReq.PostForm.Del("api_token")
 259                 }
 260                 updatedReq.Body = ioutil.NopCloser(bytes.NewBufferString(updatedReq.PostForm.Encode()))
 261         }
 262         if len(creds.Tokens) == 0 {
 263                 return updatedReq, nil
 264         }
 265
 266         ctxlog.FromContext(req.Context()).Infof("saltAuthToken: cluster %s token %s remote %s", h.Cluster.ClusterID, creds.Tokens[0], remote)
 267         token, err := auth.SaltToken(creds.Tokens[0], remote)
 268
 269         if err == auth.ErrObsoleteToken {
 270                 // If the token exists in our own database for our own
 271                 // user, salt it for the remote. Otherwise, assume it
 272                 // was issued by the remote, and pass it through
 273                 // unmodified.
 274                 currentUser, ok, err := h.validateAPItoken(req, creds.Tokens[0])
 275                 if err != nil {
 276                         return nil, err
 277                 } else if !ok || strings.HasPrefix(currentUser.UUID, remote) {
 278                         // Unknown, or cached + belongs to remote;
 279                         // pass through unmodified.
 280                         token = creds.Tokens[0]
 281                 } else {
 282                         // Found; make V2 version and salt it.
 283                         token, err = auth.SaltToken(currentUser.Authorization.TokenV2(), remote)
 284                         if err != nil {
 285                                 return nil, err
 286                         }
 287                 }
 288         } else if err != nil {
 289                 return nil, err
 290         }
 291         updatedReq.Header = http.Header{}
 292         for k, v := range req.Header {
 293                 if k != "Authorization" {
 294                         updatedReq.Header[k] = v
 295                 }
 296         }
 297         updatedReq.Header.Set("Authorization", "Bearer "+token)
 298
 299         // Remove api_token=... from the query string, in case we
 300         // end up forwarding the request.
 301         if values, err := url.ParseQuery(updatedReq.URL.RawQuery); err != nil {
 302                 return nil, err
 303         } else if _, ok := values["api_token"]; ok {
 304                 delete(values, "api_token")
 305                 updatedReq.URL = &url.URL{
 306                         Scheme:   req.URL.Scheme,
 307                         Host:     req.URL.Host,
 308                         Path:     req.URL.Path,
 309                         RawPath:  req.URL.RawPath,
 310                         RawQuery: values.Encode(),
 311                 }
 312         }
 313         return updatedReq, nil
 314 }