services/keep-web/server_test.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         "bytes"
   9         "crypto/md5"
  10         "encoding/json"
  11         "fmt"
  12         "io"
  13         "io/ioutil"
  14         "net"
  15         "net/http"
  16         "os"
  17         "os/exec"
  18         "strings"
  19         "testing"
  20
  21         "git.arvados.org/arvados.git/lib/config"
  22         "git.arvados.org/arvados.git/sdk/go/arvados"
  23         "git.arvados.org/arvados.git/sdk/go/arvadosclient"
  24         "git.arvados.org/arvados.git/sdk/go/arvadostest"
  25         "git.arvados.org/arvados.git/sdk/go/ctxlog"
  26         "git.arvados.org/arvados.git/sdk/go/keepclient"
  27         check "gopkg.in/check.v1"
  28 )
  29
  30 var testAPIHost = os.Getenv("ARVADOS_API_HOST")
  31
  32 var _ = check.Suite(&IntegrationSuite{})
  33
  34 // IntegrationSuite tests need an API server and a keep-web server
  35 type IntegrationSuite struct {
  36         testServer *server
  37         ArvConfig  *arvados.Config
  38 }
  39
  40 func (s *IntegrationSuite) TestNoToken(c *check.C) {
  41         for _, token := range []string{
  42                 "",
  43                 "bogustoken",
  44         } {
  45                 hdr, body, _ := s.runCurl(c, token, "collections.example.com", "/collections/"+arvadostest.FooCollection+"/foo")
  46                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  47                 c.Check(body, check.Equals, notFoundMessage+"\n")
  48
  49                 if token != "" {
  50                         hdr, body, _ = s.runCurl(c, token, "collections.example.com", "/collections/download/"+arvadostest.FooCollection+"/"+token+"/foo")
  51                         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  52                         c.Check(body, check.Equals, notFoundMessage+"\n")
  53                 }
  54
  55                 hdr, body, _ = s.runCurl(c, token, "collections.example.com", "/bad-route")
  56                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  57                 c.Check(body, check.Equals, notFoundMessage+"\n")
  58         }
  59 }
  60
  61 // TODO: Move most cases to functional tests -- at least use Go's own
  62 // http client instead of forking curl. Just leave enough of an
  63 // integration test to assure that the documented way of invoking curl
  64 // really works against the server.
  65 func (s *IntegrationSuite) Test404(c *check.C) {
  66         for _, uri := range []string{
  67                 // Routing errors (always 404 regardless of what's stored in Keep)
  68                 "/foo",
  69                 "/download",
  70                 "/collections",
  71                 "/collections/",
  72                 // Implicit/generated index is not implemented yet;
  73                 // until then, return 404.
  74                 "/collections/" + arvadostest.FooCollection,
  75                 "/collections/" + arvadostest.FooCollection + "/",
  76                 "/collections/" + arvadostest.FooBarDirCollection + "/dir1",
  77                 "/collections/" + arvadostest.FooBarDirCollection + "/dir1/",
  78                 // Non-existent file in collection
  79                 "/collections/" + arvadostest.FooCollection + "/theperthcountyconspiracy",
  80                 "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/theperthcountyconspiracy",
  81                 // Non-existent collection
  82                 "/collections/" + arvadostest.NonexistentCollection,
  83                 "/collections/" + arvadostest.NonexistentCollection + "/",
  84                 "/collections/" + arvadostest.NonexistentCollection + "/theperthcountyconspiracy",
  85                 "/collections/download/" + arvadostest.NonexistentCollection + "/" + arvadostest.ActiveToken + "/theperthcountyconspiracy",
  86         } {
  87                 hdr, body, _ := s.runCurl(c, arvadostest.ActiveToken, "collections.example.com", uri)
  88                 c.Check(hdr, check.Matches, "(?s)HTTP/1.1 404 Not Found\r\n.*")
  89                 if len(body) > 0 {
  90                         c.Check(body, check.Equals, notFoundMessage+"\n")
  91                 }
  92         }
  93 }
  94
  95 func (s *IntegrationSuite) Test1GBFile(c *check.C) {
  96         if testing.Short() {
  97                 c.Skip("skipping 1GB integration test in short mode")
  98         }
  99         s.test100BlockFile(c, 10000000)
 100 }
 101
 102 func (s *IntegrationSuite) Test100BlockFile(c *check.C) {
 103         if testing.Short() {
 104                 // 3 MB
 105                 s.test100BlockFile(c, 30000)
 106         } else {
 107                 // 300 MB
 108                 s.test100BlockFile(c, 3000000)
 109         }
 110 }
 111
 112 func (s *IntegrationSuite) test100BlockFile(c *check.C, blocksize int) {
 113         testdata := make([]byte, blocksize)
 114         for i := 0; i < blocksize; i++ {
 115                 testdata[i] = byte(' ')
 116         }
 117         arv, err := arvadosclient.MakeArvadosClient()
 118         c.Assert(err, check.Equals, nil)
 119         arv.ApiToken = arvadostest.ActiveToken
 120         kc, err := keepclient.MakeKeepClient(arv)
 121         c.Assert(err, check.Equals, nil)
 122         loc, _, err := kc.PutB(testdata[:])
 123         c.Assert(err, check.Equals, nil)
 124         mtext := "."
 125         for i := 0; i < 100; i++ {
 126                 mtext = mtext + " " + loc
 127         }
 128         mtext = mtext + fmt.Sprintf(" 0:%d00:testdata.bin\n", blocksize)
 129         coll := map[string]interface{}{}
 130         err = arv.Create("collections",
 131                 map[string]interface{}{
 132                         "collection": map[string]interface{}{
 133                                 "name":          fmt.Sprintf("testdata blocksize=%d", blocksize),
 134                                 "manifest_text": mtext,
 135                         },
 136                 }, &coll)
 137         c.Assert(err, check.Equals, nil)
 138         uuid := coll["uuid"].(string)
 139
 140         hdr, body, size := s.runCurl(c, arv.ApiToken, uuid+".collections.example.com", "/testdata.bin")
 141         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 142         c.Check(hdr, check.Matches, `(?si).*Content-length: `+fmt.Sprintf("%d00", blocksize)+`\r\n.*`)
 143         c.Check([]byte(body)[:1234], check.DeepEquals, testdata[:1234])
 144         c.Check(size, check.Equals, int64(blocksize)*100)
 145 }
 146
 147 type curlCase struct {
 148         auth    string
 149         host    string
 150         path    string
 151         dataMD5 string
 152 }
 153
 154 func (s *IntegrationSuite) Test200(c *check.C) {
 155         s.testServer.Config.cluster.Users.AnonymousUserToken = arvadostest.AnonymousToken
 156         for _, spec := range []curlCase{
 157                 // My collection
 158                 {
 159                         auth:    arvadostest.ActiveToken,
 160                         host:    arvadostest.FooCollection + "--collections.example.com",
 161                         path:    "/foo",
 162                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 163                 },
 164                 {
 165                         auth:    arvadostest.ActiveToken,
 166                         host:    arvadostest.FooCollection + ".collections.example.com",
 167                         path:    "/foo",
 168                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 169                 },
 170                 {
 171                         host:    strings.Replace(arvadostest.FooCollectionPDH, "+", "-", 1) + ".collections.example.com",
 172                         path:    "/t=" + arvadostest.ActiveToken + "/foo",
 173                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 174                 },
 175                 {
 176                         path:    "/c=" + arvadostest.FooCollectionPDH + "/t=" + arvadostest.ActiveToken + "/foo",
 177                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 178                 },
 179                 {
 180                         path:    "/c=" + strings.Replace(arvadostest.FooCollectionPDH, "+", "-", 1) + "/t=" + arvadostest.ActiveToken + "/_/foo",
 181                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 182                 },
 183                 {
 184                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 185                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 186                 },
 187                 {
 188                         auth:    "tokensobogus",
 189                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 190                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 191                 },
 192                 {
 193                         auth:    arvadostest.ActiveToken,
 194                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 195                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 196                 },
 197                 {
 198                         auth:    arvadostest.AnonymousToken,
 199                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 200                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 201                 },
 202
 203                 // Anonymously accessible data
 204                 {
 205                         path:    "/c=" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 206                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 207                 },
 208                 {
 209                         host:    arvadostest.HelloWorldCollection + ".collections.example.com",
 210                         path:    "/Hello%20world.txt",
 211                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 212                 },
 213                 {
 214                         host:    arvadostest.HelloWorldCollection + ".collections.example.com",
 215                         path:    "/_/Hello%20world.txt",
 216                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 217                 },
 218                 {
 219                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 220                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 221                 },
 222                 {
 223                         auth:    arvadostest.ActiveToken,
 224                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 225                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 226                 },
 227                 {
 228                         auth:    arvadostest.SpectatorToken,
 229                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 230                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 231                 },
 232                 {
 233                         auth:    arvadostest.SpectatorToken,
 234                         host:    arvadostest.HelloWorldCollection + "--collections.example.com",
 235                         path:    "/Hello%20world.txt",
 236                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 237                 },
 238                 {
 239                         auth:    arvadostest.SpectatorToken,
 240                         path:    "/collections/download/" + arvadostest.HelloWorldCollection + "/" + arvadostest.SpectatorToken + "/Hello%20world.txt",
 241                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 242                 },
 243         } {
 244                 host := spec.host
 245                 if host == "" {
 246                         host = "collections.example.com"
 247                 }
 248                 hdr, body, _ := s.runCurl(c, spec.auth, host, spec.path)
 249                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 250                 if strings.HasSuffix(spec.path, ".txt") {
 251                         c.Check(hdr, check.Matches, `(?s).*\r\nContent-Type: text/plain.*`)
 252                         // TODO: Check some types that aren't
 253                         // automatically detected by Go's http server
 254                         // by sniffing the content.
 255                 }
 256                 c.Check(fmt.Sprintf("%x", md5.Sum([]byte(body))), check.Equals, spec.dataMD5)
 257         }
 258 }
 259
 260 // Return header block and body.
 261 func (s *IntegrationSuite) runCurl(c *check.C, auth, host, uri string, args ...string) (hdr, bodyPart string, bodySize int64) {
 262         curlArgs := []string{"--silent", "--show-error", "--include"}
 263         testHost, testPort, _ := net.SplitHostPort(s.testServer.Addr)
 264         curlArgs = append(curlArgs, "--resolve", host+":"+testPort+":"+testHost)
 265         if strings.Contains(auth, " ") {
 266                 // caller supplied entire Authorization header value
 267                 curlArgs = append(curlArgs, "-H", "Authorization: "+auth)
 268         } else if auth != "" {
 269                 // caller supplied Arvados token
 270                 curlArgs = append(curlArgs, "-H", "Authorization: Bearer "+auth)
 271         }
 272         curlArgs = append(curlArgs, args...)
 273         curlArgs = append(curlArgs, "http://"+host+":"+testPort+uri)
 274         c.Log(fmt.Sprintf("curlArgs == %#v", curlArgs))
 275         cmd := exec.Command("curl", curlArgs...)
 276         stdout, err := cmd.StdoutPipe()
 277         c.Assert(err, check.IsNil)
 278         cmd.Stderr = os.Stderr
 279         err = cmd.Start()
 280         c.Assert(err, check.IsNil)
 281         buf := make([]byte, 2<<27)
 282         n, err := io.ReadFull(stdout, buf)
 283         // Discard (but measure size of) anything past 128 MiB.
 284         var discarded int64
 285         if err == io.ErrUnexpectedEOF {
 286                 buf = buf[:n]
 287         } else {
 288                 c.Assert(err, check.IsNil)
 289                 discarded, err = io.Copy(ioutil.Discard, stdout)
 290                 c.Assert(err, check.IsNil)
 291         }
 292         err = cmd.Wait()
 293         // Without "-f", curl exits 0 as long as it gets a valid HTTP
 294         // response from the server, even if the response status
 295         // indicates that the request failed. In our test suite, we
 296         // always expect a valid HTTP response, and we parse the
 297         // headers ourselves. If curl exits non-zero, our testing
 298         // environment is broken.
 299         c.Assert(err, check.Equals, nil)
 300         hdrsAndBody := strings.SplitN(string(buf), "\r\n\r\n", 2)
 301         c.Assert(len(hdrsAndBody), check.Equals, 2)
 302         hdr = hdrsAndBody[0]
 303         bodyPart = hdrsAndBody[1]
 304         bodySize = int64(len(bodyPart)) + discarded
 305         return
 306 }
 307
 308 func (s *IntegrationSuite) TestMetrics(c *check.C) {
 309         s.testServer.Config.cluster.Services.WebDAVDownload.ExternalURL.Host = s.testServer.Addr
 310         origin := "http://" + s.testServer.Addr
 311         req, _ := http.NewRequest("GET", origin+"/notfound", nil)
 312         _, err := http.DefaultClient.Do(req)
 313         c.Assert(err, check.IsNil)
 314         req, _ = http.NewRequest("GET", origin+"/by_id/", nil)
 315         req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 316         resp, err := http.DefaultClient.Do(req)
 317         c.Assert(err, check.IsNil)
 318         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 319         for i := 0; i < 2; i++ {
 320                 req, _ = http.NewRequest("GET", origin+"/foo", nil)
 321                 req.Host = arvadostest.FooCollection + ".example.com"
 322                 req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 323                 resp, err = http.DefaultClient.Do(req)
 324                 c.Assert(err, check.IsNil)
 325                 c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 326                 buf, _ := ioutil.ReadAll(resp.Body)
 327                 c.Check(buf, check.DeepEquals, []byte("foo"))
 328                 resp.Body.Close()
 329         }
 330
 331         s.testServer.Config.Cache.updateGauges()
 332
 333         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 334         resp, err = http.DefaultClient.Do(req)
 335         c.Assert(err, check.IsNil)
 336         c.Check(resp.StatusCode, check.Equals, http.StatusUnauthorized)
 337
 338         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 339         req.Header.Set("Authorization", "Bearer badtoken")
 340         resp, err = http.DefaultClient.Do(req)
 341         c.Assert(err, check.IsNil)
 342         c.Check(resp.StatusCode, check.Equals, http.StatusForbidden)
 343
 344         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 345         req.Header.Set("Authorization", "Bearer "+arvadostest.ManagementToken)
 346         resp, err = http.DefaultClient.Do(req)
 347         c.Assert(err, check.IsNil)
 348         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 349         type summary struct {
 350                 SampleCount string
 351                 SampleSum   float64
 352         }
 353         type counter struct {
 354                 Value int64
 355         }
 356         type gauge struct {
 357                 Value float64
 358         }
 359         var ents []struct {
 360                 Name   string
 361                 Help   string
 362                 Type   string
 363                 Metric []struct {
 364                         Label []struct {
 365                                 Name  string
 366                                 Value string
 367                         }
 368                         Counter counter
 369                         Gauge   gauge
 370                         Summary summary
 371                 }
 372         }
 373         json.NewDecoder(resp.Body).Decode(&ents)
 374         summaries := map[string]summary{}
 375         gauges := map[string]gauge{}
 376         counters := map[string]counter{}
 377         for _, e := range ents {
 378                 for _, m := range e.Metric {
 379                         labels := map[string]string{}
 380                         for _, lbl := range m.Label {
 381                                 labels[lbl.Name] = lbl.Value
 382                         }
 383                         summaries[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Summary
 384                         counters[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Counter
 385                         gauges[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Gauge
 386                 }
 387         }
 388         c.Check(summaries["request_duration_seconds/get/200"].SampleSum, check.Not(check.Equals), 0)
 389         c.Check(summaries["request_duration_seconds/get/200"].SampleCount, check.Equals, "3")
 390         c.Check(summaries["request_duration_seconds/get/404"].SampleCount, check.Equals, "1")
 391         c.Check(summaries["time_to_status_seconds/get/404"].SampleCount, check.Equals, "1")
 392         c.Check(counters["arvados_keepweb_collectioncache_requests//"].Value, check.Equals, int64(2))
 393         c.Check(counters["arvados_keepweb_collectioncache_api_calls//"].Value, check.Equals, int64(2))
 394         c.Check(counters["arvados_keepweb_collectioncache_hits//"].Value, check.Equals, int64(1))
 395         c.Check(counters["arvados_keepweb_collectioncache_pdh_hits//"].Value, check.Equals, int64(1))
 396         c.Check(gauges["arvados_keepweb_collectioncache_cached_manifests//"].Value, check.Equals, float64(1))
 397         // FooCollection's cached manifest size is 45 ("1f4b0....+45") plus one 51-byte blob signature
 398         c.Check(gauges["arvados_keepweb_sessions_cached_collection_bytes//"].Value, check.Equals, float64(45+51))
 399
 400         // If the Host header indicates a collection, /metrics.json
 401         // refers to a file in the collection -- the metrics handler
 402         // must not intercept that route.
 403         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 404         req.Host = strings.Replace(arvadostest.FooCollectionPDH, "+", "-", -1) + ".example.com"
 405         req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 406         resp, err = http.DefaultClient.Do(req)
 407         c.Assert(err, check.IsNil)
 408         c.Check(resp.StatusCode, check.Equals, http.StatusNotFound)
 409 }
 410
 411 func (s *IntegrationSuite) SetUpSuite(c *check.C) {
 412         arvadostest.ResetDB(c)
 413         arvadostest.StartKeep(2, true)
 414
 415         arv, err := arvadosclient.MakeArvadosClient()
 416         c.Assert(err, check.Equals, nil)
 417         arv.ApiToken = arvadostest.ActiveToken
 418         kc, err := keepclient.MakeKeepClient(arv)
 419         c.Assert(err, check.Equals, nil)
 420         kc.PutB([]byte("Hello world\n"))
 421         kc.PutB([]byte("foo"))
 422         kc.PutB([]byte("foobar"))
 423         kc.PutB([]byte("waz"))
 424 }
 425
 426 func (s *IntegrationSuite) TearDownSuite(c *check.C) {
 427         arvadostest.StopKeep(2)
 428 }
 429
 430 func (s *IntegrationSuite) SetUpTest(c *check.C) {
 431         arvadostest.ResetEnv()
 432         ldr := config.NewLoader(bytes.NewBufferString("Clusters: {zzzzz: {}}"), ctxlog.TestLogger(c))
 433         ldr.Path = "-"
 434         arvCfg, err := ldr.Load()
 435         c.Check(err, check.IsNil)
 436         cfg := newConfig(ctxlog.TestLogger(c), arvCfg)
 437         c.Assert(err, check.IsNil)
 438         cfg.Client = arvados.Client{
 439                 APIHost:  testAPIHost,
 440                 Insecure: true,
 441         }
 442         listen := "127.0.0.1:0"
 443         cfg.cluster.Services.WebDAV.InternalURLs[arvados.URL{Host: listen}] = arvados.ServiceInstance{}
 444         cfg.cluster.Services.WebDAVDownload.InternalURLs[arvados.URL{Host: listen}] = arvados.ServiceInstance{}
 445         cfg.cluster.ManagementToken = arvadostest.ManagementToken
 446         cfg.cluster.SystemRootToken = arvadostest.SystemRootToken
 447         cfg.cluster.Users.AnonymousUserToken = arvadostest.AnonymousToken
 448         s.ArvConfig = arvCfg
 449         s.testServer = &server{Config: cfg}
 450         err = s.testServer.Start(ctxlog.TestLogger(c))
 451         c.Assert(err, check.Equals, nil)
 452 }
 453
 454 func (s *IntegrationSuite) TearDownTest(c *check.C) {
 455         var err error
 456         if s.testServer != nil {
 457                 err = s.testServer.Close()
 458         }
 459         c.Check(err, check.Equals, nil)
 460 }
 461
 462 // Gocheck boilerplate
 463 func Test(t *testing.T) {
 464         check.TestingT(t)
 465 }