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.curoverse.com/arvados.git/lib/config"
  22         "git.curoverse.com/arvados.git/sdk/go/arvados"
  23         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  24         "git.curoverse.com/arvados.git/sdk/go/arvadostest"
  25         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  26         check "gopkg.in/check.v1"
  27 )
  28
  29 var testAPIHost = os.Getenv("ARVADOS_API_HOST")
  30
  31 var _ = check.Suite(&IntegrationSuite{})
  32
  33 // IntegrationSuite tests need an API server and a keep-web server
  34 type IntegrationSuite struct {
  35         testServer *server
  36 }
  37
  38 func (s *IntegrationSuite) TestNoToken(c *check.C) {
  39         for _, token := range []string{
  40                 "",
  41                 "bogustoken",
  42         } {
  43                 hdr, body, _ := s.runCurl(c, token, "collections.example.com", "/collections/"+arvadostest.FooCollection+"/foo")
  44                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  45                 c.Check(body, check.Equals, "")
  46
  47                 if token != "" {
  48                         hdr, body, _ = s.runCurl(c, token, "collections.example.com", "/collections/download/"+arvadostest.FooCollection+"/"+token+"/foo")
  49                         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  50                         c.Check(body, check.Equals, "")
  51                 }
  52
  53                 hdr, body, _ = s.runCurl(c, token, "collections.example.com", "/bad-route")
  54                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  55                 c.Check(body, check.Equals, "")
  56         }
  57 }
  58
  59 // TODO: Move most cases to functional tests -- at least use Go's own
  60 // http client instead of forking curl. Just leave enough of an
  61 // integration test to assure that the documented way of invoking curl
  62 // really works against the server.
  63 func (s *IntegrationSuite) Test404(c *check.C) {
  64         for _, uri := range []string{
  65                 // Routing errors (always 404 regardless of what's stored in Keep)
  66                 "/foo",
  67                 "/download",
  68                 "/collections",
  69                 "/collections/",
  70                 // Implicit/generated index is not implemented yet;
  71                 // until then, return 404.
  72                 "/collections/" + arvadostest.FooCollection,
  73                 "/collections/" + arvadostest.FooCollection + "/",
  74                 "/collections/" + arvadostest.FooBarDirCollection + "/dir1",
  75                 "/collections/" + arvadostest.FooBarDirCollection + "/dir1/",
  76                 // Non-existent file in collection
  77                 "/collections/" + arvadostest.FooCollection + "/theperthcountyconspiracy",
  78                 "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/theperthcountyconspiracy",
  79                 // Non-existent collection
  80                 "/collections/" + arvadostest.NonexistentCollection,
  81                 "/collections/" + arvadostest.NonexistentCollection + "/",
  82                 "/collections/" + arvadostest.NonexistentCollection + "/theperthcountyconspiracy",
  83                 "/collections/download/" + arvadostest.NonexistentCollection + "/" + arvadostest.ActiveToken + "/theperthcountyconspiracy",
  84         } {
  85                 hdr, body, _ := s.runCurl(c, arvadostest.ActiveToken, "collections.example.com", uri)
  86                 c.Check(hdr, check.Matches, "(?s)HTTP/1.1 404 Not Found\r\n.*")
  87                 if len(body) > 0 {
  88                         c.Check(body, check.Equals, "404 page not found\n")
  89                 }
  90         }
  91 }
  92
  93 func (s *IntegrationSuite) Test1GBFile(c *check.C) {
  94         if testing.Short() {
  95                 c.Skip("skipping 1GB integration test in short mode")
  96         }
  97         s.test100BlockFile(c, 10000000)
  98 }
  99
 100 func (s *IntegrationSuite) Test100BlockFile(c *check.C) {
 101         if testing.Short() {
 102                 // 3 MB
 103                 s.test100BlockFile(c, 30000)
 104         } else {
 105                 // 300 MB
 106                 s.test100BlockFile(c, 3000000)
 107         }
 108 }
 109
 110 func (s *IntegrationSuite) test100BlockFile(c *check.C, blocksize int) {
 111         testdata := make([]byte, blocksize)
 112         for i := 0; i < blocksize; i++ {
 113                 testdata[i] = byte(' ')
 114         }
 115         arv, err := arvadosclient.MakeArvadosClient()
 116         c.Assert(err, check.Equals, nil)
 117         arv.ApiToken = arvadostest.ActiveToken
 118         kc, err := keepclient.MakeKeepClient(arv)
 119         c.Assert(err, check.Equals, nil)
 120         loc, _, err := kc.PutB(testdata[:])
 121         c.Assert(err, check.Equals, nil)
 122         mtext := "."
 123         for i := 0; i < 100; i++ {
 124                 mtext = mtext + " " + loc
 125         }
 126         mtext = mtext + fmt.Sprintf(" 0:%d00:testdata.bin\n", blocksize)
 127         coll := map[string]interface{}{}
 128         err = arv.Create("collections",
 129                 map[string]interface{}{
 130                         "collection": map[string]interface{}{
 131                                 "name":          fmt.Sprintf("testdata blocksize=%d", blocksize),
 132                                 "manifest_text": mtext,
 133                         },
 134                 }, &coll)
 135         c.Assert(err, check.Equals, nil)
 136         uuid := coll["uuid"].(string)
 137
 138         hdr, body, size := s.runCurl(c, arv.ApiToken, uuid+".collections.example.com", "/testdata.bin")
 139         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 140         c.Check(hdr, check.Matches, `(?si).*Content-length: `+fmt.Sprintf("%d00", blocksize)+`\r\n.*`)
 141         c.Check([]byte(body)[:1234], check.DeepEquals, testdata[:1234])
 142         c.Check(size, check.Equals, int64(blocksize)*100)
 143 }
 144
 145 type curlCase struct {
 146         auth    string
 147         host    string
 148         path    string
 149         dataMD5 string
 150 }
 151
 152 func (s *IntegrationSuite) Test200(c *check.C) {
 153         s.testServer.Config.cluster.Users.AnonymousUserToken = arvadostest.AnonymousToken
 154         for _, spec := range []curlCase{
 155                 // My collection
 156                 {
 157                         auth:    arvadostest.ActiveToken,
 158                         host:    arvadostest.FooCollection + "--collections.example.com",
 159                         path:    "/foo",
 160                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 161                 },
 162                 {
 163                         auth:    arvadostest.ActiveToken,
 164                         host:    arvadostest.FooCollection + ".collections.example.com",
 165                         path:    "/foo",
 166                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 167                 },
 168                 {
 169                         host:    strings.Replace(arvadostest.FooCollectionPDH, "+", "-", 1) + ".collections.example.com",
 170                         path:    "/t=" + arvadostest.ActiveToken + "/foo",
 171                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 172                 },
 173                 {
 174                         path:    "/c=" + arvadostest.FooCollectionPDH + "/t=" + arvadostest.ActiveToken + "/foo",
 175                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 176                 },
 177                 {
 178                         path:    "/c=" + strings.Replace(arvadostest.FooCollectionPDH, "+", "-", 1) + "/t=" + arvadostest.ActiveToken + "/_/foo",
 179                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 180                 },
 181                 {
 182                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 183                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 184                 },
 185                 {
 186                         auth:    "tokensobogus",
 187                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 188                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 189                 },
 190                 {
 191                         auth:    arvadostest.ActiveToken,
 192                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 193                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 194                 },
 195                 {
 196                         auth:    arvadostest.AnonymousToken,
 197                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 198                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 199                 },
 200
 201                 // Anonymously accessible data
 202                 {
 203                         path:    "/c=" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 204                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 205                 },
 206                 {
 207                         host:    arvadostest.HelloWorldCollection + ".collections.example.com",
 208                         path:    "/Hello%20world.txt",
 209                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 210                 },
 211                 {
 212                         host:    arvadostest.HelloWorldCollection + ".collections.example.com",
 213                         path:    "/_/Hello%20world.txt",
 214                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 215                 },
 216                 {
 217                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 218                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 219                 },
 220                 {
 221                         auth:    arvadostest.ActiveToken,
 222                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 223                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 224                 },
 225                 {
 226                         auth:    arvadostest.SpectatorToken,
 227                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 228                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 229                 },
 230                 {
 231                         auth:    arvadostest.SpectatorToken,
 232                         host:    arvadostest.HelloWorldCollection + "--collections.example.com",
 233                         path:    "/Hello%20world.txt",
 234                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 235                 },
 236                 {
 237                         auth:    arvadostest.SpectatorToken,
 238                         path:    "/collections/download/" + arvadostest.HelloWorldCollection + "/" + arvadostest.SpectatorToken + "/Hello%20world.txt",
 239                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 240                 },
 241         } {
 242                 host := spec.host
 243                 if host == "" {
 244                         host = "collections.example.com"
 245                 }
 246                 hdr, body, _ := s.runCurl(c, spec.auth, host, spec.path)
 247                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 248                 if strings.HasSuffix(spec.path, ".txt") {
 249                         c.Check(hdr, check.Matches, `(?s).*\r\nContent-Type: text/plain.*`)
 250                         // TODO: Check some types that aren't
 251                         // automatically detected by Go's http server
 252                         // by sniffing the content.
 253                 }
 254                 c.Check(fmt.Sprintf("%x", md5.Sum([]byte(body))), check.Equals, spec.dataMD5)
 255         }
 256 }
 257
 258 // Return header block and body.
 259 func (s *IntegrationSuite) runCurl(c *check.C, token, host, uri string, args ...string) (hdr, bodyPart string, bodySize int64) {
 260         curlArgs := []string{"--silent", "--show-error", "--include"}
 261         testHost, testPort, _ := net.SplitHostPort(s.testServer.Addr)
 262         curlArgs = append(curlArgs, "--resolve", host+":"+testPort+":"+testHost)
 263         if token != "" {
 264                 curlArgs = append(curlArgs, "-H", "Authorization: OAuth2 "+token)
 265         }
 266         curlArgs = append(curlArgs, args...)
 267         curlArgs = append(curlArgs, "http://"+host+":"+testPort+uri)
 268         c.Log(fmt.Sprintf("curlArgs == %#v", curlArgs))
 269         cmd := exec.Command("curl", curlArgs...)
 270         stdout, err := cmd.StdoutPipe()
 271         c.Assert(err, check.IsNil)
 272         cmd.Stderr = os.Stderr
 273         err = cmd.Start()
 274         c.Assert(err, check.IsNil)
 275         buf := make([]byte, 2<<27)
 276         n, err := io.ReadFull(stdout, buf)
 277         // Discard (but measure size of) anything past 128 MiB.
 278         var discarded int64
 279         if err == io.ErrUnexpectedEOF {
 280                 buf = buf[:n]
 281         } else {
 282                 c.Assert(err, check.IsNil)
 283                 discarded, err = io.Copy(ioutil.Discard, stdout)
 284                 c.Assert(err, check.IsNil)
 285         }
 286         err = cmd.Wait()
 287         // Without "-f", curl exits 0 as long as it gets a valid HTTP
 288         // response from the server, even if the response status
 289         // indicates that the request failed. In our test suite, we
 290         // always expect a valid HTTP response, and we parse the
 291         // headers ourselves. If curl exits non-zero, our testing
 292         // environment is broken.
 293         c.Assert(err, check.Equals, nil)
 294         hdrsAndBody := strings.SplitN(string(buf), "\r\n\r\n", 2)
 295         c.Assert(len(hdrsAndBody), check.Equals, 2)
 296         hdr = hdrsAndBody[0]
 297         bodyPart = hdrsAndBody[1]
 298         bodySize = int64(len(bodyPart)) + discarded
 299         return
 300 }
 301
 302 func (s *IntegrationSuite) TestMetrics(c *check.C) {
 303         s.testServer.Config.cluster.Services.WebDAVDownload.ExternalURL.Host = s.testServer.Addr
 304         origin := "http://" + s.testServer.Addr
 305         req, _ := http.NewRequest("GET", origin+"/notfound", nil)
 306         _, err := http.DefaultClient.Do(req)
 307         c.Assert(err, check.IsNil)
 308         req, _ = http.NewRequest("GET", origin+"/by_id/", nil)
 309         req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 310         resp, err := http.DefaultClient.Do(req)
 311         c.Assert(err, check.IsNil)
 312         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 313         for i := 0; i < 2; i++ {
 314                 req, _ = http.NewRequest("GET", origin+"/foo", nil)
 315                 req.Host = arvadostest.FooCollection + ".example.com"
 316                 req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 317                 resp, err = http.DefaultClient.Do(req)
 318                 c.Assert(err, check.IsNil)
 319                 c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 320                 buf, _ := ioutil.ReadAll(resp.Body)
 321                 c.Check(buf, check.DeepEquals, []byte("foo"))
 322                 resp.Body.Close()
 323         }
 324
 325         s.testServer.Config.Cache.updateGauges()
 326
 327         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 328         resp, err = http.DefaultClient.Do(req)
 329         c.Assert(err, check.IsNil)
 330         c.Check(resp.StatusCode, check.Equals, http.StatusUnauthorized)
 331
 332         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 333         req.Header.Set("Authorization", "Bearer badtoken")
 334         resp, err = http.DefaultClient.Do(req)
 335         c.Assert(err, check.IsNil)
 336         c.Check(resp.StatusCode, check.Equals, http.StatusForbidden)
 337
 338         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 339         req.Header.Set("Authorization", "Bearer "+arvadostest.ManagementToken)
 340         resp, err = http.DefaultClient.Do(req)
 341         c.Assert(err, check.IsNil)
 342         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 343         type summary struct {
 344                 SampleCount string  `json:"sample_count"`
 345                 SampleSum   float64 `json:"sample_sum"`
 346                 Quantile    []struct {
 347                         Quantile float64
 348                         Value    float64
 349                 }
 350         }
 351         type counter struct {
 352                 Value int64
 353         }
 354         type gauge struct {
 355                 Value float64
 356         }
 357         var ents []struct {
 358                 Name   string
 359                 Help   string
 360                 Type   string
 361                 Metric []struct {
 362                         Label []struct {
 363                                 Name  string
 364                                 Value string
 365                         }
 366                         Counter counter
 367                         Gauge   gauge
 368                         Summary summary
 369                 }
 370         }
 371         json.NewDecoder(resp.Body).Decode(&ents)
 372         summaries := map[string]summary{}
 373         gauges := map[string]gauge{}
 374         counters := map[string]counter{}
 375         for _, e := range ents {
 376                 for _, m := range e.Metric {
 377                         labels := map[string]string{}
 378                         for _, lbl := range m.Label {
 379                                 labels[lbl.Name] = lbl.Value
 380                         }
 381                         summaries[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Summary
 382                         counters[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Counter
 383                         gauges[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Gauge
 384                 }
 385         }
 386         c.Check(summaries["request_duration_seconds/get/200"].SampleSum, check.Not(check.Equals), 0)
 387         c.Check(summaries["request_duration_seconds/get/200"].SampleCount, check.Equals, "3")
 388         c.Check(summaries["request_duration_seconds/get/404"].SampleCount, check.Equals, "1")
 389         c.Check(summaries["time_to_status_seconds/get/404"].SampleCount, check.Equals, "1")
 390         c.Check(counters["arvados_keepweb_collectioncache_requests//"].Value, check.Equals, int64(2))
 391         c.Check(counters["arvados_keepweb_collectioncache_api_calls//"].Value, check.Equals, int64(1))
 392         c.Check(counters["arvados_keepweb_collectioncache_hits//"].Value, check.Equals, int64(1))
 393         c.Check(counters["arvados_keepweb_collectioncache_pdh_hits//"].Value, check.Equals, int64(1))
 394         c.Check(counters["arvados_keepweb_collectioncache_permission_hits//"].Value, check.Equals, int64(1))
 395         c.Check(gauges["arvados_keepweb_collectioncache_cached_manifests//"].Value, check.Equals, float64(1))
 396         // FooCollection's cached manifest size is 45 ("1f4b0....+45") plus one 51-byte blob signature
 397         c.Check(gauges["arvados_keepweb_collectioncache_cached_manifest_bytes//"].Value, check.Equals, float64(45+51))
 398
 399         // If the Host header indicates a collection, /metrics.json
 400         // refers to a file in the collection -- the metrics handler
 401         // must not intercept that route.
 402         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 403         req.Host = strings.Replace(arvadostest.FooCollectionPDH, "+", "-", -1) + ".example.com"
 404         req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 405         resp, err = http.DefaultClient.Do(req)
 406         c.Assert(err, check.IsNil)
 407         c.Check(resp.StatusCode, check.Equals, http.StatusNotFound)
 408 }
 409
 410 func (s *IntegrationSuite) SetUpSuite(c *check.C) {
 411         arvadostest.StartAPI()
 412         arvadostest.StartKeep(2, true)
 413
 414         arv, err := arvadosclient.MakeArvadosClient()
 415         c.Assert(err, check.Equals, nil)
 416         arv.ApiToken = arvadostest.ActiveToken
 417         kc, err := keepclient.MakeKeepClient(arv)
 418         c.Assert(err, check.Equals, nil)
 419         kc.PutB([]byte("Hello world\n"))
 420         kc.PutB([]byte("foo"))
 421         kc.PutB([]byte("foobar"))
 422         kc.PutB([]byte("waz"))
 423 }
 424
 425 func (s *IntegrationSuite) TearDownSuite(c *check.C) {
 426         arvadostest.StopKeep(2)
 427         arvadostest.StopAPI()
 428 }
 429
 430 func (s *IntegrationSuite) SetUpTest(c *check.C) {
 431         arvadostest.ResetEnv()
 432         ldr := config.NewLoader(bytes.NewBufferString("Clusters: {zzzzz: {}}"), nil)
 433         ldr.Path = "-"
 434         arvCfg, err := ldr.Load()
 435         c.Check(err, check.IsNil)
 436         cfg := newConfig(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.Users.AnonymousUserToken = arvadostest.AnonymousToken
 447         s.testServer = &server{Config: cfg}
 448         err = s.testServer.Start()
 449         c.Assert(err, check.Equals, nil)
 450 }
 451
 452 func (s *IntegrationSuite) TearDownTest(c *check.C) {
 453         var err error
 454         if s.testServer != nil {
 455                 err = s.testServer.Close()
 456         }
 457         c.Check(err, check.Equals, nil)
 458 }
 459
 460 // Gocheck boilerplate
 461 func Test(t *testing.T) {
 462         check.TestingT(t)
 463 }