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         "crypto/md5"
   9         "encoding/json"
  10         "fmt"
  11         "io"
  12         "io/ioutil"
  13         "net"
  14         "net/http"
  15         "os"
  16         "os/exec"
  17         "strings"
  18         "testing"
  19
  20         "git.curoverse.com/arvados.git/sdk/go/arvados"
  21         "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
  22         "git.curoverse.com/arvados.git/sdk/go/arvadostest"
  23         "git.curoverse.com/arvados.git/sdk/go/keepclient"
  24         check "gopkg.in/check.v1"
  25 )
  26
  27 var testAPIHost = os.Getenv("ARVADOS_API_HOST")
  28
  29 var _ = check.Suite(&IntegrationSuite{})
  30
  31 // IntegrationSuite tests need an API server and a keep-web server
  32 type IntegrationSuite struct {
  33         testServer *server
  34 }
  35
  36 func (s *IntegrationSuite) TestNoToken(c *check.C) {
  37         for _, token := range []string{
  38                 "",
  39                 "bogustoken",
  40         } {
  41                 hdr, body, _ := s.runCurl(c, token, "collections.example.com", "/collections/"+arvadostest.FooCollection+"/foo")
  42                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  43                 c.Check(body, check.Equals, "")
  44
  45                 if token != "" {
  46                         hdr, body, _ = s.runCurl(c, token, "collections.example.com", "/collections/download/"+arvadostest.FooCollection+"/"+token+"/foo")
  47                         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  48                         c.Check(body, check.Equals, "")
  49                 }
  50
  51                 hdr, body, _ = s.runCurl(c, token, "collections.example.com", "/bad-route")
  52                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 404 Not Found\r\n.*`)
  53                 c.Check(body, check.Equals, "")
  54         }
  55 }
  56
  57 // TODO: Move most cases to functional tests -- at least use Go's own
  58 // http client instead of forking curl. Just leave enough of an
  59 // integration test to assure that the documented way of invoking curl
  60 // really works against the server.
  61 func (s *IntegrationSuite) Test404(c *check.C) {
  62         for _, uri := range []string{
  63                 // Routing errors (always 404 regardless of what's stored in Keep)
  64                 "/foo",
  65                 "/download",
  66                 "/collections",
  67                 "/collections/",
  68                 // Implicit/generated index is not implemented yet;
  69                 // until then, return 404.
  70                 "/collections/" + arvadostest.FooCollection,
  71                 "/collections/" + arvadostest.FooCollection + "/",
  72                 "/collections/" + arvadostest.FooBarDirCollection + "/dir1",
  73                 "/collections/" + arvadostest.FooBarDirCollection + "/dir1/",
  74                 // Non-existent file in collection
  75                 "/collections/" + arvadostest.FooCollection + "/theperthcountyconspiracy",
  76                 "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/theperthcountyconspiracy",
  77                 // Non-existent collection
  78                 "/collections/" + arvadostest.NonexistentCollection,
  79                 "/collections/" + arvadostest.NonexistentCollection + "/",
  80                 "/collections/" + arvadostest.NonexistentCollection + "/theperthcountyconspiracy",
  81                 "/collections/download/" + arvadostest.NonexistentCollection + "/" + arvadostest.ActiveToken + "/theperthcountyconspiracy",
  82         } {
  83                 hdr, body, _ := s.runCurl(c, arvadostest.ActiveToken, "collections.example.com", uri)
  84                 c.Check(hdr, check.Matches, "(?s)HTTP/1.1 404 Not Found\r\n.*")
  85                 if len(body) > 0 {
  86                         c.Check(body, check.Equals, "404 page not found\n")
  87                 }
  88         }
  89 }
  90
  91 func (s *IntegrationSuite) Test1GBFile(c *check.C) {
  92         if testing.Short() {
  93                 c.Skip("skipping 1GB integration test in short mode")
  94         }
  95         s.test100BlockFile(c, 10000000)
  96 }
  97
  98 func (s *IntegrationSuite) Test100BlockFile(c *check.C) {
  99         if testing.Short() {
 100                 // 3 MB
 101                 s.test100BlockFile(c, 30000)
 102         } else {
 103                 // 300 MB
 104                 s.test100BlockFile(c, 3000000)
 105         }
 106 }
 107
 108 func (s *IntegrationSuite) test100BlockFile(c *check.C, blocksize int) {
 109         testdata := make([]byte, blocksize)
 110         for i := 0; i < blocksize; i++ {
 111                 testdata[i] = byte(' ')
 112         }
 113         arv, err := arvadosclient.MakeArvadosClient()
 114         c.Assert(err, check.Equals, nil)
 115         arv.ApiToken = arvadostest.ActiveToken
 116         kc, err := keepclient.MakeKeepClient(arv)
 117         c.Assert(err, check.Equals, nil)
 118         loc, _, err := kc.PutB(testdata[:])
 119         c.Assert(err, check.Equals, nil)
 120         mtext := "."
 121         for i := 0; i < 100; i++ {
 122                 mtext = mtext + " " + loc
 123         }
 124         mtext = mtext + fmt.Sprintf(" 0:%d00:testdata.bin\n", blocksize)
 125         coll := map[string]interface{}{}
 126         err = arv.Create("collections",
 127                 map[string]interface{}{
 128                         "collection": map[string]interface{}{
 129                                 "name":          fmt.Sprintf("testdata blocksize=%d", blocksize),
 130                                 "manifest_text": mtext,
 131                         },
 132                 }, &coll)
 133         c.Assert(err, check.Equals, nil)
 134         uuid := coll["uuid"].(string)
 135
 136         hdr, body, size := s.runCurl(c, arv.ApiToken, uuid+".collections.example.com", "/testdata.bin")
 137         c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 138         c.Check(hdr, check.Matches, `(?si).*Content-length: `+fmt.Sprintf("%d00", blocksize)+`\r\n.*`)
 139         c.Check([]byte(body)[:1234], check.DeepEquals, testdata[:1234])
 140         c.Check(size, check.Equals, int64(blocksize)*100)
 141 }
 142
 143 type curlCase struct {
 144         auth    string
 145         host    string
 146         path    string
 147         dataMD5 string
 148 }
 149
 150 func (s *IntegrationSuite) Test200(c *check.C) {
 151         s.testServer.Config.AnonymousTokens = []string{arvadostest.AnonymousToken}
 152         for _, spec := range []curlCase{
 153                 // My collection
 154                 {
 155                         auth:    arvadostest.ActiveToken,
 156                         host:    arvadostest.FooCollection + "--collections.example.com",
 157                         path:    "/foo",
 158                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 159                 },
 160                 {
 161                         auth:    arvadostest.ActiveToken,
 162                         host:    arvadostest.FooCollection + ".collections.example.com",
 163                         path:    "/foo",
 164                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 165                 },
 166                 {
 167                         host:    strings.Replace(arvadostest.FooPdh, "+", "-", 1) + ".collections.example.com",
 168                         path:    "/t=" + arvadostest.ActiveToken + "/foo",
 169                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 170                 },
 171                 {
 172                         path:    "/c=" + arvadostest.FooPdh + "/t=" + arvadostest.ActiveToken + "/foo",
 173                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 174                 },
 175                 {
 176                         path:    "/c=" + strings.Replace(arvadostest.FooPdh, "+", "-", 1) + "/t=" + arvadostest.ActiveToken + "/_/foo",
 177                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 178                 },
 179                 {
 180                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 181                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 182                 },
 183                 {
 184                         auth:    "tokensobogus",
 185                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 186                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 187                 },
 188                 {
 189                         auth:    arvadostest.ActiveToken,
 190                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 191                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 192                 },
 193                 {
 194                         auth:    arvadostest.AnonymousToken,
 195                         path:    "/collections/download/" + arvadostest.FooCollection + "/" + arvadostest.ActiveToken + "/foo",
 196                         dataMD5: "acbd18db4cc2f85cedef654fccc4a4d8",
 197                 },
 198
 199                 // Anonymously accessible data
 200                 {
 201                         path:    "/c=" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 202                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 203                 },
 204                 {
 205                         host:    arvadostest.HelloWorldCollection + ".collections.example.com",
 206                         path:    "/Hello%20world.txt",
 207                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 208                 },
 209                 {
 210                         host:    arvadostest.HelloWorldCollection + ".collections.example.com",
 211                         path:    "/_/Hello%20world.txt",
 212                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 213                 },
 214                 {
 215                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 216                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 217                 },
 218                 {
 219                         auth:    arvadostest.ActiveToken,
 220                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 221                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 222                 },
 223                 {
 224                         auth:    arvadostest.SpectatorToken,
 225                         path:    "/collections/" + arvadostest.HelloWorldCollection + "/Hello%20world.txt",
 226                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 227                 },
 228                 {
 229                         auth:    arvadostest.SpectatorToken,
 230                         host:    arvadostest.HelloWorldCollection + "--collections.example.com",
 231                         path:    "/Hello%20world.txt",
 232                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 233                 },
 234                 {
 235                         auth:    arvadostest.SpectatorToken,
 236                         path:    "/collections/download/" + arvadostest.HelloWorldCollection + "/" + arvadostest.SpectatorToken + "/Hello%20world.txt",
 237                         dataMD5: "f0ef7081e1539ac00ef5b761b4fb01b3",
 238                 },
 239         } {
 240                 host := spec.host
 241                 if host == "" {
 242                         host = "collections.example.com"
 243                 }
 244                 hdr, body, _ := s.runCurl(c, spec.auth, host, spec.path)
 245                 c.Check(hdr, check.Matches, `(?s)HTTP/1.1 200 OK\r\n.*`)
 246                 if strings.HasSuffix(spec.path, ".txt") {
 247                         c.Check(hdr, check.Matches, `(?s).*\r\nContent-Type: text/plain.*`)
 248                         // TODO: Check some types that aren't
 249                         // automatically detected by Go's http server
 250                         // by sniffing the content.
 251                 }
 252                 c.Check(fmt.Sprintf("%x", md5.Sum([]byte(body))), check.Equals, spec.dataMD5)
 253         }
 254 }
 255
 256 // Return header block and body.
 257 func (s *IntegrationSuite) runCurl(c *check.C, token, host, uri string, args ...string) (hdr, bodyPart string, bodySize int64) {
 258         curlArgs := []string{"--silent", "--show-error", "--include"}
 259         testHost, testPort, _ := net.SplitHostPort(s.testServer.Addr)
 260         curlArgs = append(curlArgs, "--resolve", host+":"+testPort+":"+testHost)
 261         if token != "" {
 262                 curlArgs = append(curlArgs, "-H", "Authorization: OAuth2 "+token)
 263         }
 264         curlArgs = append(curlArgs, args...)
 265         curlArgs = append(curlArgs, "http://"+host+":"+testPort+uri)
 266         c.Log(fmt.Sprintf("curlArgs == %#v", curlArgs))
 267         cmd := exec.Command("curl", curlArgs...)
 268         stdout, err := cmd.StdoutPipe()
 269         c.Assert(err, check.Equals, nil)
 270         cmd.Stderr = cmd.Stdout
 271         go cmd.Start()
 272         buf := make([]byte, 2<<27)
 273         n, err := io.ReadFull(stdout, buf)
 274         // Discard (but measure size of) anything past 128 MiB.
 275         var discarded int64
 276         if err == io.ErrUnexpectedEOF {
 277                 buf = buf[:n]
 278         } else {
 279                 c.Assert(err, check.Equals, nil)
 280                 discarded, err = io.Copy(ioutil.Discard, stdout)
 281                 c.Assert(err, check.Equals, nil)
 282         }
 283         err = cmd.Wait()
 284         // Without "-f", curl exits 0 as long as it gets a valid HTTP
 285         // response from the server, even if the response status
 286         // indicates that the request failed. In our test suite, we
 287         // always expect a valid HTTP response, and we parse the
 288         // headers ourselves. If curl exits non-zero, our testing
 289         // environment is broken.
 290         c.Assert(err, check.Equals, nil)
 291         hdrsAndBody := strings.SplitN(string(buf), "\r\n\r\n", 2)
 292         c.Assert(len(hdrsAndBody), check.Equals, 2)
 293         hdr = hdrsAndBody[0]
 294         bodyPart = hdrsAndBody[1]
 295         bodySize = int64(len(bodyPart)) + discarded
 296         return
 297 }
 298
 299 func (s *IntegrationSuite) TestMetrics(c *check.C) {
 300         origin := "http://" + s.testServer.Addr
 301         req, _ := http.NewRequest("GET", origin+"/notfound", nil)
 302         _, err := http.DefaultClient.Do(req)
 303         c.Assert(err, check.IsNil)
 304         req, _ = http.NewRequest("GET", origin+"/by_id/", nil)
 305         req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 306         resp, err := http.DefaultClient.Do(req)
 307         c.Assert(err, check.IsNil)
 308         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 309         for i := 0; i < 2; i++ {
 310                 req, _ = http.NewRequest("GET", origin+"/foo", nil)
 311                 req.Host = arvadostest.FooCollection + ".example.com"
 312                 req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 313                 resp, err = http.DefaultClient.Do(req)
 314                 c.Assert(err, check.IsNil)
 315                 c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 316                 buf, _ := ioutil.ReadAll(resp.Body)
 317                 c.Check(buf, check.DeepEquals, []byte("foo"))
 318                 resp.Body.Close()
 319         }
 320
 321         s.testServer.Config.Cache.updateGauges()
 322
 323         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 324         resp, err = http.DefaultClient.Do(req)
 325         c.Assert(err, check.IsNil)
 326         c.Check(resp.StatusCode, check.Equals, http.StatusUnauthorized)
 327
 328         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 329         req.Header.Set("Authorization", "Bearer badtoken")
 330         resp, err = http.DefaultClient.Do(req)
 331         c.Assert(err, check.IsNil)
 332         c.Check(resp.StatusCode, check.Equals, http.StatusForbidden)
 333
 334         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 335         req.Header.Set("Authorization", "Bearer "+arvadostest.ManagementToken)
 336         resp, err = http.DefaultClient.Do(req)
 337         c.Assert(err, check.IsNil)
 338         c.Check(resp.StatusCode, check.Equals, http.StatusOK)
 339         type summary struct {
 340                 SampleCount string  `json:"sample_count"`
 341                 SampleSum   float64 `json:"sample_sum"`
 342                 Quantile    []struct {
 343                         Quantile float64
 344                         Value    float64
 345                 }
 346         }
 347         type counter struct {
 348                 Value int64
 349         }
 350         type gauge struct {
 351                 Value float64
 352         }
 353         var ents []struct {
 354                 Name   string
 355                 Help   string
 356                 Type   string
 357                 Metric []struct {
 358                         Label []struct {
 359                                 Name  string
 360                                 Value string
 361                         }
 362                         Counter counter
 363                         Gauge   gauge
 364                         Summary summary
 365                 }
 366         }
 367         json.NewDecoder(resp.Body).Decode(&ents)
 368         summaries := map[string]summary{}
 369         gauges := map[string]gauge{}
 370         counters := map[string]counter{}
 371         for _, e := range ents {
 372                 for _, m := range e.Metric {
 373                         labels := map[string]string{}
 374                         for _, lbl := range m.Label {
 375                                 labels[lbl.Name] = lbl.Value
 376                         }
 377                         summaries[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Summary
 378                         counters[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Counter
 379                         gauges[e.Name+"/"+labels["method"]+"/"+labels["code"]] = m.Gauge
 380                 }
 381         }
 382         c.Check(summaries["request_duration_seconds/get/200"].SampleSum, check.Not(check.Equals), 0)
 383         c.Check(summaries["request_duration_seconds/get/200"].SampleCount, check.Equals, "3")
 384         c.Check(summaries["request_duration_seconds/get/404"].SampleCount, check.Equals, "1")
 385         c.Check(summaries["time_to_status_seconds/get/404"].SampleCount, check.Equals, "1")
 386         c.Check(counters["arvados_keepweb_collectioncache_requests//"].Value, check.Equals, int64(2))
 387         c.Check(counters["arvados_keepweb_collectioncache_api_calls//"].Value, check.Equals, int64(1))
 388         c.Check(counters["arvados_keepweb_collectioncache_hits//"].Value, check.Equals, int64(1))
 389         c.Check(counters["arvados_keepweb_collectioncache_pdh_hits//"].Value, check.Equals, int64(1))
 390         c.Check(counters["arvados_keepweb_collectioncache_permission_hits//"].Value, check.Equals, int64(1))
 391         c.Check(gauges["arvados_keepweb_collectioncache_cached_manifests//"].Value, check.Equals, float64(1))
 392         // FooCollection's cached manifest size is 45 ("1f4b0....+45") plus one 51-byte blob signature
 393         c.Check(gauges["arvados_keepweb_collectioncache_cached_manifest_bytes//"].Value, check.Equals, float64(45+51))
 394
 395         // If the Host header indicates a collection, /metrics.json
 396         // refers to a file in the collection -- the metrics handler
 397         // must not intercept that route.
 398         req, _ = http.NewRequest("GET", origin+"/metrics.json", nil)
 399         req.Host = strings.Replace(arvadostest.FooCollectionPDH, "+", "-", -1) + ".example.com"
 400         req.Header.Set("Authorization", "Bearer "+arvadostest.ActiveToken)
 401         resp, err = http.DefaultClient.Do(req)
 402         c.Assert(err, check.IsNil)
 403         c.Check(resp.StatusCode, check.Equals, http.StatusNotFound)
 404 }
 405
 406 func (s *IntegrationSuite) SetUpSuite(c *check.C) {
 407         arvadostest.StartAPI()
 408         arvadostest.StartKeep(2, true)
 409
 410         arv, err := arvadosclient.MakeArvadosClient()
 411         c.Assert(err, check.Equals, nil)
 412         arv.ApiToken = arvadostest.ActiveToken
 413         kc, err := keepclient.MakeKeepClient(arv)
 414         c.Assert(err, check.Equals, nil)
 415         kc.PutB([]byte("Hello world\n"))
 416         kc.PutB([]byte("foo"))
 417         kc.PutB([]byte("foobar"))
 418         kc.PutB([]byte("waz"))
 419 }
 420
 421 func (s *IntegrationSuite) TearDownSuite(c *check.C) {
 422         arvadostest.StopKeep(2)
 423         arvadostest.StopAPI()
 424 }
 425
 426 func (s *IntegrationSuite) SetUpTest(c *check.C) {
 427         arvadostest.ResetEnv()
 428         cfg := DefaultConfig()
 429         cfg.Client = arvados.Client{
 430                 APIHost:  testAPIHost,
 431                 Insecure: true,
 432         }
 433         cfg.Listen = "127.0.0.1:0"
 434         cfg.ManagementToken = arvadostest.ManagementToken
 435         s.testServer = &server{Config: cfg}
 436         err := s.testServer.Start()
 437         c.Assert(err, check.Equals, nil)
 438 }
 439
 440 func (s *IntegrationSuite) TearDownTest(c *check.C) {
 441         var err error
 442         if s.testServer != nil {
 443                 err = s.testServer.Close()
 444         }
 445         c.Check(err, check.Equals, nil)
 446 }
 447
 448 // Gocheck boilerplate
 449 func Test(t *testing.T) {
 450         check.TestingT(t)
 451 }