1 // Copyright (C) The Arvados Authors. All rights reserved.
3 // SPDX-License-Identifier: AGPL-3.0
20 "git.arvados.org/arvados.git/lib/cmd"
21 "git.arvados.org/arvados.git/sdk/go/arvados"
22 "git.arvados.org/arvados.git/sdk/go/ctxlog"
23 "github.com/sirupsen/logrus"
28 func (Command) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
30 f := flag.NewFlagSet(prog, flag.ContinueOnError)
31 f.StringVar(&diag.projectName, "project-name", "scratch area for diagnostics", "name of project to find/create in home project and use for temporary/test objects")
32 f.StringVar(&diag.logLevel, "log-level", "info", "logging level (debug, info, warning, error)")
33 f.StringVar(&diag.dockerImage, "docker-image", "alpine:latest", "image to use when running a test container")
34 f.BoolVar(&diag.checkInternal, "internal-client", false, "check that this host is considered an \"internal\" client")
35 f.BoolVar(&diag.checkExternal, "external-client", false, "check that this host is considered an \"external\" client")
36 f.IntVar(&diag.priority, "priority", 500, "priority for test container (1..1000, or 0 to skip)")
37 f.DurationVar(&diag.timeout, "timeout", 10*time.Second, "timeout for http requests")
38 if ok, code := cmd.ParseFlags(f, prog, args, "", stderr); !ok {
41 diag.logger = ctxlog.New(stdout, "text", diag.logLevel)
42 diag.logger.SetFormatter(&logrus.TextFormatter{DisableTimestamp: true, DisableLevelTruncation: true, PadLevelText: true})
44 if len(diag.errors) == 0 {
45 diag.logger.Info("--- no errors ---")
48 if diag.logger.Level > logrus.ErrorLevel {
49 fmt.Fprint(stdout, "\n--- cut here --- error summary ---\n\n")
50 for _, e := range diag.errors {
58 type diagnoser struct {
73 func (diag *diagnoser) debugf(f string, args ...interface{}) {
74 diag.logger.Debugf(" ... "+f, args...)
77 func (diag *diagnoser) infof(f string, args ...interface{}) {
78 diag.logger.Infof(" ... "+f, args...)
81 func (diag *diagnoser) warnf(f string, args ...interface{}) {
82 diag.logger.Warnf(" ... "+f, args...)
85 func (diag *diagnoser) errorf(f string, args ...interface{}) {
86 diag.logger.Errorf(f, args...)
87 diag.errors = append(diag.errors, fmt.Sprintf(f, args...))
90 // Run the given func, logging appropriate messages before and after,
91 // adding timing info, etc.
93 // The id argument should be unique among tests, and shouldn't change
94 // when other tests are added/removed.
95 func (diag *diagnoser) dotest(id int, title string, fn func() error) {
97 diag.done = map[int]bool{}
98 } else if diag.done[id] {
99 diag.errorf("(bug) reused test id %d", id)
103 diag.logger.Infof("%4d: %s", id, title)
106 elapsed := fmt.Sprintf("%d ms", time.Now().Sub(t0)/time.Millisecond)
108 diag.errorf("%4d: %s (%s): %s", id, title, elapsed, err)
110 diag.logger.Debugf("%4d: %s (%s): ok", id, title, elapsed)
114 func (diag *diagnoser) runtests() {
115 client := arvados.NewClientFromEnv()
117 if client.APIHost == "" || client.AuthToken == "" {
118 diag.errorf("ARVADOS_API_HOST and ARVADOS_API_TOKEN environment variables are not set -- aborting without running any tests")
122 var dd arvados.DiscoveryDocument
123 ddpath := "discovery/v1/apis/arvados/v1/rest"
124 diag.dotest(10, fmt.Sprintf("getting discovery document from https://%s/%s", client.APIHost, ddpath), func() error {
125 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
127 err := client.RequestAndDecodeContext(ctx, &dd, "GET", ddpath, nil, nil)
131 diag.debugf("BlobSignatureTTL = %d", dd.BlobSignatureTTL)
135 var cluster arvados.Cluster
136 cfgpath := "arvados/v1/config"
137 diag.dotest(20, fmt.Sprintf("getting exported config from https://%s/%s", client.APIHost, cfgpath), func() error {
138 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
140 err := client.RequestAndDecodeContext(ctx, &cluster, "GET", cfgpath, nil, nil)
144 diag.debugf("Collections.BlobSigning = %v", cluster.Collections.BlobSigning)
148 var user arvados.User
149 diag.dotest(30, "getting current user record", func() error {
150 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
152 err := client.RequestAndDecodeContext(ctx, &user, "GET", "arvados/v1/users/current", nil, nil)
156 diag.debugf("user uuid = %s", user.UUID)
160 // uncomment to create some spurious errors
161 // cluster.Services.WebDAVDownload.ExternalURL.Host = "0.0.0.0:9"
163 // TODO: detect routing errors here, like finding wb2 at the
165 for i, svc := range []*arvados.Service{
166 &cluster.Services.Keepproxy,
167 &cluster.Services.WebDAV,
168 &cluster.Services.WebDAVDownload,
169 &cluster.Services.Websocket,
170 &cluster.Services.Workbench1,
171 &cluster.Services.Workbench2,
173 diag.dotest(40+i, fmt.Sprintf("connecting to service endpoint %s", svc.ExternalURL), func() error {
174 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
177 if strings.HasPrefix(u.Scheme, "ws") {
178 // We can do a real websocket test elsewhere,
179 // but for now we'll just check the https
181 u.Scheme = "http" + u.Scheme[2:]
183 if svc == &cluster.Services.WebDAV && strings.HasPrefix(u.Host, "*") {
184 u.Host = "d41d8cd98f00b204e9800998ecf8427e-0" + u.Host[1:]
186 req, err := http.NewRequestWithContext(ctx, http.MethodGet, u.String(), nil)
190 resp, err := http.DefaultClient.Do(req)
199 for i, url := range []string{
200 cluster.Services.Controller.ExternalURL.String(),
201 cluster.Services.Keepproxy.ExternalURL.String() + "d41d8cd98f00b204e9800998ecf8427e+0",
202 cluster.Services.WebDAVDownload.ExternalURL.String(),
204 diag.dotest(50+i, fmt.Sprintf("checking CORS headers at %s", url), func() error {
205 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
207 req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
211 req.Header.Set("Origin", "https://example.com")
212 resp, err := http.DefaultClient.Do(req)
216 if hdr := resp.Header.Get("Access-Control-Allow-Origin"); hdr != "*" {
217 return fmt.Errorf("expected \"Access-Control-Allow-Origin: *\", got %q", hdr)
223 var keeplist arvados.KeepServiceList
224 diag.dotest(60, "checking internal/external client detection", func() error {
225 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
227 err := client.RequestAndDecodeContext(ctx, &keeplist, "GET", "arvados/v1/keep_services/accessible", nil, arvados.ListOptions{Limit: 999999})
229 return fmt.Errorf("error getting keep services list: %s", err)
230 } else if len(keeplist.Items) == 0 {
231 return fmt.Errorf("controller did not return any keep services")
233 found := map[string]int{}
234 for _, ks := range keeplist.Items {
235 found[ks.ServiceType]++
237 isInternal := found["proxy"] == 0 && len(keeplist.Items) > 0
238 isExternal := found["proxy"] > 0 && found["proxy"] == len(keeplist.Items)
240 diag.debugf("controller returned only proxy services, this host is treated as \"external\"")
241 } else if isInternal {
242 diag.debugf("controller returned only non-proxy services, this host is treated as \"internal\"")
244 if (diag.checkInternal && !isInternal) || (diag.checkExternal && !isExternal) {
245 return fmt.Errorf("expecting internal=%v external=%v, but found internal=%v external=%v", diag.checkInternal, diag.checkExternal, isInternal, isExternal)
250 for i, ks := range keeplist.Items {
253 Host: net.JoinHostPort(ks.ServiceHost, fmt.Sprintf("%d", ks.ServicePort)),
256 if ks.ServiceSSLFlag {
259 diag.dotest(61+i, fmt.Sprintf("reading+writing via keep service at %s", u.String()), func() error {
260 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
262 req, err := http.NewRequestWithContext(ctx, "PUT", u.String()+"d41d8cd98f00b204e9800998ecf8427e", nil)
266 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
267 resp, err := http.DefaultClient.Do(req)
271 defer resp.Body.Close()
272 body, err := ioutil.ReadAll(resp.Body)
274 return fmt.Errorf("reading response body: %s", err)
276 loc := strings.TrimSpace(string(body))
277 if !strings.HasPrefix(loc, "d41d8") {
278 return fmt.Errorf("unexpected response from write: %q", body)
281 req, err = http.NewRequestWithContext(ctx, "GET", u.String()+loc, nil)
285 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
286 resp, err = http.DefaultClient.Do(req)
290 defer resp.Body.Close()
291 body, err = ioutil.ReadAll(resp.Body)
293 return fmt.Errorf("reading response body: %s", err)
296 return fmt.Errorf("unexpected response from read: %q", body)
303 var project arvados.Group
304 diag.dotest(80, fmt.Sprintf("finding/creating %q project", diag.projectName), func() error {
305 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
307 var grplist arvados.GroupList
308 err := client.RequestAndDecodeContext(ctx, &grplist, "GET", "arvados/v1/groups", nil, arvados.ListOptions{
309 Filters: []arvados.Filter{
310 {"name", "=", diag.projectName},
311 {"group_class", "=", "project"},
312 {"owner_uuid", "=", user.UUID}},
315 return fmt.Errorf("list groups: %s", err)
317 if len(grplist.Items) > 0 {
318 project = grplist.Items[0]
319 diag.debugf("using existing project, uuid = %s", project.UUID)
322 diag.debugf("list groups: ok, no results")
323 err = client.RequestAndDecodeContext(ctx, &project, "POST", "arvados/v1/groups", nil, map[string]interface{}{"group": map[string]interface{}{
324 "name": diag.projectName,
325 "group_class": "project",
328 return fmt.Errorf("create project: %s", err)
330 diag.debugf("created project, uuid = %s", project.UUID)
334 var collection arvados.Collection
335 diag.dotest(90, "creating temporary collection", func() error {
336 if project.UUID == "" {
337 return fmt.Errorf("skipping, no project to work in")
339 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
341 err := client.RequestAndDecodeContext(ctx, &collection, "POST", "arvados/v1/collections", nil, map[string]interface{}{
342 "ensure_unique_name": true,
343 "collection": map[string]interface{}{
344 "owner_uuid": project.UUID,
345 "name": "test collection",
346 "trash_at": time.Now().Add(time.Hour)}})
350 diag.debugf("ok, uuid = %s", collection.UUID)
354 if collection.UUID != "" {
356 diag.dotest(9990, "deleting temporary collection", func() error {
357 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
359 return client.RequestAndDecodeContext(ctx, nil, "DELETE", "arvados/v1/collections/"+collection.UUID, nil, nil)
364 diag.dotest(100, "uploading file via webdav", func() error {
365 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
367 if collection.UUID == "" {
368 return fmt.Errorf("skipping, no test collection")
370 req, err := http.NewRequestWithContext(ctx, "PUT", cluster.Services.WebDAVDownload.ExternalURL.String()+"c="+collection.UUID+"/testfile", bytes.NewBufferString("testfiledata"))
372 return fmt.Errorf("BUG? http.NewRequest: %s", err)
374 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
375 resp, err := http.DefaultClient.Do(req)
377 return fmt.Errorf("error performing http request: %s", err)
380 if resp.StatusCode != http.StatusCreated {
381 return fmt.Errorf("status %s", resp.Status)
383 diag.debugf("ok, status %s", resp.Status)
384 err = client.RequestAndDecodeContext(ctx, &collection, "GET", "arvados/v1/collections/"+collection.UUID, nil, nil)
386 return fmt.Errorf("get updated collection: %s", err)
388 diag.debugf("ok, pdh %s", collection.PortableDataHash)
392 davurl := cluster.Services.WebDAV.ExternalURL
393 davWildcard := strings.HasPrefix(davurl.Host, "*--") || strings.HasPrefix(davurl.Host, "*.")
394 diag.dotest(110, fmt.Sprintf("checking WebDAV ExternalURL wildcard (%s)", davurl), func() error {
395 if davurl.Host == "" {
396 return fmt.Errorf("host missing - content previews will not work")
398 if !davWildcard && !cluster.Collections.TrustAllContent {
399 diag.warnf("WebDAV ExternalURL has no leading wildcard and TrustAllContent==false - content previews will not work")
404 for i, trial := range []struct {
410 {false, false, http.StatusNotFound, strings.Replace(davurl.String(), "*", "d41d8cd98f00b204e9800998ecf8427e-0", 1) + "foo"},
411 {false, false, http.StatusNotFound, strings.Replace(davurl.String(), "*", "d41d8cd98f00b204e9800998ecf8427e-0", 1) + "testfile"},
412 {false, false, http.StatusNotFound, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=d41d8cd98f00b204e9800998ecf8427e+0/_/foo"},
413 {false, false, http.StatusNotFound, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=d41d8cd98f00b204e9800998ecf8427e+0/_/testfile"},
414 {true, true, http.StatusOK, strings.Replace(davurl.String(), "*", strings.Replace(collection.PortableDataHash, "+", "-", -1), 1) + "testfile"},
415 {true, false, http.StatusOK, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=" + collection.UUID + "/_/testfile"},
417 diag.dotest(120+i, fmt.Sprintf("downloading from webdav (%s)", trial.fileurl), func() error {
418 if trial.needWildcard && !davWildcard {
419 diag.warnf("skipping collection-id-in-vhost test because WebDAV ExternalURL has no leading wildcard")
422 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
424 if trial.needcoll && collection.UUID == "" {
425 return fmt.Errorf("skipping, no test collection")
427 req, err := http.NewRequestWithContext(ctx, "GET", trial.fileurl, nil)
431 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
432 resp, err := http.DefaultClient.Do(req)
436 defer resp.Body.Close()
437 body, err := ioutil.ReadAll(resp.Body)
439 return fmt.Errorf("reading response: %s", err)
441 if resp.StatusCode != trial.status {
442 return fmt.Errorf("unexpected response status: %s", resp.Status)
444 if trial.status == http.StatusOK && string(body) != "testfiledata" {
445 return fmt.Errorf("unexpected response content: %q", body)
451 var vm arvados.VirtualMachine
452 diag.dotest(130, "getting list of virtual machines", func() error {
453 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
455 var vmlist arvados.VirtualMachineList
456 err := client.RequestAndDecodeContext(ctx, &vmlist, "GET", "arvados/v1/virtual_machines", nil, arvados.ListOptions{Limit: 999999})
460 if len(vmlist.Items) < 1 {
461 diag.warnf("no VMs found")
468 diag.dotest(140, "getting workbench1 webshell page", func() error {
469 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
472 diag.warnf("skipping, no vm available")
475 webshelltermurl := cluster.Services.Workbench1.ExternalURL.String() + "virtual_machines/" + vm.UUID + "/webshell/testusername"
476 diag.debugf("url %s", webshelltermurl)
477 req, err := http.NewRequestWithContext(ctx, "GET", webshelltermurl, nil)
481 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
482 resp, err := http.DefaultClient.Do(req)
486 defer resp.Body.Close()
487 body, err := ioutil.ReadAll(resp.Body)
489 return fmt.Errorf("reading response: %s", err)
491 if resp.StatusCode != http.StatusOK {
492 return fmt.Errorf("unexpected response status: %s %q", resp.Status, body)
497 diag.dotest(150, "connecting to webshell service", func() error {
498 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
501 diag.warnf("skipping, no vm available")
504 u := cluster.Services.WebShell.ExternalURL
505 webshellurl := u.String() + vm.Hostname + "?"
506 if strings.HasPrefix(u.Host, "*") {
507 u.Host = vm.Hostname + u.Host[1:]
508 webshellurl = u.String() + "?"
510 diag.debugf("url %s", webshellurl)
511 req, err := http.NewRequestWithContext(ctx, "POST", webshellurl, bytes.NewBufferString(url.Values{
514 "session": {"xyzzy"},
515 "rooturl": {webshellurl},
520 req.Header.Set("Content-Type", "application/x-www-form-urlencoded; charset=UTF-8")
521 resp, err := http.DefaultClient.Do(req)
525 defer resp.Body.Close()
526 diag.debugf("response status %s", resp.Status)
527 body, err := ioutil.ReadAll(resp.Body)
529 return fmt.Errorf("reading response: %s", err)
531 diag.debugf("response body %q", body)
532 // We don't speak the protocol, so we get a 400 error
533 // from the webshell server even if everything is
534 // OK. Anything else (404, 502, ???) indicates a
536 if resp.StatusCode != http.StatusBadRequest {
537 return fmt.Errorf("unexpected response status: %s, %q", resp.Status, body)
542 diag.dotest(160, "running a container", func() error {
543 if diag.priority < 1 {
544 diag.infof("skipping (use priority > 0 if you want to run a container)")
547 if project.UUID == "" {
548 return fmt.Errorf("skipping, no project to work in")
551 var cr arvados.ContainerRequest
552 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
555 timestamp := time.Now().Format(time.RFC3339)
556 err := client.RequestAndDecodeContext(ctx, &cr, "POST", "arvados/v1/container_requests", nil, map[string]interface{}{"container_request": map[string]interface{}{
557 "owner_uuid": project.UUID,
558 "name": fmt.Sprintf("diagnostics container request %s", timestamp),
559 "container_image": diag.dockerImage,
560 "command": []string{"echo", timestamp},
561 "use_existing": false,
562 "output_path": "/mnt/output",
563 "output_name": fmt.Sprintf("diagnostics output %s", timestamp),
564 "priority": diag.priority,
565 "state": arvados.ContainerRequestStateCommitted,
566 "mounts": map[string]map[string]interface{}{
568 "kind": "collection",
572 "runtime_constraints": arvados.RuntimeConstraints{
575 KeepCacheRAM: 1 << 26,
581 diag.debugf("container request uuid = %s", cr.UUID)
582 diag.debugf("container uuid = %s", cr.ContainerUUID)
584 timeout := 10 * time.Minute
585 diag.infof("container request submitted, waiting up to %v for container to run", arvados.Duration(timeout))
586 ctx, cancel = context.WithDeadline(context.Background(), time.Now().Add(timeout))
589 var c arvados.Container
590 for ; cr.State != arvados.ContainerRequestStateFinal; time.Sleep(2 * time.Second) {
591 ctx, cancel := context.WithDeadline(ctx, time.Now().Add(diag.timeout))
594 crStateWas := cr.State
595 err := client.RequestAndDecodeContext(ctx, &cr, "GET", "arvados/v1/container_requests/"+cr.UUID, nil, nil)
599 if cr.State != crStateWas {
600 diag.debugf("container request state = %s", cr.State)
604 err = client.RequestAndDecodeContext(ctx, &c, "GET", "arvados/v1/containers/"+cr.ContainerUUID, nil, nil)
608 if c.State != cStateWas {
609 diag.debugf("container state = %s", c.State)
613 if c.State != arvados.ContainerStateComplete {
614 return fmt.Errorf("container request %s is final but container %s did not complete: container state = %q", cr.UUID, cr.ContainerUUID, c.State)
615 } else if c.ExitCode != 0 {
616 return fmt.Errorf("container exited %d", c.ExitCode)