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"
138 diag.dotest(20, fmt.Sprintf("getting exported config from https://%s/%s", client.APIHost, cfgpath), func() error {
139 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
141 err := client.RequestAndDecodeContext(ctx, &cluster, "GET", cfgpath, nil, nil)
145 diag.debugf("Collections.BlobSigning = %v", cluster.Collections.BlobSigning)
150 var user arvados.User
151 diag.dotest(30, "getting current user record", func() error {
152 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
154 err := client.RequestAndDecodeContext(ctx, &user, "GET", "arvados/v1/users/current", nil, nil)
158 diag.debugf("user uuid = %s", user.UUID)
163 diag.errorf("cannot proceed without cluster config -- aborting without running any further tests")
167 // uncomment to create some spurious errors
168 // cluster.Services.WebDAVDownload.ExternalURL.Host = "0.0.0.0:9"
170 // TODO: detect routing errors here, like finding wb2 at the
172 for i, svc := range []*arvados.Service{
173 &cluster.Services.Keepproxy,
174 &cluster.Services.WebDAV,
175 &cluster.Services.WebDAVDownload,
176 &cluster.Services.Websocket,
177 &cluster.Services.Workbench1,
178 &cluster.Services.Workbench2,
180 diag.dotest(40+i, fmt.Sprintf("connecting to service endpoint %s", svc.ExternalURL), func() error {
181 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
184 if strings.HasPrefix(u.Scheme, "ws") {
185 // We can do a real websocket test elsewhere,
186 // but for now we'll just check the https
188 u.Scheme = "http" + u.Scheme[2:]
190 if svc == &cluster.Services.WebDAV && strings.HasPrefix(u.Host, "*") {
191 u.Host = "d41d8cd98f00b204e9800998ecf8427e-0" + u.Host[1:]
193 req, err := http.NewRequestWithContext(ctx, http.MethodGet, u.String(), nil)
197 resp, err := http.DefaultClient.Do(req)
206 for i, url := range []string{
207 cluster.Services.Controller.ExternalURL.String(),
208 cluster.Services.Keepproxy.ExternalURL.String() + "d41d8cd98f00b204e9800998ecf8427e+0",
209 cluster.Services.WebDAVDownload.ExternalURL.String(),
211 diag.dotest(50+i, fmt.Sprintf("checking CORS headers at %s", url), func() error {
212 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
214 req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
218 req.Header.Set("Origin", "https://example.com")
219 resp, err := http.DefaultClient.Do(req)
223 if hdr := resp.Header.Get("Access-Control-Allow-Origin"); hdr != "*" {
224 return fmt.Errorf("expected \"Access-Control-Allow-Origin: *\", got %q", hdr)
230 var keeplist arvados.KeepServiceList
231 diag.dotest(60, "checking internal/external client detection", func() error {
232 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
234 err := client.RequestAndDecodeContext(ctx, &keeplist, "GET", "arvados/v1/keep_services/accessible", nil, arvados.ListOptions{Limit: 999999})
236 return fmt.Errorf("error getting keep services list: %s", err)
237 } else if len(keeplist.Items) == 0 {
238 return fmt.Errorf("controller did not return any keep services")
240 found := map[string]int{}
241 for _, ks := range keeplist.Items {
242 found[ks.ServiceType]++
244 isInternal := found["proxy"] == 0 && len(keeplist.Items) > 0
245 isExternal := found["proxy"] > 0 && found["proxy"] == len(keeplist.Items)
247 diag.debugf("controller returned only proxy services, this host is treated as \"external\"")
248 } else if isInternal {
249 diag.debugf("controller returned only non-proxy services, this host is treated as \"internal\"")
251 if (diag.checkInternal && !isInternal) || (diag.checkExternal && !isExternal) {
252 return fmt.Errorf("expecting internal=%v external=%v, but found internal=%v external=%v", diag.checkInternal, diag.checkExternal, isInternal, isExternal)
257 for i, ks := range keeplist.Items {
260 Host: net.JoinHostPort(ks.ServiceHost, fmt.Sprintf("%d", ks.ServicePort)),
263 if ks.ServiceSSLFlag {
266 diag.dotest(61+i, fmt.Sprintf("reading+writing via keep service at %s", u.String()), func() error {
267 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
269 req, err := http.NewRequestWithContext(ctx, "PUT", u.String()+"d41d8cd98f00b204e9800998ecf8427e", nil)
273 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
274 resp, err := http.DefaultClient.Do(req)
278 defer resp.Body.Close()
279 body, err := ioutil.ReadAll(resp.Body)
281 return fmt.Errorf("reading response body: %s", err)
283 loc := strings.TrimSpace(string(body))
284 if !strings.HasPrefix(loc, "d41d8") {
285 return fmt.Errorf("unexpected response from write: %q", body)
288 req, err = http.NewRequestWithContext(ctx, "GET", u.String()+loc, nil)
292 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
293 resp, err = http.DefaultClient.Do(req)
297 defer resp.Body.Close()
298 body, err = ioutil.ReadAll(resp.Body)
300 return fmt.Errorf("reading response body: %s", err)
303 return fmt.Errorf("unexpected response from read: %q", body)
310 var project arvados.Group
311 diag.dotest(80, fmt.Sprintf("finding/creating %q project", diag.projectName), func() error {
312 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
314 var grplist arvados.GroupList
315 err := client.RequestAndDecodeContext(ctx, &grplist, "GET", "arvados/v1/groups", nil, arvados.ListOptions{
316 Filters: []arvados.Filter{
317 {"name", "=", diag.projectName},
318 {"group_class", "=", "project"},
319 {"owner_uuid", "=", user.UUID}},
322 return fmt.Errorf("list groups: %s", err)
324 if len(grplist.Items) > 0 {
325 project = grplist.Items[0]
326 diag.debugf("using existing project, uuid = %s", project.UUID)
329 diag.debugf("list groups: ok, no results")
330 err = client.RequestAndDecodeContext(ctx, &project, "POST", "arvados/v1/groups", nil, map[string]interface{}{"group": map[string]interface{}{
331 "name": diag.projectName,
332 "group_class": "project",
335 return fmt.Errorf("create project: %s", err)
337 diag.debugf("created project, uuid = %s", project.UUID)
341 var collection arvados.Collection
342 diag.dotest(90, "creating temporary collection", func() error {
343 if project.UUID == "" {
344 return fmt.Errorf("skipping, no project to work in")
346 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
348 err := client.RequestAndDecodeContext(ctx, &collection, "POST", "arvados/v1/collections", nil, map[string]interface{}{
349 "ensure_unique_name": true,
350 "collection": map[string]interface{}{
351 "owner_uuid": project.UUID,
352 "name": "test collection",
353 "trash_at": time.Now().Add(time.Hour)}})
357 diag.debugf("ok, uuid = %s", collection.UUID)
361 if collection.UUID != "" {
363 diag.dotest(9990, "deleting temporary collection", func() error {
364 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
366 return client.RequestAndDecodeContext(ctx, nil, "DELETE", "arvados/v1/collections/"+collection.UUID, nil, nil)
371 diag.dotest(100, "uploading file via webdav", func() error {
372 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
374 if collection.UUID == "" {
375 return fmt.Errorf("skipping, no test collection")
377 req, err := http.NewRequestWithContext(ctx, "PUT", cluster.Services.WebDAVDownload.ExternalURL.String()+"c="+collection.UUID+"/testfile", bytes.NewBufferString("testfiledata"))
379 return fmt.Errorf("BUG? http.NewRequest: %s", err)
381 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
382 resp, err := http.DefaultClient.Do(req)
384 return fmt.Errorf("error performing http request: %s", err)
387 if resp.StatusCode != http.StatusCreated {
388 return fmt.Errorf("status %s", resp.Status)
390 diag.debugf("ok, status %s", resp.Status)
391 err = client.RequestAndDecodeContext(ctx, &collection, "GET", "arvados/v1/collections/"+collection.UUID, nil, nil)
393 return fmt.Errorf("get updated collection: %s", err)
395 diag.debugf("ok, pdh %s", collection.PortableDataHash)
399 davurl := cluster.Services.WebDAV.ExternalURL
400 davWildcard := strings.HasPrefix(davurl.Host, "*--") || strings.HasPrefix(davurl.Host, "*.")
401 diag.dotest(110, fmt.Sprintf("checking WebDAV ExternalURL wildcard (%s)", davurl), func() error {
402 if davurl.Host == "" {
403 return fmt.Errorf("host missing - content previews will not work")
405 if !davWildcard && !cluster.Collections.TrustAllContent {
406 diag.warnf("WebDAV ExternalURL has no leading wildcard and TrustAllContent==false - content previews will not work")
411 for i, trial := range []struct {
417 {false, false, http.StatusNotFound, strings.Replace(davurl.String(), "*", "d41d8cd98f00b204e9800998ecf8427e-0", 1) + "foo"},
418 {false, false, http.StatusNotFound, strings.Replace(davurl.String(), "*", "d41d8cd98f00b204e9800998ecf8427e-0", 1) + "testfile"},
419 {false, false, http.StatusNotFound, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=d41d8cd98f00b204e9800998ecf8427e+0/_/foo"},
420 {false, false, http.StatusNotFound, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=d41d8cd98f00b204e9800998ecf8427e+0/_/testfile"},
421 {true, true, http.StatusOK, strings.Replace(davurl.String(), "*", strings.Replace(collection.PortableDataHash, "+", "-", -1), 1) + "testfile"},
422 {true, false, http.StatusOK, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=" + collection.UUID + "/_/testfile"},
424 diag.dotest(120+i, fmt.Sprintf("downloading from webdav (%s)", trial.fileurl), func() error {
425 if trial.needWildcard && !davWildcard {
426 diag.warnf("skipping collection-id-in-vhost test because WebDAV ExternalURL has no leading wildcard")
429 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
431 if trial.needcoll && collection.UUID == "" {
432 return fmt.Errorf("skipping, no test collection")
434 req, err := http.NewRequestWithContext(ctx, "GET", trial.fileurl, nil)
438 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
439 resp, err := http.DefaultClient.Do(req)
443 defer resp.Body.Close()
444 body, err := ioutil.ReadAll(resp.Body)
446 return fmt.Errorf("reading response: %s", err)
448 if resp.StatusCode != trial.status {
449 return fmt.Errorf("unexpected response status: %s", resp.Status)
451 if trial.status == http.StatusOK && string(body) != "testfiledata" {
452 return fmt.Errorf("unexpected response content: %q", body)
458 var vm arvados.VirtualMachine
459 diag.dotest(130, "getting list of virtual machines", func() error {
460 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
462 var vmlist arvados.VirtualMachineList
463 err := client.RequestAndDecodeContext(ctx, &vmlist, "GET", "arvados/v1/virtual_machines", nil, arvados.ListOptions{Limit: 999999})
467 if len(vmlist.Items) < 1 {
468 diag.warnf("no VMs found")
475 diag.dotest(140, "getting workbench1 webshell page", func() error {
476 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
479 diag.warnf("skipping, no vm available")
482 webshelltermurl := cluster.Services.Workbench1.ExternalURL.String() + "virtual_machines/" + vm.UUID + "/webshell/testusername"
483 diag.debugf("url %s", webshelltermurl)
484 req, err := http.NewRequestWithContext(ctx, "GET", webshelltermurl, nil)
488 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
489 resp, err := http.DefaultClient.Do(req)
493 defer resp.Body.Close()
494 body, err := ioutil.ReadAll(resp.Body)
496 return fmt.Errorf("reading response: %s", err)
498 if resp.StatusCode != http.StatusOK {
499 return fmt.Errorf("unexpected response status: %s %q", resp.Status, body)
504 diag.dotest(150, "connecting to webshell service", func() error {
505 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
508 diag.warnf("skipping, no vm available")
511 u := cluster.Services.WebShell.ExternalURL
512 webshellurl := u.String() + vm.Hostname + "?"
513 if strings.HasPrefix(u.Host, "*") {
514 u.Host = vm.Hostname + u.Host[1:]
515 webshellurl = u.String() + "?"
517 diag.debugf("url %s", webshellurl)
518 req, err := http.NewRequestWithContext(ctx, "POST", webshellurl, bytes.NewBufferString(url.Values{
521 "session": {"xyzzy"},
522 "rooturl": {webshellurl},
527 req.Header.Set("Content-Type", "application/x-www-form-urlencoded; charset=UTF-8")
528 resp, err := http.DefaultClient.Do(req)
532 defer resp.Body.Close()
533 diag.debugf("response status %s", resp.Status)
534 body, err := ioutil.ReadAll(resp.Body)
536 return fmt.Errorf("reading response: %s", err)
538 diag.debugf("response body %q", body)
539 // We don't speak the protocol, so we get a 400 error
540 // from the webshell server even if everything is
541 // OK. Anything else (404, 502, ???) indicates a
543 if resp.StatusCode != http.StatusBadRequest {
544 return fmt.Errorf("unexpected response status: %s, %q", resp.Status, body)
549 diag.dotest(160, "running a container", func() error {
550 if diag.priority < 1 {
551 diag.infof("skipping (use priority > 0 if you want to run a container)")
554 if project.UUID == "" {
555 return fmt.Errorf("skipping, no project to work in")
558 var cr arvados.ContainerRequest
559 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
562 timestamp := time.Now().Format(time.RFC3339)
563 err := client.RequestAndDecodeContext(ctx, &cr, "POST", "arvados/v1/container_requests", nil, map[string]interface{}{"container_request": map[string]interface{}{
564 "owner_uuid": project.UUID,
565 "name": fmt.Sprintf("diagnostics container request %s", timestamp),
566 "container_image": diag.dockerImage,
567 "command": []string{"echo", timestamp},
568 "use_existing": false,
569 "output_path": "/mnt/output",
570 "output_name": fmt.Sprintf("diagnostics output %s", timestamp),
571 "priority": diag.priority,
572 "state": arvados.ContainerRequestStateCommitted,
573 "mounts": map[string]map[string]interface{}{
575 "kind": "collection",
579 "runtime_constraints": arvados.RuntimeConstraints{
582 KeepCacheRAM: 1 << 26,
588 diag.debugf("container request uuid = %s", cr.UUID)
589 diag.debugf("container uuid = %s", cr.ContainerUUID)
591 timeout := 10 * time.Minute
592 diag.infof("container request submitted, waiting up to %v for container to run", arvados.Duration(timeout))
593 ctx, cancel = context.WithDeadline(context.Background(), time.Now().Add(timeout))
596 var c arvados.Container
597 for ; cr.State != arvados.ContainerRequestStateFinal; time.Sleep(2 * time.Second) {
598 ctx, cancel := context.WithDeadline(ctx, time.Now().Add(diag.timeout))
601 crStateWas := cr.State
602 err := client.RequestAndDecodeContext(ctx, &cr, "GET", "arvados/v1/container_requests/"+cr.UUID, nil, nil)
606 if cr.State != crStateWas {
607 diag.debugf("container request state = %s", cr.State)
611 err = client.RequestAndDecodeContext(ctx, &c, "GET", "arvados/v1/containers/"+cr.ContainerUUID, nil, nil)
615 if c.State != cStateWas {
616 diag.debugf("container state = %s", c.State)
620 if c.State != arvados.ContainerStateComplete {
621 return fmt.Errorf("container request %s is final but container %s did not complete: container state = %q", cr.UUID, cr.ContainerUUID, c.State)
622 } else if c.ExitCode != 0 {
623 return fmt.Errorf("container exited %d", c.ExitCode)