1 // Copyright (C) The Arvados Authors. All rights reserved.
3 // SPDX-License-Identifier: AGPL-3.0
20 "git.arvados.org/arvados.git/sdk/go/arvados"
21 "git.arvados.org/arvados.git/sdk/go/ctxlog"
22 "github.com/sirupsen/logrus"
27 func (cmd Command) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
29 f := flag.NewFlagSet(prog, flag.ContinueOnError)
30 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")
31 f.StringVar(&diag.logLevel, "log-level", "info", "logging level (debug, info, warning, error)")
32 f.BoolVar(&diag.checkInternal, "internal-client", false, "check that this host is considered an \"internal\" client")
33 f.BoolVar(&diag.checkExternal, "external-client", false, "check that this host is considered an \"external\" client")
34 f.IntVar(&diag.priority, "priority", 500, "priority for test container (1..1000)")
35 f.DurationVar(&diag.timeout, "timeout", 10*time.Second, "timeout for http requests")
37 if err == flag.ErrHelp {
39 } else if err != nil {
40 fmt.Fprintln(stderr, err)
43 diag.logger = ctxlog.New(stdout, "text", diag.logLevel)
44 diag.logger.SetFormatter(&logrus.TextFormatter{DisableTimestamp: true, DisableLevelTruncation: true})
46 if len(diag.errors) == 0 {
47 diag.logger.Info("--- no errors ---")
50 if diag.logger.Level > logrus.ErrorLevel {
51 fmt.Fprint(stdout, "\n--- cut here --- error summary ---\n\n")
52 for _, e := range diag.errors {
60 type diagnoser struct {
74 func (diag *diagnoser) debugf(f string, args ...interface{}) {
75 diag.logger.Debugf(f, args...)
78 func (diag *diagnoser) infof(f string, args ...interface{}) {
79 diag.logger.Infof(f, args...)
82 func (diag *diagnoser) warnf(f string, args ...interface{}) {
83 diag.logger.Warnf(f, args...)
86 func (diag *diagnoser) errorf(f string, args ...interface{}) {
87 diag.logger.Errorf(f, args...)
88 diag.errors = append(diag.errors, fmt.Sprintf(f, args...))
91 // Run the given func, logging appropriate messages before and after,
92 // adding timing info, etc.
94 // The id argument should be unique among tests, and shouldn't change
95 // when other tests are added/removed.
96 func (diag *diagnoser) dotest(id int, title string, fn func() error) {
98 diag.done = map[int]bool{}
99 } else if diag.done[id] {
100 diag.errorf("(bug) reused test id %d", id)
104 diag.infof("%4d %s", id, title)
107 elapsed := fmt.Sprintf("%.0dms", time.Now().Sub(t0)/time.Millisecond)
109 diag.errorf("%s (%s): %s", title, elapsed, err)
111 diag.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 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
338 err := client.RequestAndDecodeContext(ctx, &collection, "POST", "arvados/v1/collections", nil, map[string]interface{}{
339 "ensure_unique_name": true,
340 "collection": map[string]interface{}{
341 "name": "test collection",
342 "trash_at": time.Now().Add(time.Hour)}})
346 diag.debugf("ok, uuid = %s", collection.UUID)
350 if collection.UUID != "" {
352 diag.dotest(9990, "deleting temporary collection", func() error {
353 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
355 return client.RequestAndDecodeContext(ctx, nil, "DELETE", "arvados/v1/collections/"+collection.UUID, nil, nil)
360 diag.dotest(100, "uploading file via webdav", func() error {
361 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
363 if collection.UUID == "" {
364 return fmt.Errorf("skipping, no test collection")
366 req, err := http.NewRequestWithContext(ctx, "PUT", cluster.Services.WebDAVDownload.ExternalURL.String()+"c="+collection.UUID+"/testfile", bytes.NewBufferString("testfiledata"))
368 return fmt.Errorf("BUG? http.NewRequest: %s", err)
370 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
371 resp, err := http.DefaultClient.Do(req)
373 return fmt.Errorf("error performing http request: %s", err)
376 if resp.StatusCode != http.StatusCreated {
377 return fmt.Errorf("status %s", resp.Status)
379 diag.debugf("ok, status %s", resp.Status)
380 err = client.RequestAndDecodeContext(ctx, &collection, "GET", "arvados/v1/collections/"+collection.UUID, nil, nil)
382 return fmt.Errorf("get updated collection: %s", err)
384 diag.debugf("ok, pdh %s", collection.PortableDataHash)
388 davurl := cluster.Services.WebDAV.ExternalURL
389 diag.dotest(110, fmt.Sprintf("checking WebDAV ExternalURL wildcard (%s)", davurl), func() error {
390 if davurl.Host == "" {
391 return fmt.Errorf("host missing - content previews will not work")
393 if !strings.HasPrefix(davurl.Host, "*--") && !strings.HasPrefix(davurl.Host, "*.") && !cluster.Collections.TrustAllContent {
394 diag.warnf("WebDAV ExternalURL has no leading wildcard and TrustAllContent==false - content previews will not work")
399 for i, trial := range []struct {
404 {false, http.StatusNotFound, strings.Replace(davurl.String(), "*", "d41d8cd98f00b204e9800998ecf8427e-0", 1) + "foo"},
405 {false, http.StatusNotFound, strings.Replace(davurl.String(), "*", "d41d8cd98f00b204e9800998ecf8427e-0", 1) + "testfile"},
406 {false, http.StatusNotFound, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=d41d8cd98f00b204e9800998ecf8427e+0/_/foo"},
407 {false, http.StatusNotFound, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=d41d8cd98f00b204e9800998ecf8427e+0/_/testfile"},
408 {true, http.StatusOK, strings.Replace(davurl.String(), "*", strings.Replace(collection.PortableDataHash, "+", "-", -1), 1) + "testfile"},
409 {true, http.StatusOK, cluster.Services.WebDAVDownload.ExternalURL.String() + "c=" + collection.UUID + "/_/testfile"},
411 diag.dotest(120+i, fmt.Sprintf("downloading from webdav (%s)", trial.fileurl), func() error {
412 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
414 if trial.needcoll && collection.UUID == "" {
415 return fmt.Errorf("skipping, no test collection")
417 req, err := http.NewRequestWithContext(ctx, "GET", trial.fileurl, nil)
421 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
422 resp, err := http.DefaultClient.Do(req)
426 defer resp.Body.Close()
427 body, err := ioutil.ReadAll(resp.Body)
429 return fmt.Errorf("reading response: %s", err)
431 if resp.StatusCode != trial.status {
432 return fmt.Errorf("unexpected response status: %s", resp.Status)
434 if trial.status == http.StatusOK && string(body) != "testfiledata" {
435 return fmt.Errorf("unexpected response content: %q", body)
441 var vm arvados.VirtualMachine
442 diag.dotest(130, "getting list of virtual machines", func() error {
443 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
445 var vmlist arvados.VirtualMachineList
446 err := client.RequestAndDecodeContext(ctx, &vmlist, "GET", "arvados/v1/virtual_machines", nil, arvados.ListOptions{Limit: 999999})
450 if len(vmlist.Items) < 1 {
451 return fmt.Errorf("no VMs found")
457 diag.dotest(140, "getting workbench1 webshell page", func() error {
458 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
461 return fmt.Errorf("skipping, no vm available")
463 webshelltermurl := cluster.Services.Workbench1.ExternalURL.String() + "virtual_machines/" + vm.UUID + "/webshell/testusername"
464 diag.debugf("url %s", webshelltermurl)
465 req, err := http.NewRequestWithContext(ctx, "GET", webshelltermurl, nil)
469 req.Header.Set("Authorization", "Bearer "+client.AuthToken)
470 resp, err := http.DefaultClient.Do(req)
474 defer resp.Body.Close()
475 body, err := ioutil.ReadAll(resp.Body)
477 return fmt.Errorf("reading response: %s", err)
479 if resp.StatusCode != http.StatusOK {
480 return fmt.Errorf("unexpected response status: %s %q", resp.Status, body)
485 diag.dotest(150, "connecting to webshell service", func() error {
486 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
489 return fmt.Errorf("skipping, no vm available")
491 u := cluster.Services.WebShell.ExternalURL
492 webshellurl := u.String() + vm.Hostname + "?"
493 if strings.HasPrefix(u.Host, "*") {
494 u.Host = vm.Hostname + u.Host[1:]
495 webshellurl = u.String() + "?"
497 diag.debugf("url %s", webshellurl)
498 req, err := http.NewRequestWithContext(ctx, "POST", webshellurl, bytes.NewBufferString(url.Values{
501 "session": {"xyzzy"},
502 "rooturl": {webshellurl},
507 req.Header.Set("Content-Type", "application/x-www-form-urlencoded; charset=UTF-8")
508 resp, err := http.DefaultClient.Do(req)
512 defer resp.Body.Close()
513 diag.debugf("response status %s", resp.Status)
514 body, err := ioutil.ReadAll(resp.Body)
516 return fmt.Errorf("reading response: %s", err)
518 diag.debugf("response body %q", body)
519 // We don't speak the protocol, so we get a 400 error
520 // from the webshell server even if everything is
521 // OK. Anything else (404, 502, ???) indicates a
523 if resp.StatusCode != http.StatusBadRequest {
524 return fmt.Errorf("unexpected response status: %s, %q", resp.Status, body)
529 diag.dotest(160, "running a container", func() error {
530 if diag.priority < 1 {
531 diag.debugf("skipping, caller requested priority<1 (%d)", diag.priority)
535 var cr arvados.ContainerRequest
536 ctx, cancel := context.WithDeadline(context.Background(), time.Now().Add(diag.timeout))
539 timestamp := time.Now().Format(time.RFC3339)
540 err := client.RequestAndDecodeContext(ctx, &cr, "POST", "arvados/v1/container_requests", nil, map[string]interface{}{"container_request": map[string]interface{}{
541 "owner_uuid": project.UUID,
542 "name": fmt.Sprintf("diagnostics container request %s", timestamp),
543 "container_image": "arvados/jobs",
544 "command": []string{"echo", timestamp},
545 "use_existing": false,
546 "output_path": "/mnt/output",
547 "output_name": fmt.Sprintf("diagnostics output %s", timestamp),
548 "priority": diag.priority,
549 "state": arvados.ContainerRequestStateCommitted,
550 "mounts": map[string]map[string]interface{}{
552 "kind": "collection",
556 "runtime_constraints": arvados.RuntimeConstraints{
559 KeepCacheRAM: 1 << 26,
565 diag.debugf("container request uuid = %s", cr.UUID)
566 diag.debugf("container uuid = %s", cr.ContainerUUID)
568 timeout := 10 * time.Minute
569 diag.infof("container request submitted, waiting up to %v for container to run", arvados.Duration(timeout))
570 ctx, cancel = context.WithDeadline(context.Background(), time.Now().Add(timeout))
573 var c arvados.Container
574 for ; cr.State != arvados.ContainerRequestStateFinal; time.Sleep(2 * time.Second) {
575 ctx, cancel := context.WithDeadline(ctx, time.Now().Add(diag.timeout))
578 crStateWas := cr.State
579 err := client.RequestAndDecodeContext(ctx, &cr, "GET", "arvados/v1/container_requests/"+cr.UUID, nil, nil)
583 if cr.State != crStateWas {
584 diag.debugf("container request state = %s", cr.State)
588 err = client.RequestAndDecodeContext(ctx, &c, "GET", "arvados/v1/containers/"+cr.ContainerUUID, nil, nil)
592 if c.State != cStateWas {
593 diag.debugf("container state = %s", c.State)
597 if c.State != arvados.ContainerStateComplete {
598 return fmt.Errorf("container request %s is final but container %s did not complete: container state = %q", cr.UUID, cr.ContainerUUID, c.State)
599 } else if c.ExitCode != 0 {
600 return fmt.Errorf("container exited %d", c.ExitCode)