From 23ddce7f83a4ab2e39b5910766f54aafb7b5a99d Mon Sep 17 00:00:00 2001 From: Peter Amstutz Date: Thu, 18 Oct 2018 17:17:33 -0400 Subject: [PATCH] 14262: Refactoring, split up federation code into smaller files Arvados-DCO-1.1-Signed-off-by: Peter Amstutz --- lib/controller/fed_collections.go | 312 +++++++++++++++ lib/controller/fed_generic.go | 331 ++++++++++++++++ lib/controller/federation.go | 609 ------------------------------ 3 files changed, 643 insertions(+), 609 deletions(-) create mode 100644 lib/controller/fed_collections.go create mode 100644 lib/controller/fed_generic.go diff --git a/lib/controller/fed_collections.go b/lib/controller/fed_collections.go new file mode 100644 index 0000000000..62f98367cc --- /dev/null +++ b/lib/controller/fed_collections.go @@ -0,0 +1,312 @@ +// Copyright (C) The Arvados Authors. All rights reserved. +// +// SPDX-License-Identifier: AGPL-3.0 + +package controller + +import ( + "bufio" + "bytes" + "context" + "crypto/md5" + "encoding/json" + "fmt" + "io" + "io/ioutil" + "net/http" + "strings" + "sync" + + "git.curoverse.com/arvados.git/sdk/go/arvados" + "git.curoverse.com/arvados.git/sdk/go/httpserver" + "git.curoverse.com/arvados.git/sdk/go/keepclient" +) + +type collectionFederatedRequestHandler struct { + next http.Handler + handler *Handler +} + +func rewriteSignatures(clusterID string, expectHash string, + resp *http.Response, requestError error) (newResponse *http.Response, err error) { + + if requestError != nil { + return resp, requestError + } + + if resp.StatusCode != 200 { + return resp, nil + } + + originalBody := resp.Body + defer originalBody.Close() + + var col arvados.Collection + err = json.NewDecoder(resp.Body).Decode(&col) + if err != nil { + return nil, err + } + + // rewriting signatures will make manifest text 5-10% bigger so calculate + // capacity accordingly + updatedManifest := bytes.NewBuffer(make([]byte, 0, int(float64(len(col.ManifestText))*1.1))) + + hasher := md5.New() + mw := io.MultiWriter(hasher, updatedManifest) + sz := 0 + + scanner := bufio.NewScanner(strings.NewReader(col.ManifestText)) + scanner.Buffer(make([]byte, 1048576), len(col.ManifestText)) + for scanner.Scan() { + line := scanner.Text() + tokens := strings.Split(line, " ") + if len(tokens) < 3 { + return nil, fmt.Errorf("Invalid stream (<3 tokens): %q", line) + } + + n, err := mw.Write([]byte(tokens[0])) + if err != nil { + return nil, fmt.Errorf("Error updating manifest: %v", err) + } + sz += n + for _, token := range tokens[1:] { + n, err = mw.Write([]byte(" ")) + if err != nil { + return nil, fmt.Errorf("Error updating manifest: %v", err) + } + sz += n + + m := keepclient.SignedLocatorRe.FindStringSubmatch(token) + if m != nil { + // Rewrite the block signature to be a remote signature + _, err = fmt.Fprintf(updatedManifest, "%s%s%s+R%s-%s%s", m[1], m[2], m[3], clusterID, m[5][2:], m[8]) + if err != nil { + return nil, fmt.Errorf("Error updating manifest: %v", err) + } + + // for hash checking, ignore signatures + n, err = fmt.Fprintf(hasher, "%s%s", m[1], m[2]) + if err != nil { + return nil, fmt.Errorf("Error updating manifest: %v", err) + } + sz += n + } else { + n, err = mw.Write([]byte(token)) + if err != nil { + return nil, fmt.Errorf("Error updating manifest: %v", err) + } + sz += n + } + } + n, err = mw.Write([]byte("\n")) + if err != nil { + return nil, fmt.Errorf("Error updating manifest: %v", err) + } + sz += n + } + + // Check that expected hash is consistent with + // portable_data_hash field of the returned record + if expectHash == "" { + expectHash = col.PortableDataHash + } else if expectHash != col.PortableDataHash { + return nil, fmt.Errorf("portable_data_hash %q on returned record did not match expected hash %q ", expectHash, col.PortableDataHash) + } + + // Certify that the computed hash of the manifest_text matches our expectation + sum := hasher.Sum(nil) + computedHash := fmt.Sprintf("%x+%v", sum, sz) + if computedHash != expectHash { + return nil, fmt.Errorf("Computed manifest_text hash %q did not match expected hash %q", computedHash, expectHash) + } + + col.ManifestText = updatedManifest.String() + + newbody, err := json.Marshal(col) + if err != nil { + return nil, err + } + + buf := bytes.NewBuffer(newbody) + resp.Body = ioutil.NopCloser(buf) + resp.ContentLength = int64(buf.Len()) + resp.Header.Set("Content-Length", fmt.Sprintf("%v", buf.Len())) + + return resp, nil +} + +func filterLocalClusterResponse(resp *http.Response, requestError error) (newResponse *http.Response, err error) { + if requestError != nil { + return resp, requestError + } + + if resp.StatusCode == 404 { + // Suppress returning this result, because we want to + // search the federation. + return nil, nil + } + return resp, nil +} + +type searchRemoteClusterForPDH struct { + pdh string + remoteID string + mtx *sync.Mutex + sentResponse *bool + sharedContext *context.Context + cancelFunc func() + errors *[]string + statusCode *int +} + +func (s *searchRemoteClusterForPDH) filterRemoteClusterResponse(resp *http.Response, requestError error) (newResponse *http.Response, err error) { + s.mtx.Lock() + defer s.mtx.Unlock() + + if *s.sentResponse { + // Another request already returned a response + return nil, nil + } + + if requestError != nil { + *s.errors = append(*s.errors, fmt.Sprintf("Request error contacting %q: %v", s.remoteID, requestError)) + // Record the error and suppress response + return nil, nil + } + + if resp.StatusCode != 200 { + // Suppress returning unsuccessful result. Maybe + // another request will find it. + // TODO collect and return error responses. + *s.errors = append(*s.errors, fmt.Sprintf("Response from %q: %v", s.remoteID, resp.Status)) + if resp.StatusCode != 404 { + // Got a non-404 error response, convert into BadGateway + *s.statusCode = http.StatusBadGateway + } + return nil, nil + } + + s.mtx.Unlock() + + // This reads the response body. We don't want to hold the + // lock while doing this because other remote requests could + // also have made it to this point, and we don't want a + // slow response holding the lock to block a faster response + // that is waiting on the lock. + newResponse, err = rewriteSignatures(s.remoteID, s.pdh, resp, nil) + + s.mtx.Lock() + + if *s.sentResponse { + // Another request already returned a response + return nil, nil + } + + if err != nil { + // Suppress returning unsuccessful result. Maybe + // another request will be successful. + *s.errors = append(*s.errors, fmt.Sprintf("Error parsing response from %q: %v", s.remoteID, err)) + return nil, nil + } + + // We have a successful response. Suppress/cancel all the + // other requests/responses. + *s.sentResponse = true + s.cancelFunc() + + return newResponse, nil +} + +func (h *collectionFederatedRequestHandler) ServeHTTP(w http.ResponseWriter, req *http.Request) { + if req.Method != "GET" { + // Only handle GET requests right now + h.next.ServeHTTP(w, req) + return + } + + m := collectionByPDHRe.FindStringSubmatch(req.URL.Path) + if len(m) != 2 { + // Not a collection PDH GET request + m = collectionRe.FindStringSubmatch(req.URL.Path) + clusterId := "" + + if len(m) > 0 { + clusterId = m[2] + } + + if clusterId != "" && clusterId != h.handler.Cluster.ClusterID { + // request for remote collection by uuid + resp, err := h.handler.remoteClusterRequest(clusterId, req) + newResponse, err := rewriteSignatures(clusterId, "", resp, err) + h.handler.proxy.ForwardResponse(w, newResponse, err) + return + } + // not a collection UUID request, or it is a request + // for a local UUID, either way, continue down the + // handler stack. + h.next.ServeHTTP(w, req) + return + } + + // Request for collection by PDH. Search the federation. + + // First, query the local cluster. + resp, err := h.handler.localClusterRequest(req) + newResp, err := filterLocalClusterResponse(resp, err) + if newResp != nil || err != nil { + h.handler.proxy.ForwardResponse(w, newResp, err) + return + } + + sharedContext, cancelFunc := context.WithCancel(req.Context()) + defer cancelFunc() + req = req.WithContext(sharedContext) + + // Create a goroutine for each cluster in the + // RemoteClusters map. The first valid result gets + // returned to the client. When that happens, all + // other outstanding requests are cancelled or + // suppressed. + sentResponse := false + mtx := sync.Mutex{} + wg := sync.WaitGroup{} + var errors []string + var errorCode int = 404 + + // use channel as a semaphore to limit the number of concurrent + // requests at a time + sem := make(chan bool, h.handler.Cluster.RequestLimits.GetMultiClusterRequestConcurrency()) + defer close(sem) + for remoteID := range h.handler.Cluster.RemoteClusters { + if remoteID == h.handler.Cluster.ClusterID { + // No need to query local cluster again + continue + } + // blocks until it can put a value into the + // channel (which has a max queue capacity) + sem <- true + if sentResponse { + break + } + search := &searchRemoteClusterForPDH{m[1], remoteID, &mtx, &sentResponse, + &sharedContext, cancelFunc, &errors, &errorCode} + wg.Add(1) + go func() { + resp, err := h.handler.remoteClusterRequest(search.remoteID, req) + newResp, err := search.filterRemoteClusterResponse(resp, err) + if newResp != nil || err != nil { + h.handler.proxy.ForwardResponse(w, newResp, err) + } + wg.Done() + <-sem + }() + } + wg.Wait() + + if sentResponse { + return + } + + // No successful responses, so return the error + httpserver.Errors(w, errors, errorCode) +} diff --git a/lib/controller/fed_generic.go b/lib/controller/fed_generic.go new file mode 100644 index 0000000000..0630217b6e --- /dev/null +++ b/lib/controller/fed_generic.go @@ -0,0 +1,331 @@ +// Copyright (C) The Arvados Authors. All rights reserved. +// +// SPDX-License-Identifier: AGPL-3.0 + +package controller + +import ( + "bytes" + "encoding/json" + "fmt" + "io/ioutil" + "net/http" + "net/url" + "regexp" + "sync" + + "git.curoverse.com/arvados.git/sdk/go/httpserver" +) + +type genericFederatedRequestHandler struct { + next http.Handler + handler *Handler + matcher *regexp.Regexp +} + +func (h *genericFederatedRequestHandler) remoteQueryUUIDs(w http.ResponseWriter, + req *http.Request, + clusterID string, uuids []string) (rp []map[string]interface{}, kind string, err error) { + + found := make(map[string]bool) + prev_len_uuids := len(uuids) + 1 + // Loop while + // (1) there are more uuids to query + // (2) we're making progress - on each iteration the set of + // uuids we are expecting for must shrink. + for len(uuids) > 0 && len(uuids) < prev_len_uuids { + var remoteReq http.Request + remoteReq.Header = req.Header + remoteReq.Method = "POST" + remoteReq.URL = &url.URL{Path: req.URL.Path} + remoteParams := make(url.Values) + remoteParams.Set("_method", "GET") + remoteParams.Set("count", "none") + if req.Form.Get("select") != "" { + remoteParams.Set("select", req.Form.Get("select")) + } + content, err := json.Marshal(uuids) + if err != nil { + return nil, "", err + } + remoteParams["filters"] = []string{fmt.Sprintf(`[["uuid", "in", %s]]`, content)} + enc := remoteParams.Encode() + remoteReq.Body = ioutil.NopCloser(bytes.NewBufferString(enc)) + + rc := multiClusterQueryResponseCollector{clusterID: clusterID} + + var resp *http.Response + if clusterID == h.handler.Cluster.ClusterID { + resp, err = h.handler.localClusterRequest(&remoteReq) + } else { + resp, err = h.handler.remoteClusterRequest(clusterID, &remoteReq) + } + rc.collectResponse(resp, err) + + if rc.error != nil { + return nil, "", rc.error + } + + kind = rc.kind + + if len(rc.responses) == 0 { + // We got zero responses, no point in doing + // another query. + return rp, kind, nil + } + + rp = append(rp, rc.responses...) + + // Go through the responses and determine what was + // returned. If there are remaining items, loop + // around and do another request with just the + // stragglers. + for _, i := range rc.responses { + uuid, ok := i["uuid"].(string) + if ok { + found[uuid] = true + } + } + + l := []string{} + for _, u := range uuids { + if !found[u] { + l = append(l, u) + } + } + prev_len_uuids = len(uuids) + uuids = l + } + + return rp, kind, nil +} + +func (h *genericFederatedRequestHandler) handleMultiClusterQuery(w http.ResponseWriter, + req *http.Request, clusterId *string) bool { + + var filters [][]interface{} + err := json.Unmarshal([]byte(req.Form.Get("filters")), &filters) + if err != nil { + httpserver.Error(w, err.Error(), http.StatusBadRequest) + return true + } + + // Split the list of uuids by prefix + queryClusters := make(map[string][]string) + expectCount := 0 + for _, filter := range filters { + if len(filter) != 3 { + return false + } + + if lhs, ok := filter[0].(string); !ok || lhs != "uuid" { + return false + } + + op, ok := filter[1].(string) + if !ok { + return false + } + + if op == "in" { + if rhs, ok := filter[2].([]interface{}); ok { + for _, i := range rhs { + if u, ok := i.(string); ok { + *clusterId = u[0:5] + queryClusters[u[0:5]] = append(queryClusters[u[0:5]], u) + expectCount += 1 + } + } + } + } else if op == "=" { + if u, ok := filter[2].(string); ok { + *clusterId = u[0:5] + queryClusters[u[0:5]] = append(queryClusters[u[0:5]], u) + expectCount += 1 + } + } else { + return false + } + + } + + if len(queryClusters) <= 1 { + // Query does not search for uuids across multiple + // clusters. + return false + } + + // Validations + count := req.Form.Get("count") + if count != "" && count != `none` && count != `"none"` { + httpserver.Error(w, "Federated multi-object query must have 'count=none'", http.StatusBadRequest) + return true + } + if req.Form.Get("limit") != "" || req.Form.Get("offset") != "" || req.Form.Get("order") != "" { + httpserver.Error(w, "Federated multi-object may not provide 'limit', 'offset' or 'order'.", http.StatusBadRequest) + return true + } + if expectCount > h.handler.Cluster.RequestLimits.GetMaxItemsPerResponse() { + httpserver.Error(w, fmt.Sprintf("Federated multi-object request for %v objects which is more than max page size %v.", + expectCount, h.handler.Cluster.RequestLimits.GetMaxItemsPerResponse()), http.StatusBadRequest) + return true + } + if req.Form.Get("select") != "" { + foundUUID := false + var selects []string + err := json.Unmarshal([]byte(req.Form.Get("select")), &selects) + if err != nil { + httpserver.Error(w, err.Error(), http.StatusBadRequest) + return true + } + + for _, r := range selects { + if r == "uuid" { + foundUUID = true + break + } + } + if !foundUUID { + httpserver.Error(w, "Federated multi-object request must include 'uuid' in 'select'", http.StatusBadRequest) + return true + } + } + + // Perform concurrent requests to each cluster + + // use channel as a semaphore to limit the number of concurrent + // requests at a time + sem := make(chan bool, h.handler.Cluster.RequestLimits.GetMultiClusterRequestConcurrency()) + defer close(sem) + wg := sync.WaitGroup{} + + req.Header.Set("Content-Type", "application/x-www-form-urlencoded") + mtx := sync.Mutex{} + errors := []error{} + var completeResponses []map[string]interface{} + var kind string + + for k, v := range queryClusters { + if len(v) == 0 { + // Nothing to query + continue + } + + // blocks until it can put a value into the + // channel (which has a max queue capacity) + sem <- true + wg.Add(1) + go func(k string, v []string) { + rp, kn, err := h.remoteQueryUUIDs(w, req, k, v) + mtx.Lock() + if err == nil { + completeResponses = append(completeResponses, rp...) + kind = kn + } else { + errors = append(errors, err) + } + mtx.Unlock() + wg.Done() + <-sem + }(k, v) + } + wg.Wait() + + if len(errors) > 0 { + var strerr []string + for _, e := range errors { + strerr = append(strerr, e.Error()) + } + httpserver.Errors(w, strerr, http.StatusBadGateway) + return true + } + + w.Header().Set("Content-Type", "application/json") + w.WriteHeader(http.StatusOK) + itemList := make(map[string]interface{}) + itemList["items"] = completeResponses + itemList["kind"] = kind + json.NewEncoder(w).Encode(itemList) + + return true +} + +func (h *genericFederatedRequestHandler) ServeHTTP(w http.ResponseWriter, req *http.Request) { + m := h.matcher.FindStringSubmatch(req.URL.Path) + clusterId := "" + + if len(m) > 0 && m[2] != "" { + clusterId = m[2] + } + + // Get form parameters from URL and form body (if POST). + if err := loadParamsFromForm(req); err != nil { + httpserver.Error(w, err.Error(), http.StatusBadRequest) + return + } + + // Check if the parameters have an explicit cluster_id + if req.Form.Get("cluster_id") != "" { + clusterId = req.Form.Get("cluster_id") + } + + // Handle the POST-as-GET special case (workaround for large + // GET requests that potentially exceed maximum URL length, + // like multi-object queries where the filter has 100s of + // items) + effectiveMethod := req.Method + if req.Method == "POST" && req.Form.Get("_method") != "" { + effectiveMethod = req.Form.Get("_method") + } + + if effectiveMethod == "GET" && + clusterId == "" && + req.Form.Get("filters") != "" && + h.handleMultiClusterQuery(w, req, &clusterId) { + return + } + + if clusterId == "" || clusterId == h.handler.Cluster.ClusterID { + h.next.ServeHTTP(w, req) + } else { + resp, err := h.handler.remoteClusterRequest(clusterId, req) + h.handler.proxy.ForwardResponse(w, resp, err) + } +} + +type multiClusterQueryResponseCollector struct { + responses []map[string]interface{} + error error + kind string + clusterID string +} + +func (c *multiClusterQueryResponseCollector) collectResponse(resp *http.Response, + requestError error) (newResponse *http.Response, err error) { + if requestError != nil { + c.error = requestError + return nil, nil + } + + defer resp.Body.Close() + var loadInto struct { + Kind string `json:"kind"` + Items []map[string]interface{} `json:"items"` + Errors []string `json:"errors"` + } + err = json.NewDecoder(resp.Body).Decode(&loadInto) + + if err != nil { + c.error = fmt.Errorf("error fetching from %v (%v): %v", c.clusterID, resp.Status, err) + return nil, nil + } + if resp.StatusCode != http.StatusOK { + c.error = fmt.Errorf("error fetching from %v (%v): %v", c.clusterID, resp.Status, loadInto.Errors) + return nil, nil + } + + c.responses = loadInto.Items + c.kind = loadInto.Kind + + return nil, nil +} diff --git a/lib/controller/federation.go b/lib/controller/federation.go index c5089fa235..03d2f3fabb 100644 --- a/lib/controller/federation.go +++ b/lib/controller/federation.go @@ -5,12 +5,8 @@ package controller import ( - "bufio" "bytes" - "context" - "crypto/md5" "database/sql" - "encoding/json" "fmt" "io" "io/ioutil" @@ -18,12 +14,9 @@ import ( "net/url" "regexp" "strings" - "sync" "git.curoverse.com/arvados.git/sdk/go/arvados" "git.curoverse.com/arvados.git/sdk/go/auth" - "git.curoverse.com/arvados.git/sdk/go/httpserver" - "git.curoverse.com/arvados.git/sdk/go/keepclient" ) var pathPattern = `^/arvados/v1/%s(/([0-9a-z]{5})-%s-[0-9a-z]{15})?(.*)$` @@ -33,17 +26,6 @@ var containerRequestsRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "container var collectionRe = regexp.MustCompile(fmt.Sprintf(pathPattern, "collections", "4zz18")) var collectionByPDHRe = regexp.MustCompile(`^/arvados/v1/collections/([0-9a-fA-F]{32}\+[0-9]+)+$`) -type genericFederatedRequestHandler struct { - next http.Handler - handler *Handler - matcher *regexp.Regexp -} - -type collectionFederatedRequestHandler struct { - next http.Handler - handler *Handler -} - func (h *Handler) remoteClusterRequest(remoteID string, req *http.Request) (*http.Response, error) { remote, ok := h.Cluster.RemoteClusters[remoteID] if !ok { @@ -98,597 +80,6 @@ func loadParamsFromForm(req *http.Request) error { return nil } -type multiClusterQueryResponseCollector struct { - responses []map[string]interface{} - error error - kind string - clusterID string -} - -func (c *multiClusterQueryResponseCollector) collectResponse(resp *http.Response, - requestError error) (newResponse *http.Response, err error) { - if requestError != nil { - c.error = requestError - return nil, nil - } - - defer resp.Body.Close() - var loadInto struct { - Kind string `json:"kind"` - Items []map[string]interface{} `json:"items"` - Errors []string `json:"errors"` - } - err = json.NewDecoder(resp.Body).Decode(&loadInto) - - if err != nil { - c.error = fmt.Errorf("error fetching from %v (%v): %v", c.clusterID, resp.Status, err) - return nil, nil - } - if resp.StatusCode != http.StatusOK { - c.error = fmt.Errorf("error fetching from %v (%v): %v", c.clusterID, resp.Status, loadInto.Errors) - return nil, nil - } - - c.responses = loadInto.Items - c.kind = loadInto.Kind - - return nil, nil -} - -func (h *genericFederatedRequestHandler) remoteQueryUUIDs(w http.ResponseWriter, - req *http.Request, - clusterID string, uuids []string) (rp []map[string]interface{}, kind string, err error) { - - found := make(map[string]bool) - prev_len_uuids := len(uuids) + 1 - // Loop while - // (1) there are more uuids to query - // (2) we're making progress - on each iteration the set of - // uuids we are expecting for must shrink. - for len(uuids) > 0 && len(uuids) < prev_len_uuids { - var remoteReq http.Request - remoteReq.Header = req.Header - remoteReq.Method = "POST" - remoteReq.URL = &url.URL{Path: req.URL.Path} - remoteParams := make(url.Values) - remoteParams.Set("_method", "GET") - remoteParams.Set("count", "none") - if req.Form.Get("select") != "" { - remoteParams.Set("select", req.Form.Get("select")) - } - content, err := json.Marshal(uuids) - if err != nil { - return nil, "", err - } - remoteParams["filters"] = []string{fmt.Sprintf(`[["uuid", "in", %s]]`, content)} - enc := remoteParams.Encode() - remoteReq.Body = ioutil.NopCloser(bytes.NewBufferString(enc)) - - rc := multiClusterQueryResponseCollector{clusterID: clusterID} - - var resp *http.Response - if clusterID == h.handler.Cluster.ClusterID { - resp, err = h.handler.localClusterRequest(&remoteReq) - } else { - resp, err = h.handler.remoteClusterRequest(clusterID, &remoteReq) - } - rc.collectResponse(resp, err) - - if rc.error != nil { - return nil, "", rc.error - } - - kind = rc.kind - - if len(rc.responses) == 0 { - // We got zero responses, no point in doing - // another query. - return rp, kind, nil - } - - rp = append(rp, rc.responses...) - - // Go through the responses and determine what was - // returned. If there are remaining items, loop - // around and do another request with just the - // stragglers. - for _, i := range rc.responses { - uuid, ok := i["uuid"].(string) - if ok { - found[uuid] = true - } - } - - l := []string{} - for _, u := range uuids { - if !found[u] { - l = append(l, u) - } - } - prev_len_uuids = len(uuids) - uuids = l - } - - return rp, kind, nil -} - -func (h *genericFederatedRequestHandler) handleMultiClusterQuery(w http.ResponseWriter, - req *http.Request, clusterId *string) bool { - - var filters [][]interface{} - err := json.Unmarshal([]byte(req.Form.Get("filters")), &filters) - if err != nil { - httpserver.Error(w, err.Error(), http.StatusBadRequest) - return true - } - - // Split the list of uuids by prefix - queryClusters := make(map[string][]string) - expectCount := 0 - for _, filter := range filters { - if len(filter) != 3 { - return false - } - - if lhs, ok := filter[0].(string); !ok || lhs != "uuid" { - return false - } - - op, ok := filter[1].(string) - if !ok { - return false - } - - if op == "in" { - if rhs, ok := filter[2].([]interface{}); ok { - for _, i := range rhs { - if u, ok := i.(string); ok { - *clusterId = u[0:5] - queryClusters[u[0:5]] = append(queryClusters[u[0:5]], u) - expectCount += 1 - } - } - } - } else if op == "=" { - if u, ok := filter[2].(string); ok { - *clusterId = u[0:5] - queryClusters[u[0:5]] = append(queryClusters[u[0:5]], u) - expectCount += 1 - } - } else { - return false - } - - } - - if len(queryClusters) <= 1 { - // Query does not search for uuids across multiple - // clusters. - return false - } - - // Validations - count := req.Form.Get("count") - if count != "" && count != `none` && count != `"none"` { - httpserver.Error(w, "Federated multi-object query must have 'count=none'", http.StatusBadRequest) - return true - } - if req.Form.Get("limit") != "" || req.Form.Get("offset") != "" || req.Form.Get("order") != "" { - httpserver.Error(w, "Federated multi-object may not provide 'limit', 'offset' or 'order'.", http.StatusBadRequest) - return true - } - if expectCount > h.handler.Cluster.RequestLimits.GetMaxItemsPerResponse() { - httpserver.Error(w, fmt.Sprintf("Federated multi-object request for %v objects which is more than max page size %v.", - expectCount, h.handler.Cluster.RequestLimits.GetMaxItemsPerResponse()), http.StatusBadRequest) - return true - } - if req.Form.Get("select") != "" { - foundUUID := false - var selects []string - err := json.Unmarshal([]byte(req.Form.Get("select")), &selects) - if err != nil { - httpserver.Error(w, err.Error(), http.StatusBadRequest) - return true - } - - for _, r := range selects { - if r == "uuid" { - foundUUID = true - break - } - } - if !foundUUID { - httpserver.Error(w, "Federated multi-object request must include 'uuid' in 'select'", http.StatusBadRequest) - return true - } - } - - // Perform concurrent requests to each cluster - - // use channel as a semaphore to limit the number of concurrent - // requests at a time - sem := make(chan bool, h.handler.Cluster.RequestLimits.GetMultiClusterRequestConcurrency()) - defer close(sem) - wg := sync.WaitGroup{} - - req.Header.Set("Content-Type", "application/x-www-form-urlencoded") - mtx := sync.Mutex{} - errors := []error{} - var completeResponses []map[string]interface{} - var kind string - - for k, v := range queryClusters { - if len(v) == 0 { - // Nothing to query - continue - } - - // blocks until it can put a value into the - // channel (which has a max queue capacity) - sem <- true - wg.Add(1) - go func(k string, v []string) { - rp, kn, err := h.remoteQueryUUIDs(w, req, k, v) - mtx.Lock() - if err == nil { - completeResponses = append(completeResponses, rp...) - kind = kn - } else { - errors = append(errors, err) - } - mtx.Unlock() - wg.Done() - <-sem - }(k, v) - } - wg.Wait() - - if len(errors) > 0 { - var strerr []string - for _, e := range errors { - strerr = append(strerr, e.Error()) - } - httpserver.Errors(w, strerr, http.StatusBadGateway) - return true - } - - w.Header().Set("Content-Type", "application/json") - w.WriteHeader(http.StatusOK) - itemList := make(map[string]interface{}) - itemList["items"] = completeResponses - itemList["kind"] = kind - json.NewEncoder(w).Encode(itemList) - - return true -} - -func (h *genericFederatedRequestHandler) ServeHTTP(w http.ResponseWriter, req *http.Request) { - m := h.matcher.FindStringSubmatch(req.URL.Path) - clusterId := "" - - if len(m) > 0 && m[2] != "" { - clusterId = m[2] - } - - // Get form parameters from URL and form body (if POST). - if err := loadParamsFromForm(req); err != nil { - httpserver.Error(w, err.Error(), http.StatusBadRequest) - return - } - - // Check if the parameters have an explicit cluster_id - if req.Form.Get("cluster_id") != "" { - clusterId = req.Form.Get("cluster_id") - } - - // Handle the POST-as-GET special case (workaround for large - // GET requests that potentially exceed maximum URL length, - // like multi-object queries where the filter has 100s of - // items) - effectiveMethod := req.Method - if req.Method == "POST" && req.Form.Get("_method") != "" { - effectiveMethod = req.Form.Get("_method") - } - - if effectiveMethod == "GET" && - clusterId == "" && - req.Form.Get("filters") != "" && - h.handleMultiClusterQuery(w, req, &clusterId) { - return - } - - if clusterId == "" || clusterId == h.handler.Cluster.ClusterID { - h.next.ServeHTTP(w, req) - } else { - resp, err := h.handler.remoteClusterRequest(clusterId, req) - h.handler.proxy.ForwardResponse(w, resp, err) - } -} - -func rewriteSignatures(clusterID string, expectHash string, - resp *http.Response, requestError error) (newResponse *http.Response, err error) { - - if requestError != nil { - return resp, requestError - } - - if resp.StatusCode != 200 { - return resp, nil - } - - originalBody := resp.Body - defer originalBody.Close() - - var col arvados.Collection - err = json.NewDecoder(resp.Body).Decode(&col) - if err != nil { - return nil, err - } - - // rewriting signatures will make manifest text 5-10% bigger so calculate - // capacity accordingly - updatedManifest := bytes.NewBuffer(make([]byte, 0, int(float64(len(col.ManifestText))*1.1))) - - hasher := md5.New() - mw := io.MultiWriter(hasher, updatedManifest) - sz := 0 - - scanner := bufio.NewScanner(strings.NewReader(col.ManifestText)) - scanner.Buffer(make([]byte, 1048576), len(col.ManifestText)) - for scanner.Scan() { - line := scanner.Text() - tokens := strings.Split(line, " ") - if len(tokens) < 3 { - return nil, fmt.Errorf("Invalid stream (<3 tokens): %q", line) - } - - n, err := mw.Write([]byte(tokens[0])) - if err != nil { - return nil, fmt.Errorf("Error updating manifest: %v", err) - } - sz += n - for _, token := range tokens[1:] { - n, err = mw.Write([]byte(" ")) - if err != nil { - return nil, fmt.Errorf("Error updating manifest: %v", err) - } - sz += n - - m := keepclient.SignedLocatorRe.FindStringSubmatch(token) - if m != nil { - // Rewrite the block signature to be a remote signature - _, err = fmt.Fprintf(updatedManifest, "%s%s%s+R%s-%s%s", m[1], m[2], m[3], clusterID, m[5][2:], m[8]) - if err != nil { - return nil, fmt.Errorf("Error updating manifest: %v", err) - } - - // for hash checking, ignore signatures - n, err = fmt.Fprintf(hasher, "%s%s", m[1], m[2]) - if err != nil { - return nil, fmt.Errorf("Error updating manifest: %v", err) - } - sz += n - } else { - n, err = mw.Write([]byte(token)) - if err != nil { - return nil, fmt.Errorf("Error updating manifest: %v", err) - } - sz += n - } - } - n, err = mw.Write([]byte("\n")) - if err != nil { - return nil, fmt.Errorf("Error updating manifest: %v", err) - } - sz += n - } - - // Check that expected hash is consistent with - // portable_data_hash field of the returned record - if expectHash == "" { - expectHash = col.PortableDataHash - } else if expectHash != col.PortableDataHash { - return nil, fmt.Errorf("portable_data_hash %q on returned record did not match expected hash %q ", expectHash, col.PortableDataHash) - } - - // Certify that the computed hash of the manifest_text matches our expectation - sum := hasher.Sum(nil) - computedHash := fmt.Sprintf("%x+%v", sum, sz) - if computedHash != expectHash { - return nil, fmt.Errorf("Computed manifest_text hash %q did not match expected hash %q", computedHash, expectHash) - } - - col.ManifestText = updatedManifest.String() - - newbody, err := json.Marshal(col) - if err != nil { - return nil, err - } - - buf := bytes.NewBuffer(newbody) - resp.Body = ioutil.NopCloser(buf) - resp.ContentLength = int64(buf.Len()) - resp.Header.Set("Content-Length", fmt.Sprintf("%v", buf.Len())) - - return resp, nil -} - -func filterLocalClusterResponse(resp *http.Response, requestError error) (newResponse *http.Response, err error) { - if requestError != nil { - return resp, requestError - } - - if resp.StatusCode == 404 { - // Suppress returning this result, because we want to - // search the federation. - return nil, nil - } - return resp, nil -} - -type searchRemoteClusterForPDH struct { - pdh string - remoteID string - mtx *sync.Mutex - sentResponse *bool - sharedContext *context.Context - cancelFunc func() - errors *[]string - statusCode *int -} - -func (s *searchRemoteClusterForPDH) filterRemoteClusterResponse(resp *http.Response, requestError error) (newResponse *http.Response, err error) { - s.mtx.Lock() - defer s.mtx.Unlock() - - if *s.sentResponse { - // Another request already returned a response - return nil, nil - } - - if requestError != nil { - *s.errors = append(*s.errors, fmt.Sprintf("Request error contacting %q: %v", s.remoteID, requestError)) - // Record the error and suppress response - return nil, nil - } - - if resp.StatusCode != 200 { - // Suppress returning unsuccessful result. Maybe - // another request will find it. - // TODO collect and return error responses. - *s.errors = append(*s.errors, fmt.Sprintf("Response from %q: %v", s.remoteID, resp.Status)) - if resp.StatusCode != 404 { - // Got a non-404 error response, convert into BadGateway - *s.statusCode = http.StatusBadGateway - } - return nil, nil - } - - s.mtx.Unlock() - - // This reads the response body. We don't want to hold the - // lock while doing this because other remote requests could - // also have made it to this point, and we don't want a - // slow response holding the lock to block a faster response - // that is waiting on the lock. - newResponse, err = rewriteSignatures(s.remoteID, s.pdh, resp, nil) - - s.mtx.Lock() - - if *s.sentResponse { - // Another request already returned a response - return nil, nil - } - - if err != nil { - // Suppress returning unsuccessful result. Maybe - // another request will be successful. - *s.errors = append(*s.errors, fmt.Sprintf("Error parsing response from %q: %v", s.remoteID, err)) - return nil, nil - } - - // We have a successful response. Suppress/cancel all the - // other requests/responses. - *s.sentResponse = true - s.cancelFunc() - - return newResponse, nil -} - -func (h *collectionFederatedRequestHandler) ServeHTTP(w http.ResponseWriter, req *http.Request) { - if req.Method != "GET" { - // Only handle GET requests right now - h.next.ServeHTTP(w, req) - return - } - - m := collectionByPDHRe.FindStringSubmatch(req.URL.Path) - if len(m) != 2 { - // Not a collection PDH GET request - m = collectionRe.FindStringSubmatch(req.URL.Path) - clusterId := "" - - if len(m) > 0 { - clusterId = m[2] - } - - if clusterId != "" && clusterId != h.handler.Cluster.ClusterID { - // request for remote collection by uuid - resp, err := h.handler.remoteClusterRequest(clusterId, req) - newResponse, err := rewriteSignatures(clusterId, "", resp, err) - h.handler.proxy.ForwardResponse(w, newResponse, err) - return - } - // not a collection UUID request, or it is a request - // for a local UUID, either way, continue down the - // handler stack. - h.next.ServeHTTP(w, req) - return - } - - // Request for collection by PDH. Search the federation. - - // First, query the local cluster. - resp, err := h.handler.localClusterRequest(req) - newResp, err := filterLocalClusterResponse(resp, err) - if newResp != nil || err != nil { - h.handler.proxy.ForwardResponse(w, newResp, err) - return - } - - sharedContext, cancelFunc := context.WithCancel(req.Context()) - defer cancelFunc() - req = req.WithContext(sharedContext) - - // Create a goroutine for each cluster in the - // RemoteClusters map. The first valid result gets - // returned to the client. When that happens, all - // other outstanding requests are cancelled or - // suppressed. - sentResponse := false - mtx := sync.Mutex{} - wg := sync.WaitGroup{} - var errors []string - var errorCode int = 404 - - // use channel as a semaphore to limit the number of concurrent - // requests at a time - sem := make(chan bool, h.handler.Cluster.RequestLimits.GetMultiClusterRequestConcurrency()) - defer close(sem) - for remoteID := range h.handler.Cluster.RemoteClusters { - if remoteID == h.handler.Cluster.ClusterID { - // No need to query local cluster again - continue - } - // blocks until it can put a value into the - // channel (which has a max queue capacity) - sem <- true - if sentResponse { - break - } - search := &searchRemoteClusterForPDH{m[1], remoteID, &mtx, &sentResponse, - &sharedContext, cancelFunc, &errors, &errorCode} - wg.Add(1) - go func() { - resp, err := h.handler.remoteClusterRequest(search.remoteID, req) - newResp, err := search.filterRemoteClusterResponse(resp, err) - if newResp != nil || err != nil { - h.handler.proxy.ForwardResponse(w, newResp, err) - } - wg.Done() - <-sem - }() - } - wg.Wait() - - if sentResponse { - return - } - - // No successful responses, so return the error - httpserver.Errors(w, errors, errorCode) -} - func (h *Handler) setupProxyRemoteCluster(next http.Handler) http.Handler { mux := http.NewServeMux() mux.Handle("/arvados/v1/workflows", &genericFederatedRequestHandler{next, h, wfRe}) -- 2.30.2