X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/8d2aebfd3a0e4814b292659710386f949cafe092..e3a1bf022cbc081cf7c7ebb441dc1ab1df200f98:/services/datamanager/keep/keep.go diff --git a/services/datamanager/keep/keep.go b/services/datamanager/keep/keep.go index c2346cd154..3a9c21a438 100644 --- a/services/datamanager/keep/keep.go +++ b/services/datamanager/keep/keep.go @@ -5,10 +5,11 @@ package keep import ( "bufio" "encoding/json" - "flag" + "errors" "fmt" "git.curoverse.com/arvados.git/sdk/go/arvadosclient" "git.curoverse.com/arvados.git/sdk/go/blockdigest" + "git.curoverse.com/arvados.git/sdk/go/keepclient" "git.curoverse.com/arvados.git/sdk/go/logger" "git.curoverse.com/arvados.git/services/datamanager/loggerutil" "io" @@ -17,37 +18,42 @@ import ( "net/http" "strconv" "strings" - "sync" "time" ) +// ServerAddress struct type ServerAddress struct { - Host string `json:"service_host"` - Port int `json:"service_port"` - Uuid string `json:"uuid"` + SSL bool `json:"service_ssl_flag"` + Host string `json:"service_host"` + Port int `json:"service_port"` + UUID string `json:"uuid"` + ServiceType string `json:"service_type"` } -// Info about a particular block returned by the server +// BlockInfo is info about a particular block returned by the server type BlockInfo struct { Digest blockdigest.DigestWithSize Mtime int64 // TODO(misha): Replace this with a timestamp. } -// Info about a specified block given by a server +// BlockServerInfo is info about a specified block given by a server type BlockServerInfo struct { ServerIndex int Mtime int64 // TODO(misha): Replace this with a timestamp. } +// ServerContents struct type ServerContents struct { BlockDigestToInfo map[blockdigest.DigestWithSize]BlockInfo } +// ServerResponse struct type ServerResponse struct { Address ServerAddress Contents ServerContents } +// ReadServers struct type ReadServers struct { ReadAllServers bool KeepServerIndexToAddress []ServerAddress @@ -57,63 +63,34 @@ type ReadServers struct { BlockReplicationCounts map[int]int } +// GetKeepServersParams struct type GetKeepServersParams struct { Client arvadosclient.ArvadosClient Logger *logger.Logger Limit int } -type KeepServiceList struct { +// ServiceList consists of the addresses of all the available kee servers +type ServiceList struct { ItemsAvailable int `json:"items_available"` KeepServers []ServerAddress `json:"items"` } -var ( - // Don't access the token directly, use getDataManagerToken() to - // make sure it's been read. - dataManagerToken string - dataManagerTokenFile string - dataManagerTokenFileReadOnce sync.Once -) - -func init() { - flag.StringVar(&dataManagerTokenFile, - "data-manager-token-file", - "", - "File with the API token we should use to contact keep servers.") -} - +// String // TODO(misha): Change this to include the UUID as well. func (s ServerAddress) String() string { - return s.HostPort() -} - -func (s ServerAddress) HostPort() string { - return fmt.Sprintf("%s:%d", s.Host, s.Port) + return s.URL() } -func getDataManagerToken(arvLogger *logger.Logger) string { - readDataManagerToken := func() { - if dataManagerTokenFile == "" { - flag.Usage() - loggerutil.FatalWithMessage(arvLogger, - "Data Manager Token needed, but data manager token file not specified.") - } else { - rawRead, err := ioutil.ReadFile(dataManagerTokenFile) - if err != nil { - loggerutil.FatalWithMessage(arvLogger, - fmt.Sprintf("Unexpected error reading token file %s: %v", - dataManagerTokenFile, - err)) - } - dataManagerToken = strings.TrimSpace(string(rawRead)) - } +// URL of the keep server +func (s ServerAddress) URL() string { + if s.SSL { + return fmt.Sprintf("https://%s:%d", s.Host, s.Port) } - - dataManagerTokenFileReadOnce.Do(readDataManagerToken) - return dataManagerToken + return fmt.Sprintf("http://%s:%d", s.Host, s.Port) } +// GetKeepServersAndSummarize gets keep servers from api func GetKeepServersAndSummarize(params GetKeepServersParams) (results ReadServers) { results = GetKeepServers(params) log.Printf("Returned %d keep disks", len(results.ServerToContents)) @@ -125,20 +102,16 @@ func GetKeepServersAndSummarize(params GetKeepServersParams) (results ReadServer return } +// GetKeepServers from api server func GetKeepServers(params GetKeepServersParams) (results ReadServers) { - if ¶ms.Client == nil { - log.Fatalf("params.Client passed to GetKeepServers() should " + - "contain a valid ArvadosClient, but instead it is nil.") - } - sdkParams := arvadosclient.Dict{ - "filters": [][]string{[]string{"service_type", "=", "disk"}}, + "filters": [][]string{[]string{"service_type", "!=", "proxy"}}, } if params.Limit > 0 { sdkParams["limit"] = params.Limit } - var sdkResponse KeepServiceList + var sdkResponse ServiceList err := params.Client.List("keep_services", sdkParams, &sdkResponse) if err != nil { @@ -146,6 +119,14 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers) { fmt.Sprintf("Error requesting keep disks from API server: %v", err)) } + // Currently, only "disk" types are supported. Stop if any other service types are found. + for _, server := range sdkResponse.KeepServers { + if server.ServiceType != "disk" { + loggerutil.FatalWithMessage(params.Logger, + fmt.Sprintf("Unsupported service type %q found for: %v", server.ServiceType, server)) + } + } + if params.Logger != nil { params.Logger.Update(func(p map[string]interface{}, e map[string]interface{}) { keepInfo := logger.GetOrCreateMap(p, "keep_info") @@ -170,9 +151,6 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers) { log.Printf("Got Server Addresses: %v", results) - // This is safe for concurrent use - client := http.Client{} - // Send off all the index requests concurrently responseChan := make(chan ServerResponse) for _, keepServer := range sdkResponse.KeepServers { @@ -185,7 +163,7 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers) { go func(keepServer ServerAddress) { responseChan <- GetServerContents(params.Logger, keepServer, - client) + params.Client) }(keepServer) } @@ -211,14 +189,15 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers) { return } +// GetServerContents of the keep server func GetServerContents(arvLogger *logger.Logger, keepServer ServerAddress, - client http.Client) (response ServerResponse) { + arv arvadosclient.ArvadosClient) (response ServerResponse) { - GetServerStatus(arvLogger, keepServer, client) + GetServerStatus(arvLogger, keepServer, arv) - req := CreateIndexRequest(arvLogger, keepServer) - resp, err := client.Do(req) + req := CreateIndexRequest(arvLogger, keepServer, arv) + resp, err := arv.Client.Do(req) if err != nil { loggerutil.FatalWithMessage(arvLogger, fmt.Sprintf("Error fetching %s: %v. Response was %+v", @@ -230,9 +209,10 @@ func GetServerContents(arvLogger *logger.Logger, return ReadServerResponse(arvLogger, keepServer, resp) } +// GetServerStatus get keep server status by invoking /status.json func GetServerStatus(arvLogger *logger.Logger, keepServer ServerAddress, - client http.Client) { + arv arvadosclient.ArvadosClient) { url := fmt.Sprintf("http://%s:%d/status.json", keepServer.Host, keepServer.Port) @@ -246,11 +226,11 @@ func GetServerStatus(arvLogger *logger.Logger, serverInfo["host"] = keepServer.Host serverInfo["port"] = keepServer.Port - keepInfo[keepServer.Uuid] = serverInfo + keepInfo[keepServer.UUID] = serverInfo }) } - resp, err := client.Get(url) + resp, err := arv.Client.Get(url) if err != nil { loggerutil.FatalWithMessage(arvLogger, fmt.Sprintf("Error getting keep status from %s: %v", url, err)) @@ -274,15 +254,17 @@ func GetServerStatus(arvLogger *logger.Logger, now := time.Now() arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) { keepInfo := logger.GetOrCreateMap(p, "keep_info") - serverInfo := keepInfo[keepServer.Uuid].(map[string]interface{}) + serverInfo := keepInfo[keepServer.UUID].(map[string]interface{}) serverInfo["status_response_processed_at"] = now serverInfo["status"] = keepStatus }) } } +// CreateIndexRequest to the keep server func CreateIndexRequest(arvLogger *logger.Logger, - keepServer ServerAddress) (req *http.Request) { + keepServer ServerAddress, + arv arvadosclient.ArvadosClient) (req *http.Request) { url := fmt.Sprintf("http://%s:%d/index", keepServer.Host, keepServer.Port) log.Println("About to fetch keep server contents from " + url) @@ -290,7 +272,7 @@ func CreateIndexRequest(arvLogger *logger.Logger, now := time.Now() arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) { keepInfo := logger.GetOrCreateMap(p, "keep_info") - serverInfo := keepInfo[keepServer.Uuid].(map[string]interface{}) + serverInfo := keepInfo[keepServer.UUID].(map[string]interface{}) serverInfo["index_request_sent_at"] = now }) } @@ -301,11 +283,11 @@ func CreateIndexRequest(arvLogger *logger.Logger, fmt.Sprintf("Error building http request for %s: %v", url, err)) } - req.Header.Add("Authorization", - fmt.Sprintf("OAuth2 %s", getDataManagerToken(arvLogger))) + req.Header.Add("Authorization", "OAuth2 "+arv.ApiToken) return } +// ReadServerResponse reads reasponse from keep server func ReadServerResponse(arvLogger *logger.Logger, keepServer ServerAddress, resp *http.Response) (response ServerResponse) { @@ -321,7 +303,7 @@ func ReadServerResponse(arvLogger *logger.Logger, now := time.Now() arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) { keepInfo := logger.GetOrCreateMap(p, "keep_info") - serverInfo := keepInfo[keepServer.Uuid].(map[string]interface{}) + serverInfo := keepInfo[keepServer.UUID].(map[string]interface{}) serverInfo["index_response_received_at"] = now }) } @@ -368,7 +350,7 @@ func ReadServerResponse(arvLogger *logger.Logger, if storedBlock, ok := response.Contents.BlockDigestToInfo[blockInfo.Digest]; ok { // This server returned multiple lines containing the same block digest. - numDuplicates += 1 + numDuplicates++ // Keep the block that's newer. if storedBlock.Mtime < blockInfo.Mtime { response.Contents.BlockDigestToInfo[blockInfo.Digest] = blockInfo @@ -389,7 +371,7 @@ func ReadServerResponse(arvLogger *logger.Logger, now := time.Now() arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) { keepInfo := logger.GetOrCreateMap(p, "keep_info") - serverInfo := keepInfo[keepServer.Uuid].(map[string]interface{}) + serverInfo := keepInfo[keepServer.UUID].(map[string]interface{}) serverInfo["processing_finished_at"] = now serverInfo["lines_received"] = numLines @@ -432,11 +414,12 @@ func parseBlockInfoFromIndexLine(indexLine string) (blockInfo BlockInfo, err err return } +// Summarize results from keep server func (readServers *ReadServers) Summarize(arvLogger *logger.Logger) { readServers.BlockReplicationCounts = make(map[int]int) for _, infos := range readServers.BlockToServers { replication := len(infos) - readServers.BlockReplicationCounts[replication] += 1 + readServers.BlockReplicationCounts[replication]++ } if arvLogger != nil { @@ -445,5 +428,73 @@ func (readServers *ReadServers) Summarize(arvLogger *logger.Logger) { keepInfo["distinct_blocks_stored"] = len(readServers.BlockToServers) }) } +} + +// TrashRequest struct +type TrashRequest struct { + Locator string `json:"locator"` + BlockMtime int64 `json:"block_mtime"` +} + +// TrashList is an array of TrashRequest objects +type TrashList []TrashRequest + +// SendTrashLists to trash queue +func SendTrashLists(kc *keepclient.KeepClient, spl map[string]TrashList) (errs []error) { + count := 0 + barrier := make(chan error) + + client := kc.Client + + for url, v := range spl { + count++ + log.Printf("Sending trash list to %v", url) + + go (func(url string, v TrashList) { + pipeReader, pipeWriter := io.Pipe() + go (func() { + enc := json.NewEncoder(pipeWriter) + enc.Encode(v) + pipeWriter.Close() + })() + + req, err := http.NewRequest("PUT", fmt.Sprintf("%s/trash", url), pipeReader) + if err != nil { + log.Printf("Error creating trash list request for %v error: %v", url, err.Error()) + barrier <- err + return + } + + req.Header.Add("Authorization", "OAuth2 "+kc.Arvados.ApiToken) + + // Make the request + var resp *http.Response + if resp, err = client.Do(req); err != nil { + log.Printf("Error sending trash list to %v error: %v", url, err.Error()) + barrier <- err + return + } + + log.Printf("Sent trash list to %v: response was HTTP %v", url, resp.Status) + + io.Copy(ioutil.Discard, resp.Body) + resp.Body.Close() + + if resp.StatusCode != 200 { + barrier <- errors.New(fmt.Sprintf("Got HTTP code %v", resp.StatusCode)) + } else { + barrier <- nil + } + })(url, v) + + } + + for i := 0; i < count; i++ { + b := <-barrier + if b != nil { + errs = append(errs, b) + } + } + return errs }