1 /* Deals with getting Keep Server blocks from API Server and Keep Servers. */
10 "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
11 "git.curoverse.com/arvados.git/sdk/go/blockdigest"
12 "git.curoverse.com/arvados.git/sdk/go/keepclient"
13 "git.curoverse.com/arvados.git/sdk/go/logger"
24 // ServerAddress struct
25 type ServerAddress struct {
26 SSL bool `json:"service_ssl_flag"`
27 Host string `json:"service_host"`
28 Port int `json:"service_port"`
29 UUID string `json:"uuid"`
30 ServiceType string `json:"service_type"`
33 // BlockInfo is info about a particular block returned by the server
34 type BlockInfo struct {
35 Digest blockdigest.DigestWithSize
36 Mtime int64 // TODO(misha): Replace this with a timestamp.
39 // BlockServerInfo is info about a specified block given by a server
40 type BlockServerInfo struct {
42 Mtime int64 // TODO(misha): Replace this with a timestamp.
45 // ServerContents struct
46 type ServerContents struct {
47 BlockDigestToInfo map[blockdigest.DigestWithSize]BlockInfo
50 // ServerResponse struct
51 type ServerResponse struct {
53 Contents ServerContents
58 type ReadServers struct {
60 KeepServerIndexToAddress []ServerAddress
61 KeepServerAddressToIndex map[ServerAddress]int
62 ServerToContents map[ServerAddress]ServerContents
63 BlockToServers map[blockdigest.DigestWithSize][]BlockServerInfo
64 BlockReplicationCounts map[int]int
67 // GetKeepServersParams struct
68 type GetKeepServersParams struct {
69 Client arvadosclient.ArvadosClient
74 // ServiceList consists of the addresses of all the available kee servers
75 type ServiceList struct {
76 ItemsAvailable int `json:"items_available"`
77 KeepServers []ServerAddress `json:"items"`
81 // TODO(misha): Change this to include the UUID as well.
82 func (s ServerAddress) String() string {
86 // URL of the keep server
87 func (s ServerAddress) URL() string {
89 return fmt.Sprintf("https://%s:%d", s.Host, s.Port)
91 return fmt.Sprintf("http://%s:%d", s.Host, s.Port)
94 // GetKeepServersAndSummarize gets keep servers from api
95 func GetKeepServersAndSummarize(params GetKeepServersParams) (results ReadServers, err error) {
96 results, err = GetKeepServers(params)
100 log.Printf("Returned %d keep disks", len(results.ServerToContents))
102 results.Summarize(params.Logger)
103 log.Printf("Replication level distribution: %v",
104 results.BlockReplicationCounts)
109 // GetKeepServers from api server
110 func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error) {
111 sdkParams := arvadosclient.Dict{
112 "filters": [][]string{[]string{"service_type", "!=", "proxy"}},
114 if params.Limit > 0 {
115 sdkParams["limit"] = params.Limit
118 var sdkResponse ServiceList
119 err = params.Client.List("keep_services", sdkParams, &sdkResponse)
125 // Currently, only "disk" types are supported. Stop if any other service types are found.
126 for _, server := range sdkResponse.KeepServers {
127 if server.ServiceType != "disk" {
128 return results, fmt.Errorf("Unsupported service type %q found for: %v", server.ServiceType, server)
132 if params.Logger != nil {
133 params.Logger.Update(func(p map[string]interface{}, e map[string]interface{}) {
134 keepInfo := logger.GetOrCreateMap(p, "keep_info")
135 keepInfo["num_keep_servers_available"] = sdkResponse.ItemsAvailable
136 keepInfo["num_keep_servers_received"] = len(sdkResponse.KeepServers)
137 keepInfo["keep_servers"] = sdkResponse.KeepServers
141 log.Printf("Received keep services list: %+v", sdkResponse)
143 if len(sdkResponse.KeepServers) < sdkResponse.ItemsAvailable {
144 return results, fmt.Errorf("Did not receive all available keep servers: %+v", sdkResponse)
147 results.KeepServerIndexToAddress = sdkResponse.KeepServers
148 results.KeepServerAddressToIndex = make(map[ServerAddress]int)
149 for i, address := range results.KeepServerIndexToAddress {
150 results.KeepServerAddressToIndex[address] = i
153 log.Printf("Got Server Addresses: %v", results)
155 // Send off all the index requests concurrently
156 responseChan := make(chan ServerResponse)
157 for _, keepServer := range sdkResponse.KeepServers {
158 // The above keepsServer variable is reused for each iteration, so
159 // it would be shared across all goroutines. This would result in
160 // us querying one server n times instead of n different servers
161 // as we intended. To avoid this we add it as an explicit
162 // parameter which gets copied. This bug and solution is described
163 // in https://golang.org/doc/effective_go.html#channels
164 go func(keepServer ServerAddress) {
165 responseChan <- GetServerContents(params.Logger,
171 results.ServerToContents = make(map[ServerAddress]ServerContents)
172 results.BlockToServers = make(map[blockdigest.DigestWithSize][]BlockServerInfo)
174 // Read all the responses
175 for i := range sdkResponse.KeepServers {
176 _ = i // Here to prevent go from complaining.
177 response := <-responseChan
179 // Check if there were any errors during GetServerContents
180 if response.Err != nil {
181 return results, response.Err
184 log.Printf("Received channel response from %v containing %d files",
186 len(response.Contents.BlockDigestToInfo))
187 results.ServerToContents[response.Address] = response.Contents
188 serverIndex := results.KeepServerAddressToIndex[response.Address]
189 for _, blockInfo := range response.Contents.BlockDigestToInfo {
190 results.BlockToServers[blockInfo.Digest] = append(
191 results.BlockToServers[blockInfo.Digest],
192 BlockServerInfo{ServerIndex: serverIndex,
193 Mtime: blockInfo.Mtime})
199 // GetServerContents of the keep server
200 func GetServerContents(arvLogger *logger.Logger,
201 keepServer ServerAddress,
202 arv arvadosclient.ArvadosClient) (response ServerResponse) {
204 err := GetServerStatus(arvLogger, keepServer, arv)
210 req, err := CreateIndexRequest(arvLogger, keepServer, arv)
216 resp, err := arv.Client.Do(req)
222 response, err = ReadServerResponse(arvLogger, keepServer, resp)
231 // GetServerStatus get keep server status by invoking /status.json
232 func GetServerStatus(arvLogger *logger.Logger,
233 keepServer ServerAddress,
234 arv arvadosclient.ArvadosClient) error {
235 url := fmt.Sprintf("http://%s:%d/status.json",
239 if arvLogger != nil {
241 arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
242 keepInfo := logger.GetOrCreateMap(p, "keep_info")
243 serverInfo := make(map[string]interface{})
244 serverInfo["status_request_sent_at"] = now
245 serverInfo["host"] = keepServer.Host
246 serverInfo["port"] = keepServer.Port
248 keepInfo[keepServer.UUID] = serverInfo
250 runInfo := logger.GetOrCreateMap(p, "run_info")
251 runInfo["args"] = os.Args
255 resp, err := arv.Client.Get(url)
257 return fmt.Errorf("Error getting keep status from %s: %v", url, err)
258 } else if resp.StatusCode != 200 {
259 return fmt.Errorf("Received error code %d in response to request "+
261 resp.StatusCode, url, resp.Status)
264 var keepStatus map[string]interface{}
265 decoder := json.NewDecoder(resp.Body)
267 err = decoder.Decode(&keepStatus)
269 return fmt.Errorf("Error decoding keep status from %s: %v", url, err)
272 if arvLogger != nil {
274 arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
275 keepInfo := logger.GetOrCreateMap(p, "keep_info")
276 serverInfo := keepInfo[keepServer.UUID].(map[string]interface{})
277 serverInfo["status_response_processed_at"] = now
278 serverInfo["status"] = keepStatus
280 runInfo := logger.GetOrCreateMap(p, "run_info")
281 runInfo["args"] = os.Args
288 // CreateIndexRequest to the keep server
289 func CreateIndexRequest(arvLogger *logger.Logger,
290 keepServer ServerAddress,
291 arv arvadosclient.ArvadosClient) (req *http.Request, err error) {
292 url := fmt.Sprintf("http://%s:%d/index", keepServer.Host, keepServer.Port)
293 log.Println("About to fetch keep server contents from " + url)
295 if arvLogger != nil {
297 arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
298 keepInfo := logger.GetOrCreateMap(p, "keep_info")
299 serverInfo := keepInfo[keepServer.UUID].(map[string]interface{})
300 serverInfo["index_request_sent_at"] = now
302 runInfo := logger.GetOrCreateMap(p, "run_info")
303 runInfo["args"] = os.Args
307 req, err = http.NewRequest("GET", url, nil)
309 return req, fmt.Errorf("Error building http request for %s: %v", url, err)
312 req.Header.Add("Authorization", "OAuth2 "+arv.ApiToken)
316 // ReadServerResponse reads reasponse from keep server
317 func ReadServerResponse(arvLogger *logger.Logger,
318 keepServer ServerAddress,
319 resp *http.Response) (response ServerResponse, err error) {
321 if resp.StatusCode != 200 {
322 return response, fmt.Errorf("Received error code %d in response to index request for %s: %s",
323 resp.StatusCode, keepServer.String(), resp.Status)
326 if arvLogger != nil {
328 arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
329 keepInfo := logger.GetOrCreateMap(p, "keep_info")
330 serverInfo := keepInfo[keepServer.UUID].(map[string]interface{})
331 serverInfo["index_response_received_at"] = now
333 runInfo := logger.GetOrCreateMap(p, "run_info")
334 runInfo["args"] = os.Args
338 response.Address = keepServer
339 response.Contents.BlockDigestToInfo =
340 make(map[blockdigest.DigestWithSize]BlockInfo)
341 reader := bufio.NewReader(resp.Body)
342 numLines, numDuplicates, numSizeDisagreements := 0, 0, 0
345 line, err := reader.ReadString('\n')
347 return response, fmt.Errorf("Index from %s truncated at line %d",
348 keepServer.String(), numLines)
349 } else if err != nil {
350 return response, fmt.Errorf("Error reading index response from %s at line %d: %v",
351 keepServer.String(), numLines, err)
354 if _, err := reader.Peek(1); err == nil {
355 extra, _ := reader.ReadString('\n')
356 return response, fmt.Errorf("Index from %s had trailing data at line %d after EOF marker: %s",
357 keepServer.String(), numLines+1, extra)
358 } else if err != io.EOF {
359 return response, fmt.Errorf("Index from %s had read error after EOF marker at line %d: %v",
360 keepServer.String(), numLines, err)
365 blockInfo, err := parseBlockInfoFromIndexLine(line)
367 return response, fmt.Errorf("Error parsing BlockInfo from index line "+
368 "received from %s: %v",
373 if storedBlock, ok := response.Contents.BlockDigestToInfo[blockInfo.Digest]; ok {
374 // This server returned multiple lines containing the same block digest.
376 // Keep the block that's newer.
377 if storedBlock.Mtime < blockInfo.Mtime {
378 response.Contents.BlockDigestToInfo[blockInfo.Digest] = blockInfo
381 response.Contents.BlockDigestToInfo[blockInfo.Digest] = blockInfo
385 log.Printf("%s index contained %d lines with %d duplicates with "+
386 "%d size disagreements",
390 numSizeDisagreements)
392 if arvLogger != nil {
394 arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
395 keepInfo := logger.GetOrCreateMap(p, "keep_info")
396 serverInfo := keepInfo[keepServer.UUID].(map[string]interface{})
398 serverInfo["processing_finished_at"] = now
399 serverInfo["lines_received"] = numLines
400 serverInfo["duplicates_seen"] = numDuplicates
401 serverInfo["size_disagreements_seen"] = numSizeDisagreements
403 runInfo := logger.GetOrCreateMap(p, "run_info")
404 runInfo["args"] = os.Args
411 func parseBlockInfoFromIndexLine(indexLine string) (blockInfo BlockInfo, err error) {
412 tokens := strings.Fields(indexLine)
413 if len(tokens) != 2 {
414 err = fmt.Errorf("Expected 2 tokens per line but received a "+
415 "line containing %#q instead.",
419 var locator blockdigest.BlockLocator
420 if locator, err = blockdigest.ParseBlockLocator(tokens[0]); err != nil {
421 err = fmt.Errorf("%v Received error while parsing line \"%#q\"",
425 if len(locator.Hints) > 0 {
426 err = fmt.Errorf("Block locator in index line should not contain hints "+
432 blockInfo.Mtime, err = strconv.ParseInt(tokens[1], 10, 64)
437 blockdigest.DigestWithSize{Digest: locator.Digest,
438 Size: uint32(locator.Size)}
442 // Summarize results from keep server
443 func (readServers *ReadServers) Summarize(arvLogger *logger.Logger) {
444 readServers.BlockReplicationCounts = make(map[int]int)
445 for _, infos := range readServers.BlockToServers {
446 replication := len(infos)
447 readServers.BlockReplicationCounts[replication]++
450 if arvLogger != nil {
451 arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
452 keepInfo := logger.GetOrCreateMap(p, "keep_info")
453 keepInfo["distinct_blocks_stored"] = len(readServers.BlockToServers)
455 runInfo := logger.GetOrCreateMap(p, "run_info")
456 runInfo["args"] = os.Args
461 // TrashRequest struct
462 type TrashRequest struct {
463 Locator string `json:"locator"`
464 BlockMtime int64 `json:"block_mtime"`
467 // TrashList is an array of TrashRequest objects
468 type TrashList []TrashRequest
470 // SendTrashLists to trash queue
471 func SendTrashLists(kc *keepclient.KeepClient, spl map[string]TrashList) (errs []error) {
473 barrier := make(chan error)
477 for url, v := range spl {
479 log.Printf("Sending trash list to %v", url)
481 go (func(url string, v TrashList) {
482 pipeReader, pipeWriter := io.Pipe()
484 enc := json.NewEncoder(pipeWriter)
489 req, err := http.NewRequest("PUT", fmt.Sprintf("%s/trash", url), pipeReader)
491 log.Printf("Error creating trash list request for %v error: %v", url, err.Error())
496 req.Header.Add("Authorization", "OAuth2 "+kc.Arvados.ApiToken)
499 var resp *http.Response
500 if resp, err = client.Do(req); err != nil {
501 log.Printf("Error sending trash list to %v error: %v", url, err.Error())
506 log.Printf("Sent trash list to %v: response was HTTP %v", url, resp.Status)
508 io.Copy(ioutil.Discard, resp.Body)
511 if resp.StatusCode != 200 {
512 barrier <- errors.New(fmt.Sprintf("Got HTTP code %v", resp.StatusCode))
520 for i := 0; i < count; i++ {
523 errs = append(errs, b)