16 "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
17 "git.curoverse.com/arvados.git/sdk/go/keepclient"
28 flags := flag.NewFlagSet("keep-rsync", flag.ExitOnError)
30 srcConfigFile := flags.String(
33 "Source configuration filename. May be either a pathname to a config file, or (for example) 'foo' as shorthand for $HOME/.config/arvados/foo.conf file. This file is expected to specify the values for ARVADOS_API_TOKEN, ARVADOS_API_HOST, ARVADOS_API_HOST_INSECURE, and ARVADOS_BLOB_SIGNING_KEY for the source.")
35 dstConfigFile := flags.String(
38 "Destination configuration filename. May be either a pathname to a config file, or (for example) 'foo' as shorthand for $HOME/.config/arvados/foo.conf file. This file is expected to specify the values for ARVADOS_API_TOKEN, ARVADOS_API_HOST, and ARVADOS_API_HOST_INSECURE for the destination.")
40 srcKeepServicesJSON := flags.String(
41 "src-keep-services-json",
43 "An optional list of available source keepservices. "+
44 "If not provided, this list is obtained from api server configured in src-config-file.")
46 dstKeepServicesJSON := flags.String(
47 "dst-keep-services-json",
49 "An optional list of available destination keepservices. "+
50 "If not provided, this list is obtained from api server configured in dst-config-file.")
52 replications := flags.Int(
55 "Number of replications to write to the destination. If replications not specified, "+
56 "default replication level configured on destination server will be used.")
58 prefix := flags.String(
63 srcBlobSignatureTTLFlag := flags.Duration(
64 "src-blob-signature-ttl",
66 "Lifetime of blob permission signatures on source keepservers. If not provided, this will be retrieved from the API server's discovery document.")
68 // Parse args; omit the first arg which is the command name
69 flags.Parse(os.Args[1:])
71 srcConfig, srcBlobSigningKey, err := loadConfig(*srcConfigFile)
73 return fmt.Errorf("Error loading src configuration from file: %s", err.Error())
76 dstConfig, _, err := loadConfig(*dstConfigFile)
78 return fmt.Errorf("Error loading dst configuration from file: %s", err.Error())
81 // setup src and dst keepclients
82 kcSrc, srcBlobSignatureTTL, err := setupKeepClient(srcConfig, *srcKeepServicesJSON, false, 0, *srcBlobSignatureTTLFlag)
84 return fmt.Errorf("Error configuring src keepclient: %s", err.Error())
87 kcDst, _, err := setupKeepClient(dstConfig, *dstKeepServicesJSON, true, *replications, 0)
89 return fmt.Errorf("Error configuring dst keepclient: %s", err.Error())
92 // Copy blocks not found in dst from src
93 err = performKeepRsync(kcSrc, kcDst, srcBlobSignatureTTL, srcBlobSigningKey, *prefix)
95 return fmt.Errorf("Error while syncing data: %s", err.Error())
101 type apiConfig struct {
108 // Load src and dst config from given files
109 func loadConfig(configFile string) (config apiConfig, blobSigningKey string, err error) {
110 if configFile == "" {
111 return config, blobSigningKey, errors.New("config file not specified")
114 config, blobSigningKey, err = readConfigFromFile(configFile)
116 return config, blobSigningKey, fmt.Errorf("Error reading config file: %v", err)
122 // Read config from file
123 func readConfigFromFile(filename string) (config apiConfig, blobSigningKey string, err error) {
124 if !strings.Contains(filename, "/") {
125 filename = os.Getenv("HOME") + "/.config/arvados/" + filename + ".conf"
128 content, err := ioutil.ReadFile(filename)
131 return config, "", err
134 lines := strings.Split(string(content), "\n")
135 for _, line := range lines {
140 kv := strings.SplitN(line, "=", 2)
141 key := strings.TrimSpace(kv[0])
142 value := strings.TrimSpace(kv[1])
145 case "ARVADOS_API_TOKEN":
146 config.APIToken = value
147 case "ARVADOS_API_HOST":
148 config.APIHost = value
149 case "ARVADOS_API_HOST_INSECURE":
150 config.APIHostInsecure = arvadosclient.StringBool(value)
151 case "ARVADOS_EXTERNAL_CLIENT":
152 config.ExternalClient = arvadosclient.StringBool(value)
153 case "ARVADOS_BLOB_SIGNING_KEY":
154 blobSigningKey = value
160 // setup keepclient using the config provided
161 func setupKeepClient(config apiConfig, keepServicesJSON string, isDst bool, replications int, srcBlobSignatureTTL time.Duration) (kc *keepclient.KeepClient, blobSignatureTTL time.Duration, err error) {
162 arv := arvadosclient.ArvadosClient{
163 ApiToken: config.APIToken,
164 ApiServer: config.APIHost,
165 ApiInsecure: config.APIHostInsecure,
166 Client: &http.Client{Transport: &http.Transport{
167 TLSClientConfig: &tls.Config{InsecureSkipVerify: config.APIHostInsecure}}},
168 External: config.ExternalClient,
171 // If keepServicesJSON is provided, use it instead of service discovery
172 if keepServicesJSON == "" {
173 kc, err = keepclient.MakeKeepClient(&arv)
178 kc = keepclient.New(&arv)
179 err = kc.LoadKeepServicesFromJSON(keepServicesJSON)
186 // Get default replications value from destination, if it is not already provided
187 if replications == 0 {
188 value, err := arv.Discovery("defaultCollectionReplication")
190 replications = int(value.(float64))
196 kc.Want_replicas = replications
199 // If srcBlobSignatureTTL is not provided, get it from API server discovery doc
200 blobSignatureTTL = srcBlobSignatureTTL
201 if !isDst && srcBlobSignatureTTL == 0 {
202 value, err := arv.Discovery("blobSignatureTtl")
204 blobSignatureTTL = time.Duration(int(value.(float64))) * time.Second
210 return kc, blobSignatureTTL, nil
213 // Get unique block locators from src and dst
214 // Copy any blocks missing in dst
215 func performKeepRsync(kcSrc, kcDst *keepclient.KeepClient, srcBlobSignatureTTL time.Duration, blobSigningKey, prefix string) error {
216 // Get unique locators from src
217 srcIndex, err := getUniqueLocators(kcSrc, prefix)
222 // Get unique locators from dst
223 dstIndex, err := getUniqueLocators(kcDst, prefix)
228 // Get list of locators found in src, but missing in dst
229 toBeCopied := getMissingLocators(srcIndex, dstIndex)
231 // Copy each missing block to dst
232 log.Printf("Before keep-rsync, there are %d blocks in src and %d blocks in dst. Start copying %d blocks from src not found in dst.",
233 len(srcIndex), len(dstIndex), len(toBeCopied))
235 err = copyBlocksToDst(toBeCopied, kcSrc, kcDst, srcBlobSignatureTTL, blobSigningKey)
240 // Get list of unique locators from the specified cluster
241 func getUniqueLocators(kc *keepclient.KeepClient, prefix string) (map[string]bool, error) {
242 uniqueLocators := map[string]bool{}
244 // Get index and dedup
245 for uuid := range kc.LocalRoots() {
246 reader, err := kc.GetIndex(uuid, prefix)
248 return uniqueLocators, err
250 scanner := bufio.NewScanner(reader)
252 uniqueLocators[strings.Split(scanner.Text(), " ")[0]] = true
256 return uniqueLocators, nil
259 // Get list of locators that are in src but not in dst
260 func getMissingLocators(srcLocators, dstLocators map[string]bool) []string {
261 var missingLocators []string
262 for locator := range srcLocators {
263 if _, ok := dstLocators[locator]; !ok {
264 missingLocators = append(missingLocators, locator)
267 return missingLocators
270 // Copy blocks from src to dst; only those that are missing in dst are copied
271 func copyBlocksToDst(toBeCopied []string, kcSrc, kcDst *keepclient.KeepClient, srcBlobSignatureTTL time.Duration, blobSigningKey string) error {
272 total := len(toBeCopied)
274 startedAt := time.Now()
275 for done, locator := range toBeCopied {
277 log.Printf("Copying data block %d of %d (%.2f%% done): %v", done+1, total,
278 float64(done)/float64(total)*100, locator)
280 timePerBlock := time.Since(startedAt) / time.Duration(done)
281 log.Printf("Copying data block %d of %d (%.2f%% done, %v est. time remaining): %v", done+1, total,
282 float64(done)/float64(total)*100, timePerBlock*time.Duration(total-done), locator)
285 getLocator := locator
286 expiresAt := time.Now().AddDate(0, 0, 1)
287 if blobSigningKey != "" {
288 getLocator = keepclient.SignLocator(getLocator, kcSrc.Arvados.ApiToken, expiresAt, srcBlobSignatureTTL, []byte(blobSigningKey))
291 reader, len, _, err := kcSrc.Get(getLocator)
293 return fmt.Errorf("Error getting block: %v %v", locator, err)
296 _, _, err = kcDst.PutHR(getLocator[:32], reader, len)
298 return fmt.Errorf("Error copying data block: %v %v", locator, err)
302 log.Printf("Successfully copied to destination %d blocks.", total)