1 /* Deals with getting Keep Server blocks from API Server and Keep Servers. */
9 //"git.curoverse.com/arvados.git/sdk/go/keepclient"
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/manifest"
13 "git.curoverse.com/arvados.git/sdk/go/util"
22 type ServerAddress struct {
23 Host string `json:"service_host"`
24 Port int `json:"service_port"`
27 // Info about a particular block returned by the server
28 type BlockInfo struct {
29 Digest blockdigest.BlockDigest
31 Mtime int // TODO(misha): Replace this with a timestamp.
34 // Info about a specified block given by a server
35 type BlockServerInfo struct {
38 Mtime int // TODO(misha): Replace this with a timestamp.
41 type ServerContents struct {
42 BlockDigestToInfo map[blockdigest.BlockDigest]BlockInfo
45 type ServerResponse struct {
47 Contents ServerContents
50 type ReadServers struct {
52 KeepServerIndexToAddress []ServerAddress
53 KeepServerAddressToIndex map[ServerAddress]int
54 ServerToContents map[ServerAddress]ServerContents
55 BlockToServers map[blockdigest.BlockDigest][]BlockServerInfo
56 BlockReplicationCounts map[int]int
59 type GetKeepServersParams struct {
60 Client arvadosclient.ArvadosClient
64 type KeepServiceList struct {
65 ItemsAvailable int `json:"items_available"`
66 KeepServers []ServerAddress `json:"items"`
69 // Methods to implement util.SdkListResponse Interface
70 func (k KeepServiceList) NumItemsAvailable() (numAvailable int, err error) {
71 return k.ItemsAvailable, nil
74 func (k KeepServiceList) NumItemsContained() (numContained int, err error) {
75 return len(k.KeepServers), nil
79 // Don't access the token directly, use getDataManagerToken() to
80 // make sure it's been read.
81 dataManagerToken string
82 dataManagerTokenFile string
83 dataManagerTokenFileReadOnce sync.Once
87 flag.StringVar(&dataManagerTokenFile,
88 "data-manager-token-file",
90 "File with the API token we should use to contact keep servers.")
93 func getDataManagerToken() (string) {
94 readDataManagerToken := func () {
95 if dataManagerTokenFile == "" {
97 log.Fatalf("Data Manager Token needed, but data manager token file not specified.")
99 rawRead, err := ioutil.ReadFile(dataManagerTokenFile)
101 log.Fatalf("Unexpected error reading token file %s: %v",
102 dataManagerTokenFile,
105 dataManagerToken = strings.TrimSpace(string(rawRead))
109 dataManagerTokenFileReadOnce.Do(readDataManagerToken)
110 return dataManagerToken
113 func GetKeepServersAndSummarize(params GetKeepServersParams) (results ReadServers) {
114 results = GetKeepServers(params)
115 log.Printf("Returned %d keep disks", len(results.ServerToContents))
117 ComputeBlockReplicationCounts(&results)
118 log.Printf("Replication level distribution: %v",
119 results.BlockReplicationCounts)
124 func GetKeepServers(params GetKeepServersParams) (results ReadServers) {
125 if ¶ms.Client == nil {
126 log.Fatalf("params.Client passed to GetKeepServers() should " +
127 "contain a valid ArvadosClient, but instead it is nil.")
130 sdkParams := arvadosclient.Dict{}
131 if params.Limit > 0 {
132 sdkParams["limit"] = params.Limit
135 var sdkResponse KeepServiceList
136 err := params.Client.Call("GET", "keep_services", "", "accessible", sdkParams, &sdkResponse)
139 log.Fatalf("Error requesting keep disks from API server: %v", err)
142 log.Printf("Received keep services list: %v", sdkResponse)
145 var numReceived, numAvailable int
146 results.ReadAllServers, numReceived, numAvailable =
147 util.ContainsAllAvailableItems(sdkResponse)
149 if (!results.ReadAllServers) {
150 log.Printf("ERROR: Did not receive all keep server addresses.")
152 log.Printf("Received %d of %d available keep server addresses.",
157 results.KeepServerIndexToAddress = sdkResponse.KeepServers
158 results.KeepServerAddressToIndex = make(map[ServerAddress]int)
159 for i, address := range results.KeepServerIndexToAddress {
160 results.KeepServerAddressToIndex[address] = i
163 log.Printf("Got Server Addresses: %v", results)
165 // This is safe for concurrent use
166 client := http.Client{}
168 // Send off all the index requests concurrently
169 responseChan := make(chan ServerResponse)
170 for _, keepServer := range sdkResponse.KeepServers {
171 go GetServerContents(keepServer, client, responseChan)
174 results.ServerToContents = make(map[ServerAddress]ServerContents)
175 results.BlockToServers = make(map[blockdigest.BlockDigest][]BlockServerInfo)
177 // Read all the responses
178 for i := range sdkResponse.KeepServers {
179 _ = i // Here to prevent go from complaining.
180 response := <- responseChan
181 log.Printf("Received channel response from %v containing %d files",
183 len(response.Contents.BlockDigestToInfo))
184 results.ServerToContents[response.Address] = response.Contents
185 serverIndex := results.KeepServerAddressToIndex[response.Address]
186 for _, blockInfo := range response.Contents.BlockDigestToInfo {
187 results.BlockToServers[blockInfo.Digest] = append(
188 results.BlockToServers[blockInfo.Digest],
189 BlockServerInfo{ServerIndex: serverIndex,
190 Size: blockInfo.Size,
191 Mtime: blockInfo.Mtime})
197 // TODO(misha): Break this function apart into smaller, easier to
198 // understand functions.
199 func GetServerContents(keepServer ServerAddress,
201 responseChan chan<- ServerResponse) () {
202 // Create and send request.
203 url := fmt.Sprintf("http://%s:%d/index", keepServer.Host, keepServer.Port)
204 log.Println("About to fetch keep server contents from " + url)
206 req, err := http.NewRequest("GET", url, nil)
208 log.Fatalf("Error building http request for %s: %v", url, err)
211 req.Header.Add("Authorization",
212 fmt.Sprintf("OAuth2 %s", getDataManagerToken()))
214 resp, err := client.Do(req)
216 log.Fatalf("Error fetching %s: %v", url, err)
220 if resp.StatusCode != 200 {
221 log.Fatalf("Received error code %d in response to request for %s: %s",
222 resp.StatusCode, url, resp.Status)
225 response := ServerResponse{}
226 response.Address = keepServer
227 response.Contents.BlockDigestToInfo =
228 make(map[blockdigest.BlockDigest]BlockInfo)
229 scanner := bufio.NewScanner(resp.Body)
230 numLines, numDuplicates, numSizeDisagreements := 0, 0, 0
233 blockInfo, err := parseBlockInfoFromIndexLine(scanner.Text())
235 log.Fatalf("Error parsing BlockInfo from index line received from %s: %v",
240 if storedBlock, ok := response.Contents.BlockDigestToInfo[blockInfo.Digest]; ok {
241 // This server returned multiple lines containing the same block digest.
243 if storedBlock.Size != blockInfo.Size {
244 numSizeDisagreements += 1
245 // TODO(misha): Consider failing here.
246 log.Printf("Saw different sizes for the same block on %s: %v %v",
251 // Keep the block that is bigger, or the block that's newer in
252 // the case of a size tie.
253 if storedBlock.Size < blockInfo.Size ||
254 (storedBlock.Size == blockInfo.Size &&
255 storedBlock.Mtime < blockInfo.Mtime) {
256 response.Contents.BlockDigestToInfo[blockInfo.Digest] = blockInfo
259 response.Contents.BlockDigestToInfo[blockInfo.Digest] = blockInfo
262 if err := scanner.Err(); err != nil {
263 log.Fatalf("Received error scanning response from %s: %v", url, err)
265 log.Printf("%s contained %d lines with %d duplicates with " +
266 "%d size disagreements",
270 numSizeDisagreements)
273 responseChan <- response
276 func parseBlockInfoFromIndexLine(indexLine string) (blockInfo BlockInfo, err error) {
277 tokens := strings.Fields(indexLine)
278 if len(tokens) != 2 {
279 err = fmt.Errorf("Expected 2 tokens per line but received a " +
280 "line containing %v instead.",
284 var locator manifest.BlockLocator
285 if locator, err = manifest.ParseBlockLocator(tokens[0]); err != nil {
288 if len(locator.Hints) > 0 {
289 err = fmt.Errorf("Block locator in index line should not contain hints " +
295 blockInfo.Mtime, err = strconv.Atoi(tokens[1])
299 blockInfo.Digest = locator.Digest
300 blockInfo.Size = locator.Size
304 func ComputeBlockReplicationCounts(readServers *ReadServers) {
305 readServers.BlockReplicationCounts = make(map[int]int)
306 for _, infos := range readServers.BlockToServers {
307 replication := len(infos)
308 readServers.BlockReplicationCounts[replication] += 1