sdk/go/keepclient/support.go

   1 package keepclient
   2
   3 import (
   4         "crypto/md5"
   5         "errors"
   6         "fmt"
   7         "git.curoverse.com/arvados.git/sdk/go/streamer"
   8         "io"
   9         "io/ioutil"
  10         "log"
  11         "net"
  12         "net/http"
  13         "strings"
  14         "time"
  15 )
  16
  17 type keepService struct {
  18         Uuid     string `json:"uuid"`
  19         Hostname string `json:"service_host"`
  20         Port     int    `json:"service_port"`
  21         SSL      bool   `json:"service_ssl_flag"`
  22         SvcType  string `json:"service_type"`
  23         ReadOnly bool   `json:"read_only"`
  24 }
  25
  26 // Md5String returns md5 hash for the bytes in the given string
  27 func Md5String(s string) string {
  28         return fmt.Sprintf("%x", md5.Sum([]byte(s)))
  29 }
  30
  31 // Set timeouts apply when connecting to keepproxy services (assumed to be over
  32 // the Internet).
  33 func (this *KeepClient) setClientSettingsProxy() {
  34         if this.Client.Timeout == 0 {
  35                 // Maximum time to wait for a complete response
  36                 this.Client.Timeout = 300 * time.Second
  37
  38                 // TCP and TLS connection settings
  39                 this.Client.Transport = &http.Transport{
  40                         Dial: (&net.Dialer{
  41                                 // The maximum time to wait to set up
  42                                 // the initial TCP connection.
  43                                 Timeout: 30 * time.Second,
  44
  45                                 // The TCP keep alive heartbeat
  46                                 // interval.
  47                                 KeepAlive: 120 * time.Second,
  48                         }).Dial,
  49
  50                         TLSHandshakeTimeout: 10 * time.Second,
  51                 }
  52         }
  53 }
  54
  55 // Set timeouts apply when connecting to keepstore services directly (assumed
  56 // to be on the local network).
  57 func (this *KeepClient) setClientSettingsDisk() {
  58         if this.Client.Timeout == 0 {
  59                 // Maximum time to wait for a complete response
  60                 this.Client.Timeout = 20 * time.Second
  61
  62                 // TCP and TLS connection timeouts
  63                 this.Client.Transport = &http.Transport{
  64                         Dial: (&net.Dialer{
  65                                 // The maximum time to wait to set up
  66                                 // the initial TCP connection.
  67                                 Timeout: 2 * time.Second,
  68
  69                                 // The TCP keep alive heartbeat
  70                                 // interval.
  71                                 KeepAlive: 180 * time.Second,
  72                         }).Dial,
  73
  74                         TLSHandshakeTimeout: 4 * time.Second,
  75                 }
  76         }
  77 }
  78
  79 // DiscoverKeepServers gets list of available keep services from api server
  80 func (this *KeepClient) DiscoverKeepServers() error {
  81         type svcList struct {
  82                 Items []keepService `json:"items"`
  83         }
  84         var m svcList
  85
  86         // Get keep services from api server
  87         err := this.Arvados.Call("GET", "keep_services", "", "accessible", nil, &m)
  88         if err != nil {
  89                 return err
  90         }
  91
  92         listed := make(map[string]bool)
  93         localRoots := make(map[string]string)
  94         gatewayRoots := make(map[string]string)
  95         writableLocalRoots := make(map[string]string)
  96
  97         // replicasPerService is 1 for disks; unknown or unlimited otherwise
  98         this.replicasPerService = 1
  99         this.Using_proxy = false
 100
 101         for _, service := range m.Items {
 102                 scheme := "http"
 103                 if service.SSL {
 104                         scheme = "https"
 105                 }
 106                 url := fmt.Sprintf("%s://%s:%d", scheme, service.Hostname, service.Port)
 107
 108                 // Skip duplicates
 109                 if listed[url] {
 110                         continue
 111                 }
 112                 listed[url] = true
 113
 114                 localRoots[service.Uuid] = url
 115                 if service.SvcType == "proxy" {
 116                         this.Using_proxy = true
 117                 }
 118
 119                 if service.ReadOnly == false {
 120                         writableLocalRoots[service.Uuid] = url
 121                         if service.SvcType != "disk" {
 122                                 this.replicasPerService = 0
 123                         }
 124                 }
 125
 126                 // Gateway services are only used when specified by
 127                 // UUID, so there's nothing to gain by filtering them
 128                 // by service type. Including all accessible services
 129                 // (gateway and otherwise) merely accommodates more
 130                 // service configurations.
 131                 gatewayRoots[service.Uuid] = url
 132         }
 133
 134         if this.Using_proxy {
 135                 this.setClientSettingsProxy()
 136         } else {
 137                 this.setClientSettingsDisk()
 138         }
 139
 140         this.SetServiceRoots(localRoots, writableLocalRoots, gatewayRoots)
 141         return nil
 142 }
 143
 144 type uploadStatus struct {
 145         err             error
 146         url             string
 147         statusCode      int
 148         replicas_stored int
 149         response        string
 150 }
 151
 152 func (this KeepClient) uploadToKeepServer(host string, hash string, body io.ReadCloser,
 153         upload_status chan<- uploadStatus, expectedLength int64, requestId string) {
 154
 155         var req *http.Request
 156         var err error
 157         var url = fmt.Sprintf("%s/%s", host, hash)
 158         if req, err = http.NewRequest("PUT", url, nil); err != nil {
 159                 log.Printf("[%v] Error creating request PUT %v error: %v", requestId, url, err.Error())
 160                 upload_status <- uploadStatus{err, url, 0, 0, ""}
 161                 body.Close()
 162                 return
 163         }
 164
 165         req.ContentLength = expectedLength
 166         if expectedLength > 0 {
 167                 // http.Client.Do will close the body ReadCloser when it is
 168                 // done with it.
 169                 req.Body = body
 170         } else {
 171                 // "For client requests, a value of 0 means unknown if Body is
 172                 // not nil."  In this case we do want the body to be empty, so
 173                 // don't set req.Body.  However, we still need to close the
 174                 // body ReadCloser.
 175                 body.Close()
 176         }
 177
 178         req.Header.Add("Authorization", fmt.Sprintf("OAuth2 %s", this.Arvados.ApiToken))
 179         req.Header.Add("Content-Type", "application/octet-stream")
 180         req.Header.Add(X_Keep_Desired_Replicas, fmt.Sprint(this.Want_replicas))
 181
 182         var resp *http.Response
 183         if resp, err = this.Client.Do(req); err != nil {
 184                 log.Printf("[%v] Upload failed %v error: %v", requestId, url, err.Error())
 185                 upload_status <- uploadStatus{err, url, 0, 0, ""}
 186                 return
 187         }
 188
 189         rep := 1
 190         if xr := resp.Header.Get(X_Keep_Replicas_Stored); xr != "" {
 191                 fmt.Sscanf(xr, "%d", &rep)
 192         }
 193
 194         defer resp.Body.Close()
 195         defer io.Copy(ioutil.Discard, resp.Body)
 196
 197         respbody, err2 := ioutil.ReadAll(&io.LimitedReader{resp.Body, 4096})
 198         response := strings.TrimSpace(string(respbody))
 199         if err2 != nil && err2 != io.EOF {
 200                 log.Printf("[%v] Upload %v error: %v response: %v", requestId, url, err2.Error(), response)
 201                 upload_status <- uploadStatus{err2, url, resp.StatusCode, rep, response}
 202         } else if resp.StatusCode == http.StatusOK {
 203                 log.Printf("[%v] Upload %v success", requestId, url)
 204                 upload_status <- uploadStatus{nil, url, resp.StatusCode, rep, response}
 205         } else {
 206                 log.Printf("[%v] Upload %v error: %v response: %v", requestId, url, resp.StatusCode, response)
 207                 upload_status <- uploadStatus{errors.New(resp.Status), url, resp.StatusCode, rep, response}
 208         }
 209 }
 210
 211 func (this KeepClient) putReplicas(
 212         hash string,
 213         tr *streamer.AsyncStream,
 214         expectedLength int64) (locator string, replicas int, err error) {
 215
 216         // Take the hash of locator and timestamp in order to identify this
 217         // specific transaction in log statements.
 218         requestId := fmt.Sprintf("%x", md5.Sum([]byte(locator+time.Now().String())))[0:8]
 219
 220         // Calculate the ordering for uploading to servers
 221         sv := NewRootSorter(this.WritableLocalRoots(), hash).GetSortedRoots()
 222
 223         // The next server to try contacting
 224         next_server := 0
 225
 226         // The number of active writers
 227         active := 0
 228
 229         // Used to communicate status from the upload goroutines
 230         upload_status := make(chan uploadStatus)
 231         defer close(upload_status)
 232
 233         // Desired number of replicas
 234         remaining_replicas := this.Want_replicas
 235
 236         replicasPerThread := this.replicasPerService
 237         if replicasPerThread < 1 {
 238                 // unlimited or unknown
 239                 replicasPerThread = remaining_replicas
 240         }
 241
 242         for remaining_replicas > 0 {
 243                 for active*replicasPerThread < remaining_replicas {
 244                         // Start some upload requests
 245                         if next_server < len(sv) {
 246                                 log.Printf("[%v] Begin upload %s to %s", requestId, hash, sv[next_server])
 247                                 go this.uploadToKeepServer(sv[next_server], hash, tr.MakeStreamReader(), upload_status, expectedLength, requestId)
 248                                 next_server += 1
 249                                 active += 1
 250                         } else {
 251                                 if active == 0 {
 252                                         return locator, (this.Want_replicas - remaining_replicas), InsufficientReplicasError
 253                                 } else {
 254                                         break
 255                                 }
 256                         }
 257                 }
 258                 log.Printf("[%v] Replicas remaining to write: %v active uploads: %v",
 259                         requestId, remaining_replicas, active)
 260
 261                 // Now wait for something to happen.
 262                 status := <-upload_status
 263                 active -= 1
 264
 265                 if status.statusCode == 200 {
 266                         // good news!
 267                         remaining_replicas -= status.replicas_stored
 268                         locator = status.response
 269                 }
 270         }
 271
 272         return locator, this.Want_replicas, nil
 273 }