snapshot_sender.go 4.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168
  1. // Copyright 2015 The etcd Authors
  2. //
  3. // Licensed under the Apache License, Version 2.0 (the "License");
  4. // you may not use this file except in compliance with the License.
  5. // You may obtain a copy of the License at
  6. //
  7. // http://www.apache.org/licenses/LICENSE-2.0
  8. //
  9. // Unless required by applicable law or agreed to in writing, software
  10. // distributed under the License is distributed on an "AS IS" BASIS,
  11. // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  12. // See the License for the specific language governing permissions and
  13. // limitations under the License.
  14. package rafthttp
  15. import (
  16. "bytes"
  17. "context"
  18. "io"
  19. "io/ioutil"
  20. "net/http"
  21. "time"
  22. "github.com/coreos/etcd/pkg/httputil"
  23. pioutil "github.com/coreos/etcd/pkg/ioutil"
  24. "github.com/coreos/etcd/pkg/types"
  25. "github.com/coreos/etcd/raft"
  26. "github.com/coreos/etcd/snap"
  27. )
  28. var (
  29. // timeout for reading snapshot response body
  30. snapResponseReadTimeout = 5 * time.Second
  31. )
  32. type snapshotSender struct {
  33. from, to types.ID
  34. cid types.ID
  35. tr *Transport
  36. picker *urlPicker
  37. status *peerStatus
  38. r Raft
  39. errorc chan error
  40. stopc chan struct{}
  41. }
  42. func newSnapshotSender(tr *Transport, picker *urlPicker, to types.ID, status *peerStatus) *snapshotSender {
  43. return &snapshotSender{
  44. from: tr.ID,
  45. to: to,
  46. cid: tr.ClusterID,
  47. tr: tr,
  48. picker: picker,
  49. status: status,
  50. r: tr.Raft,
  51. errorc: tr.ErrorC,
  52. stopc: make(chan struct{}),
  53. }
  54. }
  55. func (s *snapshotSender) stop() { close(s.stopc) }
  56. func (s *snapshotSender) send(merged snap.Message) {
  57. start := time.Now()
  58. m := merged.Message
  59. to := types.ID(m.To).String()
  60. body := createSnapBody(merged)
  61. defer body.Close()
  62. u := s.picker.pick()
  63. req := createPostRequest(u, RaftSnapshotPrefix, body, "application/octet-stream", s.tr.URLs, s.from, s.cid)
  64. plog.Infof("start to send database snapshot [index: %d, to %s]...", m.Snapshot.Metadata.Index, types.ID(m.To))
  65. snapshotSendInflights.WithLabelValues(to).Inc()
  66. defer func() {
  67. snapshotSendInflights.WithLabelValues(to).Dec()
  68. }()
  69. err := s.post(req)
  70. defer merged.CloseWithError(err)
  71. if err != nil {
  72. plog.Warningf("database snapshot [index: %d, to: %s] failed to be sent out (%v)", m.Snapshot.Metadata.Index, types.ID(m.To), err)
  73. // errMemberRemoved is a critical error since a removed member should
  74. // always be stopped. So we use reportCriticalError to report it to errorc.
  75. if err == errMemberRemoved {
  76. reportCriticalError(err, s.errorc)
  77. }
  78. s.picker.unreachable(u)
  79. s.status.deactivate(failureType{source: sendSnap, action: "post"}, err.Error())
  80. s.r.ReportUnreachable(m.To)
  81. // report SnapshotFailure to raft state machine. After raft state
  82. // machine knows about it, it would pause a while and retry sending
  83. // new snapshot message.
  84. s.r.ReportSnapshot(m.To, raft.SnapshotFailure)
  85. sentFailures.WithLabelValues(to).Inc()
  86. snapshotSendFailures.WithLabelValues(to).Inc()
  87. return
  88. }
  89. s.status.activate()
  90. s.r.ReportSnapshot(m.To, raft.SnapshotFinish)
  91. plog.Infof("database snapshot [index: %d, to: %s] sent out successfully", m.Snapshot.Metadata.Index, types.ID(m.To))
  92. sentBytes.WithLabelValues(to).Add(float64(merged.TotalSize))
  93. snapshotSend.WithLabelValues(to).Inc()
  94. snapshotSendSeconds.WithLabelValues(to).Observe(time.Since(start).Seconds())
  95. }
  96. // post posts the given request.
  97. // It returns nil when request is sent out and processed successfully.
  98. func (s *snapshotSender) post(req *http.Request) (err error) {
  99. ctx, cancel := context.WithCancel(context.Background())
  100. req = req.WithContext(ctx)
  101. defer cancel()
  102. type responseAndError struct {
  103. resp *http.Response
  104. body []byte
  105. err error
  106. }
  107. result := make(chan responseAndError, 1)
  108. go func() {
  109. resp, err := s.tr.pipelineRt.RoundTrip(req)
  110. if err != nil {
  111. result <- responseAndError{resp, nil, err}
  112. return
  113. }
  114. // close the response body when timeouts.
  115. // prevents from reading the body forever when the other side dies right after
  116. // successfully receives the request body.
  117. time.AfterFunc(snapResponseReadTimeout, func() { httputil.GracefulClose(resp) })
  118. body, err := ioutil.ReadAll(resp.Body)
  119. result <- responseAndError{resp, body, err}
  120. }()
  121. select {
  122. case <-s.stopc:
  123. return errStopped
  124. case r := <-result:
  125. if r.err != nil {
  126. return r.err
  127. }
  128. return checkPostResponse(r.resp, r.body, req, s.to)
  129. }
  130. }
  131. func createSnapBody(merged snap.Message) io.ReadCloser {
  132. buf := new(bytes.Buffer)
  133. enc := &messageEncoder{w: buf}
  134. // encode raft message
  135. if err := enc.encode(&merged.Message); err != nil {
  136. plog.Panicf("encode message error (%v)", err)
  137. }
  138. return &pioutil.ReaderAndCloser{
  139. Reader: io.MultiReader(buf, merged.ReadCloser),
  140. Closer: merged.ReadCloser,
  141. }
  142. }