forked from vitessio/vitess
-
Notifications
You must be signed in to change notification settings - Fork 1
/
lock.go
197 lines (173 loc) · 6.51 KB
/
lock.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
// Copyright 2014, Google Inc. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package etcd2topo
import (
"flag"
"fmt"
"path"
"strconv"
"github.com/coreos/etcd/clientv3"
"github.com/coreos/etcd/mvcc/mvccpb"
log "github.com/golang/glog"
"golang.org/x/net/context"
"github.com/youtube/vitess/go/vt/topo"
)
var (
leaseTTL = flag.Int("topo_etcd_lease_ttl", 30, "Lease TTL for locks and master election. The client will use KeepAlive to keep the lease going.")
)
// newUniqueEphemeralKV creates a new file in the provided directory.
// It is linked to the Lease.
// Errors returned are converted to topo errors.
func (s *Server) newUniqueEphemeralKV(ctx context.Context, leaseID clientv3.LeaseID, nodePath string, contents string) (string, int64, error) {
// Use the lease ID as the file name, so it's guaranteed unique.
newKey := fmt.Sprintf("%v/%v", nodePath, leaseID)
// Only create a new file if it doesn't exist already
// (version = 0), to avoid two processes using the
// same file name. Since we use the lease ID, this should never happen.
txnresp, err := s.global.cli.Txn(ctx).
If(clientv3.Compare(clientv3.Version(newKey), "=", 0)).
Then(clientv3.OpPut(newKey, contents, clientv3.WithLease(leaseID))).
Commit()
if err != nil {
if err == context.Canceled || err == context.DeadlineExceeded {
// Our context was canceled as we were sending
// a creation request. We don't know if it
// succeeded or not. In any case, let's try to
// delete the node, so we don't leave an orphan
// node behind for *leaseTTL time.
s.global.cli.Delete(context.Background(), newKey)
}
return "", 0, convertError(err)
}
if !txnresp.Succeeded {
// The key already exists, that should not happen.
return "", 0, ErrBadResponse
}
// The key was created.
return newKey, txnresp.Header.Revision, nil
}
// waitOnLastRev waits on all revisions of the files in the provided
// directory that have revisions smaller than the provided revision.
// It returns true only if there is no more other older files.
func (s *Server) waitOnLastRev(ctx context.Context, nodePath string, revision int64) (bool, error) {
// Get the keys that are blocking us, if any.
opts := append(clientv3.WithLastRev(), clientv3.WithMaxModRev(revision-1))
lastKey, err := s.global.cli.Get(ctx, nodePath+"/", opts...)
if err != nil {
return false, convertError(err)
}
if len(lastKey.Kvs) == 0 {
// No older key, we're done waiting.
return true, nil
}
// Wait for release on blocking key. Cancel the watch when we
// exit this function.
key := string(lastKey.Kvs[0].Key)
ctx, cancel := context.WithCancel(ctx)
defer cancel()
wc := s.global.cli.Watch(ctx, key, clientv3.WithRev(revision))
if wc == nil {
return false, fmt.Errorf("Watch failed")
}
select {
case <-ctx.Done():
return false, convertError(ctx.Err())
case wresp := <-wc:
for _, ev := range wresp.Events {
if ev.Type == mvccpb.DELETE {
// There might still be older keys,
// but not this one.
return false, nil
}
}
}
// The Watch stopped, we're not sure if there are more items.
return false, nil
}
func (s *Server) lock(ctx context.Context, nodePath string, contents string) (string, error) {
// Get a lease, set its KeepAlive.
lease, err := s.global.cli.Grant(ctx, int64(*leaseTTL))
if err != nil {
return "", convertError(err)
}
leaseKA, err := s.global.cli.KeepAlive(ctx, lease.ID)
if err != nil {
return "", convertError(err)
}
go func() {
for range leaseKA {
}
}()
// Create an ephemeral node in the locks directory.
_, revision, err := s.newUniqueEphemeralKV(ctx, lease.ID, nodePath, contents)
if err != nil {
return "", err
}
key := path.Join(nodePath, fmt.Sprintf("%v", lease.ID))
// Wait until all older nodes in the locks directory are gone.
for {
done, err := s.waitOnLastRev(ctx, nodePath, revision)
if err != nil {
// We had an error waiting on the last node.
// Revoke our lease, this will delete the file.
if _, rerr := s.global.cli.Revoke(context.Background(), lease.ID); rerr != nil {
log.Warningf("Revoke(%d) failed, may have left %v behind: %v", lease.ID, key, rerr)
}
return "", err
}
if done {
// No more older nodes, we're it!
return key, nil
}
}
}
// unlock releases a lock acquired by lock() on the given directory.
// The string returned by lock() should be passed as the actionPath.
func (s *Server) unlock(ctx context.Context, dirPath, actionPath string) error {
leaseIDStr := path.Base(actionPath)
// Sanity check.
if checkPath := path.Join(dirPath, leaseIDStr); checkPath != actionPath {
return fmt.Errorf("unlock: actionPath doesn't match directory being unlocked: %q != %q", actionPath, checkPath)
}
i, err := strconv.ParseInt(leaseIDStr, 10, 64)
if err != nil {
return fmt.Errorf("unlock: invalid leaseID %v: %v", leaseIDStr, err)
}
leaseID := clientv3.LeaseID(i)
// Revoke the lease, will delete the node.
_, err = s.global.cli.Revoke(ctx, leaseID)
if err != nil {
return convertError(err)
}
return nil
}
// LockKeyspaceForAction implements topo.Server.
func (s *Server) LockKeyspaceForAction(ctx context.Context, keyspace, contents string) (string, error) {
// Check the keyspace exists first.
keyspacePath := path.Join(keyspacesPath, keyspace, topo.KeyspaceFile)
_, _, err := s.Get(ctx, topo.GlobalCell, keyspacePath)
if err != nil {
return "", err
}
return s.lock(ctx, path.Join(s.global.root, keyspacesPath, keyspace, locksPath), contents)
}
// UnlockKeyspaceForAction implements topo.Server.
func (s *Server) UnlockKeyspaceForAction(ctx context.Context, keyspace, actionPath, results string) error {
log.Infof("results of %v: %v", actionPath, results)
return s.unlock(ctx, path.Join(s.global.root, keyspacesPath, keyspace, locksPath), actionPath)
}
// LockShardForAction implements topo.Server.
func (s *Server) LockShardForAction(ctx context.Context, keyspace, shard, contents string) (string, error) {
shardPath := path.Join(keyspacesPath, keyspace, shardsPath, shard, topo.ShardFile)
_, _, err := s.Get(ctx, topo.GlobalCell, shardPath)
if err != nil {
return "", err
}
return s.lock(ctx, path.Join(s.global.root, keyspacesPath, keyspace, shardsPath, shard, locksPath), contents)
}
// UnlockShardForAction implements topo.Server.
func (s *Server) UnlockShardForAction(ctx context.Context, keyspace, shard, actionPath, results string) error {
log.Infof("results of %v: %v", actionPath, results)
return s.unlock(ctx, path.Join(s.global.root, keyspacesPath, keyspace, shardsPath, shard, locksPath), actionPath)
}