// 连接etcd
import (
"github.com/coreos/etcd/clientv3"
"github.com/coreos/etcd/mvcc/mvccpb"
)
config = clientv3.Config{
Endpoints: []string{"127.0.0.1:2379"},
DialTimeout: 5 * time.Second,
}
client, err = clientv3.New(config)
kv = clientv3.NewKV(client)
// put
putResp, err = kv.Put(context.TODO(), "/cron/jobs/job1", "bye", clientv3.WithPrevKV())
putResp.Header.Revision
putResp.PrevKv.Value
// get
getResp, err = kv.Get(context.TODO(), "/cron/jobs/job1", /*clientv3.WithCountOnly()*/)
getResp.Kvs
getResp.Count
// get prefix
getResp, err = kv.Get(context.TODO(), "/cron/jobs/", clientv3.WithPrefix())
getResp.Kvs
// delete
delResp, err = kv.Delete(context.TODO(), "/cron/jobs/job1", clientv3.WithFromKey(), clientv3.WithLimit(2))
for _, kvpair = range delResp.PrevKvs {
fmt.Println("删除了:", string(kvpair.Key), string(kvpair.Value))
}
// lease: 关注key是否存在
lease = clientv3.NewLease(client)
leaseGrantResp, err = lease.Grant(context.TODO(), 10)
keepRespChan, err = lease.KeepAlive(context.TODO(), leaseGrantResp.id) //自动续约
putResp, err = kv.Put(context.TODO(), "/cron/lock/job1", "", clientv3.WithLease(leaseGrantResp.id))
putResp.Header.Revision
// watch: 关注key的变化
getResp, err = kv.Get(context.TODO(), "/cron/jobs/job7") //先get到当前值,监听后续变化
watchStartRevision = getResp.Header.Revision + 1
watcher = clientv3.NewWatcher(client)
ctx, cancelFunc := context.WithCancel(context.TODO())
time.AfterFunc(5 * time.Second, func() {
cancelFunc()
})
watchRespChan = watcher.Watch(ctx, "/cron/jobs/job7", clientv3.WithRev(watchStartRevision))
for watchResp = range watchRespChan {
for _, event = range watchResp.Events {
switch event.Type {
case mvccpb.PUT:
fmt.Println("修改为:", string(event.Kv.Value), "Revision:", event.Kv.CreateRevision, event.Kv.ModRevision)
case mvccpb.DELETE:
fmt.Println("删除了", "Revision:", event.Kv.ModRevision)
}
}
}
// op操作(类似条件一样的)
putOp = clientv3.OpPut("/cron/jobs/job8", "123123123")
opResp, err = kv.Do(context.TODO(), putOp)
// kv.Put
// kv.Get
// kv.Delete
opResp.Put().Header.Revision
getOp = clientv3.OpGet("/cron/jobs/job8")
opResp, err = kv.Do(context.TODO(), getOp)
opResp.Get().Kvs[0].ModRevision
opResp.Get().Kvs[0].Value
// 事务机制: 去锁资源, 判断lock目录key是否存在, 如果不存在则put进去, 标记抢锁成功, 如果存在则象征性get下, 标记强锁失败
lease = clientv3.NewLease(client) //创建租约, 自动续租, 拿着租约去抢占一个key
leaseGrantResp, err = lease.Grant(context.TODO(), 5)
defer lease.Revoke(context.TODO(), leaseGrantResp.id)
defer lease.Revoke(context.TODO(), leaseGrantResp.id)
keepRespChan, err = lease.KeepAlive(ctx, leaseId) //5秒后会取消自动续租
go func() {
for {
select {
case keepResp = <- keepRespChan:
if keepRespChan == nil {
fmt.Println("租约已经失效了")
goto END
} else { // 每秒会续租一次, 所以就会受到一次应答
fmt.Println("收到自动续租应答:", keepResp.ID)
}
}
}
END:
}()
kv = clientv3.NewKV(client) // if 不存在key, then 设置它, else 抢锁失败
txn = kv.Txn(context.TODO()) // 创建事务
//if条件成立,走then,否则走else
txn.If(clientv3.Compare(clientv3.CreateRevision("/cron/lock/job9"), "=", 0)).
Then(clientv3.OpPut("/cron/lock/job9", "xxx", clientv3.WithLease(leaseId))).
Else(clientv3.OpGet("/cron/lock/job9")) // 否则抢锁失败
txn.Commit()
txnResp.Succeeded //抢锁成功
理解事务: 一个curl的例子
// python2
// 已在etcd中创建了key1:0,A:"success",B:"failure"三个键值对
import json
import base64
import requests
URL = "http://127.0.0.1:2379/v3beta/kv/%s"
url = URL % "txn"
payload = {
"compare":[
{
"target": "VALUE",
"key":base64.b64encode("key1"),
'result': "EQUAL",
"value": base64.b64encode("0"),
},
],
"success":[
{
"requestRange":{
"key":base64.b64encode("A"),
}
}
],
"failure":[
{
"requestRange":{
"key":base64.b64encode("B"),
}
}
]
}
resp = requests.post(url,json=payload)
print json.dumps(resp.json(), indent=2)
所谓的支持事务锁: 里面的操作是原子的
// 发起转账视图
txn := etcd.Txn(ctx.TODO()).If(
v3.Compare(v3.ModRevision(from), “=”, fromKV.ModRevision), // 事务提交时,from账户余额没有没有变动
v3.Compare(v3.ModRevision(to), “=”, toKV.ModRevision)) // 事务提交时,to账户余额没有变动
txn = txn.Then(
OpPut(from, fromUint64(fromV - amount)), // 更新from账户余额
OpPut(to, fromUint64(toV - amount)) // 更新to账户余额
putresp, err := txn.Commit() // 提交事务
[etcd]分布式锁
etcd election特性使用场景——Master选举分析与实现
|
请发表评论