Give me a bunch of learning documents first , Easy to check later

Official website document address encyclopedia ：

OverView( summary )

http://zookeeper.apache.org/doc/r3.4.6/zookeeperOver.html

Getting Started( Get started )

http://zookeeper.apache.org/doc/r3.4.6/zookeeperStarted.html

Tutorial（ course ）

http://zookeeper.apache.org/doc/r3.4.6/zookeeperTutorial.html

Java Example(Java Example )

http://zookeeper.apache.org/doc/r3.4.6/javaExample.html

Programmer's Guide( Developer's Guide )

http://zookeeper.apache.org/doc/r3.4.6/zookeeperProgrammers.html

Recipes and Solutions( Skills and solutions )

http://zookeeper.apache.org/doc/r3.4.6/recipes.html

3.4.6 API online( On-line API Quick check )

http://zookeeper.apache.org/doc/r3.4.6/api/index.html

In addition, recommend Garden Friends sunddenly Of zookeeper series

http://www.cnblogs.com/sunddenly/category/620563.html

One 、 Installation and deployment

This paper simulates on a machine 3 individual zk server Cluster installation of

1.1  Download decompression

Unzip to 3 A catalog （ simulation 3 platform zk server）：

 /home/hadoop/zookeeper-1
/home/hadoop/zookeeper-2
/home/hadoop/zookeeper-3

1.2  Create each directory conf/zoo.cfg The configuration file  

/home/hadoop/zookeeper-1/conf/zoo.cfg The contents are as follows ：

tickTime=2000
initLimit=10
syncLimit=5
dataDir=/home/hadoop/tmp/zk1/data
dataLogDir=/home/hadoop/tmp/zk1/log
clientPort=2181
server.1=localhost:2287:3387
server.2=localhost:2288:3388
server.3=localhost:2289:3389

/home/hadoop/zookeeper-2/conf/zoo.cfg The contents are as follows ：

tickTime=2000
initLimit=10
syncLimit=5
dataDir=/home/hadoop/tmp/zk2/data
dataLogDir=/home/hadoop/tmp/zk2/log
clientPort=2182
server.1=localhost:2287:3387
server.2=localhost:2288:3388
server.3=localhost:2289:3389

/home/hadoop/zookeeper-3/conf/zoo.cfg The contents are as follows ：

tickTime=2000
initLimit=10
syncLimit=5
dataDir=/home/hadoop/tmp/zk3/data
dataLogDir=/home/hadoop/tmp/zk3/log
clientPort=2183
server.1=localhost:2287:3387
server.2=localhost:2288:3388
server.3=localhost:2289:3389

notes ： Because the cluster is simulated on one machine , So the port cannot be repeated , Here we use 2181~2183,2287~2289, as well as 3387~3389 Stagger each other . And each of them zk Of instance, You need to set up an independent data storage directory 、 Log storage directory , therefore dataDir、dataLogDir The directory corresponding to these two nodes , You need to manually create .

There is also a key setting , At every zk server Profile's dataDir Corresponding directory , You must create a file named myid The file of , The contents must be consistent with zoo.cfg in server.x Medium x identical , namely ：

/home/hadoop/tmp/zk1/data/myid is 1, Corresponding server.1 Medium 1
/home/hadoop/tmp/zk2/data/myid is 2, Corresponding server.2 Medium 2
/home/hadoop/tmp/zk3/data/myid is 3, Corresponding server.3 Medium 3

In production environment , The steps of distributed cluster deployment are basically the same as above , Just because each zk server Distributed in different machines , In the above configuration file localhost Replace with the real of each server Ip that will do . Distributed behind different machines , There are no port conflicts , You can make each server zk All use the same port , It's easier to manage in this way .

1.3  Start validation  

/home/hadoop/zookeeper-1/bin/zkServer.sh start
/home/hadoop/zookeeper-2/bin/zkServer.sh start
/home/hadoop/zookeeper-3/bin/zkServer.sh start

When enabled successfully , Input jps Look at the process

20351 ZooKeeperMain
20791 QuorumPeerMain
20822 QuorumPeerMain
20865 QuorumPeerMain

You should be able to see at least the above processes .

You can start the client test ：

bin/zkCli.sh -server localhost:2181

( notes ： If it's a remote connection , hold localhost Replace with the specified IP that will do )

After success , It should go to the prompt , Similar to the following :

[zk: localhost:2181(CONNECTED) 0]  

then , You can use some basic commands , such as ls ,create ,delete ,get To test （ About these orders , You can view the document ）, In particular, a very useful command rmr Used to recursively delete a node and all its children

Two 、java And zk Connection example for

2.1 maven Project pom.xml Add the following dependencies first

 <!--zk-->
<dependency>
<groupId>org.apache.zookeeper</groupId>
<artifactId>zookeeper</artifactId>
<version>3.4.6</version>
</dependency>

2.2  The most basic example program

2.3  And zk Cluster connectivity

zk One of the advantages of , It's high availability , The above code connects a single zk server, If this one server Hang up , Natural code will go wrong , in fact zk Of API With that in mind , Change the connection code to the following ：

 ZooKeeper zk = new ZooKeeper("172.28.20.102:2181,172.28.20.102:2182,172.28.20.102:2183", 300000, new DemoWatcher());// Connect zk server

  namely ：IP1:port1,IP2:port2,IP3:port3...   Just connect to the cluster in this way , As long as more than half zk server Still alive , Application is generally no problem . But there is also an extremely rare case , For example, when this line of code executes , Just finished initializing , Preparing to connect ip1 when , Because of network failure ip1 Corresponding server Hang up , Still report a mistake （ here ,zk It's too late to choose a new leader）, This problem is detailed in ：http://segmentfault.com/q/1010000002506725/a-1020000002507402, Refer to the practice of this article , Change to ：

 1 ZooKeeper zk = new ZooKeeper("172.28.20.102:2181,172.28.20.102:2182,172.28.20.102:2183", 300000, new DemoWatcher());// Connect zk server
2 if (!zk.getState().equals(ZooKeeper.States.CONNECTED)) {
3 while (true) {
4 if (zk.getState().equals(ZooKeeper.States.CONNECTED)) {
5 break;
6 }
7 try {
8 TimeUnit.SECONDS.sleep(5);
9 } catch (InterruptedException e) {
10 e.printStackTrace();
11 }
12 }
13 }

  But this code is too lengthy , Open source is recommended zkClient, Official address : https://github.com/sgroschupf/zkclient, The method is simple to use ：

 <!--zkclient-->
<dependency>
<groupId>com.github.sgroschupf</groupId>
<artifactId>zkclient</artifactId>
<version>0.1</version>
</dependency>

pom.xml Add this lump first , Then use ：

1 @Test
2 public void testZkClient() {
3 ZkClient zkClient = new ZkClient("172.28.20.102:2181,172.28.20.102:2182,172.28.20.102:2183");
4 String node = "/app2";
5 if (!zkClient.exists(node)) {
6 zkClient.createPersistent(node, "hello zk");
7 }
8 System.out.println(zkClient.readData(node));
9 }