zookeeper的伪集群搭建模式

所谓伪集群就是在单机模拟集群模式zookeeper的运行。

配置文件

下面是我配置的伪集群分布模式，分别通过zoo1.cfg、zoo2.cfg、zoo3.cfg来模拟三台机器的zookeeper集群:

zoo1.cfg内容如下:

# The number of milliseconds of each tick
tickTime=2000

# The number of ticks that the initial
# synchronization phase can take
initLimit=10

# The number of ticks that can pass between
# sending a request and getting an acknowledgement
syncLimit=5

# the directory where the snapshot is stored.
# do not use /tmp for storage, /tmp here is just
# example sakes.
dataDir=/usr/local/zookeeper-3.4.9/data/data_1
dataLogDir=/usr/local/zookeeper-3.4.9/dataLog/dataLog_1

# the port at which the clients will connect
clientPort=2181

server.0=localhost:2287:3387
server.1=localhost:2288:3388
server.2=localhost:2289:3389

zoo2.cfg内容如下

# The number of milliseconds of each tick
tickTime=2000

# The number of ticks that the initial
# synchronization phase can take
initLimit=10

# The number of ticks that can pass between
# sending a request and getting an acknowledgement
syncLimit=5

# the directory where the snapshot is stored.
# do not use /tmp for storage, /tmp here is just
# example sakes.
dataDir=/usr/local/zookeeper-3.4.9/data/data_2
dataLogDir=/usr/local/zookeeper-3.4.9/dataLog/dataLog_2

# the port at which the clients will connect
clientPort=2182

server.0=localhost:2287:3387
server.1=localhost:2288:3388
server.2=localhost:2289:3389

zoo3.cfg内容如下:

# The number of milliseconds of each tick
tickTime=2000

# The number of ticks that the initial
# synchronization phase can take
initLimit=10

# The number of ticks that can pass between
# sending a request and getting an acknowledgement
syncLimit=5

# the directory where the snapshot is stored.
# do not use /tmp for storage, /tmp here is just
# example sakes.
dataDir=/usr/local/zookeeper-3.4.9/data/data_3
dataLogDir=/usr/local/zookeeper-3.4.9/dataLog/dataLog_3

# the port at which the clients will connect
clientPort=2183

server.0=localhost:2287:3387
server.1=localhost:2288:3388
server.2=localhost:2289:3389

配置完成上述三个文件后，还需在dataDir对应路径下添加myid文件，内容对应server.n的n值:

echo 1 > data_1/myid
echo 2 > data_2/myid
echo 0 > data_3/myid

启动

在集群为分布式下，我们只有一台机器，但是要运行三个Zookeeper实例。此时，如果在使用单机模式的启动命令是行不通的。此时，只要通过下面三条命令就能运行前面所配置的Zookeeper服务。如下所示：

./zkServer.sh start zoo1.cfg
./zkServer.sh start zoo2.cfg
./zkServer.sh start zoo3.cfg

在运行完第一条指令之后，会出现一些错误异常，产生异常信息的原因是由于Zookeeper 服务的每个实例都拥有全局配置信息，他们在启动的时候会随时随地的进行Leader选举操作。此时，第一个启动的Zookeeper需要和另外两个 Zookeeper实例进行通信。但是，另外两个Zookeeper实例还没有启动起来，因此就产生了这的异样信息。我们直接将其忽略即可，待把图中“2 号”和“3号”Zookeeper实例启动起来之后，相应的异常信息自然会消失。此时，可以通过下面三条命令，来查询：

./zkServer.sh status zoo1.cfg
./zkServer.sh status zoo2.cfg
./zkServer.sh status zoo3.cfg

结果如下: