Zookeeper集群安装

Zookeeper集群安装

前提:准备3台机器，ip分别是

172.16.97.11

172.16.97.12

172.16.120.4

1. 下载zookeeper, http://zookeeper.apache.org/releases.html , 我下载的最新的3.4.5版

2. 下面操作在3台机器上都要做:

   sudo tar xvf zookeeper-3.4.5.tar.gz -C /opt/
   sudo mkdir -p /var/lib/zookeeper
   sudo mkdir -p /var/log/zookeeper
   cd /opt/zookeeper-3.4.5
   cp conf/zoo_sample.cfg conf/zoo.cfg
   
   vim conf/zoo.cfg并修改下面几行
   
       dataDir=/var/lib/zookeeper
       dataLogDir=/var/log/zookeeper
       server.1=172.16.97.11:2888:3888
       server.2=172.16.97.12:2888:3888
       server.3=172.16.120.4:2888:3888

3. 创建myid

in 172.16.97.11

echo “1” > /var/lib/zookeeper/myid

sudo bin/zkServer.sh start

in 172.16.97.12

echo “2” > /var/lib/zookeeper/myid

sudo bin/zkServer.sh start

in 172.16.120.4

echo “3” > /var/lib/Zookeeper/myid

sudo bin/zkServer.sh start

4. connect zookeeper

   bin/zkCli.sh -server 172.16.97.11:2181

###########################################################################

通过zookeeper来配置程序并自动重启

test.py

import os
#from kazoo.client import KazooClient
import zc.zk

zk = zc.zk.ZooKeeper('172.16.97.11:2181')
pid = os.getpid()
flag = False

@zk.properties('/new_test')
def data_update(data):
    global flag
    if flag:
        os.kill(pid, 1)
        #print("Version: %s, data: %s" % (stat.version, data.decode("utf-8")))
    else:
        flag = True

def application(environ, start_response):
    start_response('200 OK', [('Content-Type', 'text/plain')])
    yield 'Blocking on read(), pid is %s \n' % pid   
    if not os.path.exists('/tmp/my-fifo'):
        os.mkfifo('/tmp/my-fifo')
    fd = open('/tmp/my-fifo', 'r').read(1)
    yield 'Read done\n'

gunicorn -w 2 test:application

然后从python命令行：

In [1]: import zc.zk

In [2]: zk = zc.zk.ZooKeeper('172.16.97.11:2181')

In [3]: data = zk.properties('/new_test')

In [4]: data.update(abcd=12345)

可以从gunicorn看到进程重启了。

ps. 目前有个问题是如果gunicorn -k gevent -w 2 test:application启动的话，不能实现自动重启， 待fix