kanyun的部署

• 1 全部节点配置
  • 1.1 安装zmq
  • 1.2 安装kanyun
• 2 控制节点配置
  • 2.1 安装配置cassandra
  • 2.2 启动和配置kanyun
• 3 计算节点配置
  • 3.1 启动和配置kanyun

1 全部节点

1.1 安装zmq

安装zerozmq(2.2.0)

1. yum install libtool autoconf automake
2. yum install uuid-devel
3. yum install libuuid-devel
4. wget http://download.zeromq.org/zeromq-2.2.0.tar.gz
5. tar zxvf zeromq-2.2.0.tar.gz & cd zeromq-2.2.0 & ./configure & make install
6. ldconfig

安装pyzmq

1. easy_install “pyzmq==2.2.0”
2. 如果出现缺少libzmq.so.1的情况，可能是链接库没有正确配置。

3. sudo /sbin/ldconfig -v

   grep libzmq查看是否有所需链接

4. 如果没有libzmq.so.1 -> libzmq.so.1.0.1，则修改/etc/ld.so.conf，在末尾增加/usr/local/lib，然后ldconfig -v

   grep libzmq即可

5. 如果出现缺少Python.h，则使用yum install python-devel来安装

1.2 安装kanyun

下载代码

kanyun原来的代码有bug，而且不能在python2.6环境下使用，下面是我改过的

cd /opt
git clone https://github.com/suyan/kanyun.git

修改内容：

• 兼容python2.6和2.7（python2.6需要安装OrderedDict）
• 修改获取内存信息方式和计算cpu使用率方式
• 修复一些bug

安装

python /opt/kanyun/setup.py install

创建日志目录

mkdir /var/log/kanyun

安装OrderedDict(python2.6)

easy_install OrderedDict

2 控制节点

2.1 安装配置cassandra

下载Cassandra

cd /opt
wget http://mirror.bit.edu.cn/apache/cassandra/0.8.10/apache-cassandra-0.8.10-bin.tar.gz

解压

tar zxvf apache-cassandra-0.8.10-bin.tar.gz

运行

/opt/apache-cassandra-0.8.10/bin/cassandra

创建数据库

通过命令打开数据库

/opt/apache-cassandra-0.8.10/bin/cassandra-cli -h 127.0.0.1

并执行下面命令

CREATE keyspace DATA;
USE DATA;

CREATE COLUMN family vmnetwork WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family cpu WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family mem_max WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family mem_free WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family nic_incoming WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family nic_outgoing WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family blk_read WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';
CREATE COLUMN family blk_write WITH column_type='Super' AND comparator='AsciiType' AND subcomparator='IntegerType' AND default_validation_class='AsciiType';

assume vmnetwork KEYS AS ascii;
assume cpu KEYS AS ascii;
assume mem_max KEYS AS ascii;
assume nic_incoming KEYS AS ascii;
assume nic_outgoing KEYS AS ascii;
assume blk_read KEYS AS ascii;
assume blk_write KEYS AS ascii;
assume mem_free KEYS AS ascii;       

安装pycassa支持

easy_install pycassa

如果出现The required version of distribute is not available

easy_install -U distribute

2.2 启动和配置kanyun

配置/etc/kanyun.conf

• 文件中host需要填写计算节点的ip
• sql_connection填写nova.conf中connection值

配置文件

[kanyun]
log: /var/log/kanyun/kanyun.log
[DEFAULT]
sql_connection: mysql://nova:novamysqlpassword@controller/nova
[server]
host: 172.19.9.1 
port: 5551
db_host: 127.0.0.1
log: /var/log/kanyun/kanyun-server.log
[api]
api_host: 172.19.9.1
api_port: 5552
db_host: 127.0.0.1
log: /var/log/kanyun/kanyun-api.log
[client]
api_host: 172.19.9.1
api_port: 5552
log: /var/log/kanyun/kanyun-client.log 

启动服务

python /opt/kanyun/bin/kanyun-server &
python /opt/kanyun/bin/kanyun-api &

3 计算节点

3.1 启动和配置kanyun

配置/etc/kanyun.conf

• 文件中api_host需要填写计算节点的ip
• sql_connection填写nova.conf中connection值
• id根据节点编号自己设置

配置文件

[kanyun]
log: /var/log/kanyun/kanyun.log
[DEFAULT]
sql_connection: mysql://nova:novamysqlpassword@controller/nova
[worker]
id: worker1
worker_timeout: 60
dataserver_host: 172.19.9.1
dataserver_port: 5551
log: /var/log/kanyun/kanyun-worker.log
[client]
api_host: 172.19.9.1
api_port: 5552
log: /var/log/kanyun/kanyun-client.log

启动服务

python /opt/kanyun/bin/kanyun-worker &