Matika版OpenStack伪生产环境部署-前期配置

目前matika版发布不久，目前还存在很多问题，在安装和使用过程中会遇到很多问题。但个人能力有限，希望在以后慢慢解决使用过程的问题。

本系列文章主要描述OpenStack的Matika版在物理机上安装的过程。在部署OpenStack过程中自己也在尝试使用编写脚本进行部署，有很多不足之处，望见谅！

在M版中，Controller节点和Neutron节点合并在Controller节点，根据现实需求单独划分节点安装Neutron没有必要性，所以将网络节点和控制器节点都安装配置在Controller节点。在前面安装部署Juno版中，没有使用到对象存储，所以在此也没有安装对象存储节点。

基本硬件配置：

管理监控节点： 1个 内存 8G 存储 600G 网卡端口 2个

Controller: 2个 内存 32G*2 存储 300G*4 网卡端口 4个 CPU 2个(24核)

Compute: 12个 内存 128G*4 64G*8 存储 900G 网卡端口 2个CPU 2个(24核)

Block Storage： 2个 内存 32G*2 存储 1.2T*2 网卡端口 2个 CPU 2个(24核)

网络模式选择Self-Service Networks，主机网络层逻辑图：



在各个节点上需要安装的组件和服务：

关闭防火墙和Selinux

systemctl stop firewalld.service
systemctl disable firewalld.service
sed -i "s/SELINUX=enforcing/SELINUX=disabled/" /etc/selinux/config
setenforce 0

配置/etc/hosts

为了方便配置，在每个节点都执行：

cat >> /etc/hosts << OFF
10.0.0.10    controller
10.0.0.11    controller1
10.0.0.12    controller2
10.0.0.31    compute01
10.0.0.32    compute02
10.0.0.33    compute03
10.0.0.34    compute04
10.0.0.35    compute05
10.0.0.36    compute06
10.0.0.37    compute07
10.0.0.38    compute08
10.0.0.39    compute09
10.0.0.40    compute10
10.0.0.41    compute11
10.0.0.42    compute12
10.0.0.51    cinder1
10.0.0.52    cinder2
OFF

配置NTP

目前CentOS7已经开始放弃使用ntp改为使用chrony

NTP服务器

设置本地时间同步：127.127.1.0，删除其他时间服务节点

yum install -y ntp
sed -i "s/server 0.centos.pool.ntp.org iburst/server 127.127.1.0 iburst/" /etc/ntp.conf
sed -i "22,24d" /etc/ntp.conf
systemctl disable chronyd.service
systemctl stop chronyd.service
systemctl enable ntpd.service
systemctl start ntpd.service

配置OpenStack包源

在所有Controller节点

yum install -y https://rdoproject.org/repos/rdo-release.rpm yum upgrade -y
yum install -y python-openstackclient
yum install -y openstack-selinux

安装MariaDB galera

许多OpenStack服务使用SQL数据库存储信息。数据库一般运行在Controller节点或者单独的数据库节点上。为了实现数据库的高可用性，本次安装采用了MariaDB Galera集群，Galera集群完成数据库之间复制操作。Galera 集群是一个同步的多主机数据库集群，基于MySQL和InnoDB存储引擎。在所有Controller节点上配置MariaDB Galera包源，编辑/etc/yum.repos.d/mariadb.repo

[mariadb]
name = MariaDB
baseurl = http://yum.mariadb.org/10.0/centos7-amd64 gpgkey=https://yum.mariadb.org/RPM-GPG-KEY-MariaDB
gpgcheck=1
enabled=0

安装数据库：

yum --enablerepo=mariadb -y install MariaDB-Galera-server

Controller1节点

创建/etc/my.cnf.d/openstack.cnf

sed -i "/\[mysqld\]$/a character-set-server = utf8" /etc/my.cnf.d/openstack.cnf
sed -i "/\[mysqld\]$/a init-connect = 'SET NAMES utf8'" /etc/my.cnf.d/openstack.cnf
sed -i "/\[mysqld\]$/a collation-server = utf8_general_ci" /etc/my.cnf.d/openstack.cnf
sed -i "/\[mysqld\]$/a innodb_file_per_table" /etc/my.cnf.d/openstack.cnf
sed -i "/\[mysqld\]$/a default-storage-engine = innodb" /etc/my.cnf.d/openstack.cnf
sed -i "/\[mysqld\]$/a bind-address = 0.0.0.0" /etc/my.cnf.d/openstack.cnf
sed -i "/\[mysqld\]$/a max_connections = 10000" /etc/my.cnf.d/openstack.cnf
sed -i "/\[galera\]$/a wsrep_provider = /usr/lib64/galera/libgalera_smm.so " /etc/my.cnf.d/openstack.cnf
sed -i "/\[galera\]$/a wsrep_cluster_address = \"gcomm://10.0.0.11,10.0.0.12\" " /etc/my.cnf.d/openstack.cnf
sed -i "/\[galera\]$/awsrep_cluster_name= \"MariaDB_Cluster\" " /etc/my.cnf.d/openstack.cnf
sed -i "/\[galera\]$/a wsrep_node_address= \"10.0.0.11\" " /etc/my.cnf.d/openstack.cnf
sed -i "/\[galera\]$/a wsrep_sst_method = rsync  " /etc/my.cnf.d/openstack.cnf

启动数据库

/etc/rc.d/init.d/mysql bootstrap

设置数据库root用户密码

mysqladmin -u root password "SWPUcs406mariadb"

Controller2节点

创建/etc/my.cnf.d/openstack.cnf，与controller1节点内容相似，稍微修改为wsrep_node_address=”10.0.0.12”

启动数据库

systemctl start mysql

验证：

mysql -u root -pSWPUcs406mariadb
> show status like 'wsrep_%';

主要查看

Variable_name	Value
wsrep_local_state_comment	synced
wsrep_cluster_size	2
wsrep_incoming_addresses	10.0.0.11:3306,10.0.0.12:3306

安装RabbitMQ

OpenStack使用消息队列来协调操作和状态信息等服务。消息队列服务通常Controller节点或独立的节点上运行。OpenStack支持多种消息队列服务,包括RabbitMQ Qpid,ZeroMQ。在所有Controller节点上执行

yum install -y rabbitmq-server

Controller1节点

systemctl enable rabbitmq-server.service
systemctl start rabbitmq-server.service
systemctl stop rabbitmq-server.service

然后执行：

scp /var/lib/rabbitmq/.erlang.cookie controller2:/var/lib/rabbitmq/.erlang.cookie
chown rabbitmq:rabbitmq /var/lib/rabbitmq/.erlang.cookie
chmod 400 /var/lib/rabbitmq/.erlang.cookie

启动RabbitMQ

systemctl start rabbitmq-server.service
rabbitmqctl cluster_status

Controller2节点

chown rabbitmq:rabbitmq /var/lib/rabbitmq/.erlang.cookie
chmod 400 /var/lib/rabbitmq/.erlang.cookie
systemctl enable rabbitmq-server.service
systemctl start rabbitmq-server.service
rabbitmqctl stop_app
rabbitmqctl join_cluster --ram rabbit@controller1
rabbitmqctl start_app

Controller1节点

rabbitmqctl set_policy ha-all '^(?!amq\.).*' '{"ha-mode": "all"}'

创建一个openstack用户：

rabbitmqctl add_user openstack SWPUcs406rabbit
rabbitmqctl set_permissions openstack ".*" ".*" ".*"

安装Memcached

在所有Controller节点

yum install -y memcached python-memcached
systemctl enable memcached.service
systemctl start memcached.service

安装Pacemaker+Corosync

+----------------------+          |          +----------------------+
| [       Node01     ] |10.0.0.11 | 10.0.0.12| [       Node02     ] |
|   controller1        +----------+----------+   controller2        |
|                      |                     |                      |
+----------------------+                     +----------------------+

在Controller所有节点：

yum -y install pacemaker pcs
systemctl start pcsd.service
systemctl enable pcsd.service
passwd hacluster

在Controller1节点

建立授权

pcs cluster auth controller1 controller2
Username: hacluster
Password:
controller1: Authorized
controller2: Authorized

配置集群

pcs cluster setup --name ha_cluster controll1 controller2
Shutting down pacemaker/corosync services...
Redirecting to /bin/systemctl stop pacemaker.service
Redirecting to /bin/systemctl stop corosync.service
Killing any remaining services...
Removing all cluster configuration files...
controller1: Succeeded
controller2: Succeeded

启动集群服务

pcs cluster start --all
controller1: Starting Cluster...
controller2: Starting Cluster...

使集群可用

pcs cluster enable --all
controller1: Cluster Enabled
controller2: Cluster Enabled

显示集群状态

pcs status cluster
Cluster Status:
Last updated: Tue May 24 16:02:13 2016     Last change: Mon May 23 10:29:50 2016 by hacluster via crmd on controller2
Stack: corosync
Current DC: controller1 (version 1.1.13-10.el7_2.2-44eb2dd) - partition with quorum
2 nodes and 0 resources configured
Online: [ controller1 controller2 ]
PCSD Status:
controller1: Online
controller2: Online

配置虚拟IP

pcs resource create OpenStack-VIP ocf:heartbeat:IPaddr2
params ip="10.0.0.10" cidr_netmask="24" op monitor interval="30s"

文章引用：

1..http://docs.openstack.org/ha-guide/index.html

2.http://docs.openstack.org/mitaka/install-guide-rdo/index.html