centos安装rabbitmq_Openstack安装之基础环境及公共组件安装

1.1 环境介绍

1.1.1 操作系统及软件

• CentOS Linux release 8.2.2004 (Core)
• openstack train
• Ceph：Octopus

1.1.2 官网提供的逻辑架构图

1.1.3 本文所涉及的部署拓扑

• 部署说明：

(1)部署的拓扑取决于OpenStack的三层网络是采用Linuxbridge还是OVS来实现，不同的实现方式需要的网卡数量不尽相同。(2)建议服务器最少2张网络设计。External Network、Management Network和Ceph Cluster的Public Network可以复用一张网络。(3)LinuxBridge网络下，Tenant Network和Ceph Cluster的Cluster Network可以复用一张网络。(4)Ceph Cluster 建议2张网络分离，Public Network和Cluster Network走不同的网络。

• 组件部署说明

(1) controller节点运行keystone,glance,horizon,nova的管理组件，neutron的管理组件，cinder管理组件和ceph-mon节点。(2) 采用ceph作为存储，因此cinder volume部署在controller节点上，以便用pacemaker实现故障切换。(3) conpute节点运行nova-compute,neutron-linuxbridge-agent/neutron-ovs-agent,kvm,ceph-osd,ceph-mgr等。(4) Tenant Network可以在Tunel和Vxlan网络中二选一即可，实现guest。 (5) 本系列文档中，os之间通讯的网络，主要选用vxlan方式。(6) 本系列文档采用self-service-networks提供自助网络服务，provider networks 不支持专有网络，需要依靠外部基础设施提供3层路由与增值服务(LBaaS,FwaaS)。(7) 本系列文档前端采用pacemaker+haproxy做高可用。(8) 无状态的服务器，如xxx-api，采用active/active的模式运行；有状态的服务，如neutron-xxx-agent、cinder-volume等，建议采用active/passive的模式运行。自身具有集群机制的服务，如rabbitmq、memcached等采用本身的集群机制即可。

1.2 基础环境配置

1.2.1 hosts

127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4::1         localhost localhost.localdomain localhost6 localhost6.localdomain6#vip10.29.32.10     controller.sysit.cn  controller#controller10.29.32.7     controller1.sysit.cn controller110.29.32.8     controller2.sysit.cn controller210.29.32.9     controller3.sysit.cn controller3#compute10.29.32.11     compute1.sysit.cn compute110.29.32.12     compute2.sysit.cn compute210.29.32.13     compute3.sysit.cn compute3

1.2.2 操作通初始化

• 见本站CentOS8初始化脚本

1.2.3 ntp或chrony

• 见本网站其他文档，推荐chrony

1.2.4 设置openstack packages

#安装yum源dnf install -y centos-release-openstack-traindnf install https://www.rdoproject.org/repos/rdo-release.el8.rpmdnf upgrade -y#安装oepnstack-clientdnf install -y python3-openstackclient#selinux开启时需要安装openstack-selinux，这里已将seliux设置为默认关闭dnf install -y openstack-selinux

• 以上使用的源默认会使用mirrors.centos.org的源，安装时候会特别慢。建议修改centos、epel的源为清华大学的源，批量修改：

for i in `ls /etc/yum.repos.d/` ;do sed -i "s/mirror.centos.org/mirror.tuna.tsinghua.edu.cn/g" /etc/yum.repos.d/$i ;done

1.3 公共组件安装

1.3.1 数据库

本系列文档采用PXC作为集群数据库，参见本站PXC在CentOS8上安装。

1.3.2 Rabbitmq 集群安装

• 安装

dnf install -y rabbitmq-server

• 常用管理

systemctl enable rabbitmq-server.servicesystemctl start rabbitmq-server.servicesystemctl status rabbitmq-server.service

• 集群搭建

# 第一台systemctl start rabbitmq-server.service# 此时得到node的名称为rabbit@controller1rabbitmqctl cluster_status

• 同步cookie

[root@controller1 ~]# scp /var/lib/rabbitmq/.erlang.cookie root@controller2.sysit.cn:/var/lib/rabbitmq/[root@controller2 ~]# chown rabbitmq:rabbitmq /var/lib/rabbitmq/.erlang.cookie[root@controller2 ~]# ll /var/lib/rabbitmq/.erlang.cookie -r--------. 1 rabbitmq rabbitmq 20 Jul 14 17:32 /var/lib/rabbitmq/.erlang.cookie

• 加入集群

需要用到上一步得到的rabbit@controller1.

将controller2和controller3节点加入到集群

[root@controller2 ~]# systemctl restart rabbitmq-server[root@controller2 ~]# rabbitmqctl stop_appStopping rabbit application on node rabbit@controller2 ...[root@controller2 ~]# rabbitmqctl join_cluster  rabbit@controller1 Clustering node rabbit@controller2 with rabbit@controller1 [root@controller2 ~]# rabbitmqctl start_app Starting node rabbit@controller2 ... completed with 0 plugins.

有些网上的文档中采用了 rabbitmqctl join_cluster --ram rabbit@controller2，这表示这个节点是一个RAM节点，会将schema存储在内存中。默认值是–disc，存储在磁盘中，强烈推荐存储在磁盘。

• 检查是否创建成功

[root@controller1 ~]# rabbitmqctl cluster_statusCluster status of node rabbit@controller1 ...BasicsCluster name: rabbit@controller1.sysit.cnDisk Nodesrabbit@controller1rabbit@controller2rabbit@controller3Running Nodesrabbit@controller1rabbit@controller2rabbit@controller3Versionsrabbit@controller1: RabbitMQ 3.8.3 on Erlang 22.2.8rabbit@controller2: RabbitMQ 3.8.3 on Erlang 22.2.8rabbit@controller3: RabbitMQ 3.8.3 on Erlang 22.2.8Alarms(none)Network Partitions(none)ListenersNode: rabbit@controller1, interface: [::], port: 25672, protocol: clustering, purpose: inter-node and CLI tool communicationNode: rabbit@controller1, interface: [::], port: 5672, protocol: amqp, purpose: AMQP 0-9-1 and AMQP 1.0Node: rabbit@controller2, interface: [::], port: 25672, protocol: clustering, purpose: inter-node and CLI tool communicationNode: rabbit@controller2, interface: [::], port: 5672, protocol: amqp, purpose: AMQP 0-9-1 and AMQP 1.0Node: rabbit@controller3, interface: [::], port: 25672, protocol: clustering, purpose: inter-node and CLI tool communicationNode: rabbit@controller3, interface: [::], port: 5672, protocol: amqp, purpose: AMQP 0-9-1 and AMQP 1.0Feature flagsFlag: implicit_default_bindings, state: enabledFlag: quorum_queue, state: enabledFlag: virtual_host_metadata, state: enabled

以上均是Disk Nodes，如果有节点使用了--ram参数，则这里会显示RAM Nodes的信息

• 创建用户并配置读写策略

任意节点上操作

# 创建用户[root@controller1 ~]# rabbitmqctl add_user openstack passwordAdding user "openstack" ...[root@controller1 ~]# rabbitmqctl set_user_tags openstack administratorSetting tags for user "openstack" to [administrator] ...#允许openstack用户的配置、写和读访问:[root@controller1 ~]# rabbitmqctl set_permissions -p "/" openstack ".*" ".*" ".*"        Setting permissions for user "openstack" in vhost "/" ... #查看账号[root@controller1 ~]# rabbitmqctl list_usersListing users ...user    tagsopenstack       [administrator]guest   [administrator]

• 设置镜像队列高可用

任意节点上操作

rabbitmqctl set_policy ha-all "^" '{"ha-mode":"all"}'

• 查看镜像队列策略

rabbitmqctl list_policies 

• web管理插件

在需要的节点上启用web管理插件

rabbitmq-plugins enable rabbitmq_management

• 剔除节点

rabbitmqctl stoprabbitmqctl  -n rabbit@controller1 forget_cluster_node rabbit@controller2

1.3.3 memcached

dnf install -y memcached python3-memcachedsed -i 's|127.0.0.1,::1|0.0.0.0|g' /etc/sysconfig/memcachedsystemctl restart memcached