Centos7 - corosync + pacemaker + pcs cluster

安装配置

[root@node-120 ~]# yum install crmsh pcs corosync pacemaker -y  # crmsh已经独立为一个项目，使用它需要独立安装，centos7自带的还是pcs.

[root@node-120 ~]# cat /etc/hosts
127.0.0.1localhostlocalhost.localdomainlocalhost4
::1localhostlocalhost.localdomain
192.168.199.120 node-120
192.168.199.123 node-123
192.168.199.124 node

systemctl start pcsd ; systemctl enable pcsd   # 每个节点都执行
echo "hacluster:openstack" | chpasswd    # 每个节点都执行

[root@node-120 ~]# pcs cluster auth -u hacluster -p openstack node-120 node-123

[root@node-120 ~]# pcs cluster setup –name controller node-120 node-123 --transport udpu --force  # 自动生成corosync配置文件

[root@node-120 ~]# pcs clsuter start --all   # 启动集群

[root@node-120 ~]# pcs  cluster enable --all  # 集群中的节点自动启动

# 如果pacemaker服务启动失败，报连接cib失败的错误，解决方法如下：
拷贝好的节点以下目录覆盖坏的节点的目录。
这里有个关于pacemaker的bug: https://bugs.launchpad.net/ubuntu/+source/pacemaker/+bug/1439649 
corosync默认data目录：/var/lib/corosync
pacemaker默认data目录：/var/lib/pacemaker # 这里有cib集群配置信息
pcs默认data目录：/var/lib/pcs  # 这里存放认证用户信息

[root@node-120 ~]# pcs resource defaults  resource-stickiness=100     # 防止故障恢复后资源迁移

[root@node-120 ~]# pcs property set no-quorum-policy=ignore
[root@node-120 ~]# pcs property set stonith-enabled=false

[root@node-120 ~]# pcs status
Cluster name: controller
Last updated: Fri Jul 31 10:32:04 2015
Last change: Thu Jul 30 21:42:13 2015 via cibadmin on node-120
Stack: corosync
Current DC: node-120 (1) - partition with quorum
Version: 1.1.10-32.el7_0.1-368c726
2 Nodes configured
2 Resources configured

注：使用pcs创建集群方便多了，而且centos7支持pcs也比较好。

[root@node-120 ~]# pcs resource create rs_mount_glance   Filesystem device=localhost:/glance directory=/glance_gfs fstype=glusterfs   # 创建一条自动挂载gluster volume的资源

[root@node-120 ~]# pcs resource clone rs_mount_glance  # 成为克隆资源

[root@node-120 ~]# pcs resource meta rs_mount_glance-clone failure-timeout=30   # 每隔30s对失败资源进行restart

[root@node-120 ~]# pcs config
Cluster Name: controller
Corosync Nodes:
node-120 node-123
Pacemaker Nodes:
node-120 node-123

Resources:
Clone: rs_mount_glance-clone
Meta Attrs: failure-timeout=30
Resource: rs_mount_glance (class=ocf provider=heartbeat type=Filesystem)
Attributes: device=localhost:/glance directory=/glance_gfs fstype=glusterfs
Operations: start interval=0s timeout=60 (rs_mount_glance-start-timeout-60)
stop interval=0s timeout=60 (rs_mount_glance-stop-timeout-60)
monitor interval=20 timeout=40 (rs_mount_glance-monitor-interval-20)

Stonith Devices:
Fencing Levels:

Location Constraints:
Ordering Constraints:
Colocation Constraints:

Cluster Properties:
cluster-infrastructure: corosync
dc-version: 1.1.10-32.el7_0.1-368c726
last-lrm-refresh: 1438347821
no-quorum-policy: ignore
stonith-enabled: false

[root@node-120 ~]# pcs resource update rs_mount_glance meta requires=nothing   # requires=nothing表示无论如何cluster都会启动这个resource

pcs提供了web UI界面


目前支持hacluster用户登录。

参考链接
Red_Hat_Enterprise_Linux-7-High_Availability_Add-On_Reference-en-US.pdf

two-node-cluster-corosyncpacemaker

Pacemaker-1.1-Clusters_from_Scratch-zh-CN.pdf

本文出自 “the-way-to-cloud” 博客，请务必保留此出处http://iceyao.blog.51cto.com/9426658/1680435