ovn-northd服务是集中式方案，集群中所有的ovn-controller服务都连接到ovn-northd，这样ovn-northd可能会引起单点故障的问题，所以OVN用在生产环境前需要考虑ovn-northd服务HA问题，
这里记录的OVN社区ovn-northd服务的HA方案。
Corosync + Pacemaker
corosync +pacemaker使用原理，可参见文档 http://blog.csdn.net/rzhzhz/article/details/7104135
ovn-northd HA 搭建过程
下面步骤搭建三个节点组成的ovn-northd HA的环境：
系统版本: CentOS7
host: 192.168.10.143 192.168.10.146 192.168.10.147
vip : 192.168.10.200
1.安装相关组件
1 # 安装 corosync,pacemaker 2 yum install corosynclib corosync 3 yum install pacemaker-libs pacemaker-cluster-libs pacemaker-cli pacemaker # 快速安装 yum install corosync pacemaker -y 4 5 #安装 crmsh(命令方式管理corosync +pacemaker) 6 yum install crmsh-scripts-3.0.0-2.2.noarch.rpm python-parallax-1.0.1-29.1.noarch.rpm crmsh-3.0.0-2.2.noarch.rpm 7 8 # 安装OVN组件 9 yum install openvswitch 10 yum install openvswitch-ovn-common openvswitch-ovn-central
2. 配置服务(192.168.10.143)
1 # 配置corosync + pacemaker
2 ## 编辑 /etc/corosync/corosync.conf
3 totem {
4 version: 2
5
6 crypto_cipher: aes128
7 crypto_hash: sha1
8 secauth: on
9
10 interface {
11 ringnumber: 0
12 bindnetaddr: 192.168.10.0
13 mcastaddr: 239.255.1.1
14 mcastport: 5405
15 ttl: 1
16 }
17 }
18
19 nodelist {
20 node {
21 ring0_addr: 192.168.10.143
22 nodeid: 1
23 }
24 node {
25 ring0_addr: 192.168.10.146
26 nodeid: 2
27 }
28 node {
29 ring0_addr: 192.168.10.147
30 nodeid: 3
31 }
32 }
33
34 logging {
35 fileline: off
36 to_stderr: no
37 to_logfile: yes
38 logfile: /var/log/cluster/corosync.log
39 to_syslog: yes
40 debug: off
41 timestamp: on
42 logger_subsys {
43 subsys: QUORUM
44 debug: off
45 }
46 }
47
48 quorum {
49 provider: corosync_votequorum
50 }
51
52
53 # 生成authkey
54 corosync-keygen
55
56 # 配置ovn-northd服务
57 ##在/etc/openvswitch/目录创建ovnnb-active.conf ovnsb-active.conf ovn-northd-db-params.conf
58 echo ‘tcp:192.168.10.200:6641‘ > /etc/openvswitch/ovnnb-active.conf
59 echo ‘tcp:192.168.10.200:6642‘ > /etc/openvswitch/ovnsb-active.conf
60 echo ‘--ovnnb-db=tcp:192.168.10.147:6641 --ovnsb-db=tcp:192.168.10.147:6642‘ > /etc/openvswitch/ovn-northd-db-params.conf
61
62
63 # *将上面生成的文件copy到192.168.10.146、192.168.10.147
64
65 # 启动各个节上的corosync + pacemaker服务
66 systemctl enable corosync pacemaker
67 systemctl start corosync pacemaker
3. 命令管理corosync + pacemaker
crm_mon查看节点是否加入集群
crm configure 配置方法：
方法1：
1 #crm configure 2 property stonith-enabled=false 3 primitive ovndb_servers ocf:ovn:ovndb-servers params master_ip="192.168.10.200" op start interval="0s" timeout="30s" op stop interval="0s" timeout="20s" op promote interval="0s" timeout="50s" op demote interval="0s" timeout="50s" op monitor interval="10s" timeout="20s" 4 primitive ovnip ocf:heartbeat:IPaddr2 params ip="192.168.10.200" cidr_netmask="24" op start interval="0s" timeout="20s" op stop interval="0s" timeout="20s" op monitor interval="10s" timeout="20s" 5 ms ovndb_servers-master ovndb_servers meta notify="true" 6 colocation colocation-ovndb_servers-master-ovnip-INFINITY inf: ovndb_servers-master:Started ovnip:Master 7 order order-ovnip-ovndb_servers-master-mandatory inf: ovnip:start ovndb_servers-master:start 8 verify 9 commit 10 bye
方法2：
1 crm configure load repalce /root/ovndb-ha.conf 2 3 # ovndb-ha.conf内容如下： 4 primitive ovndb_servers ocf:ovn:ovndb-servers params master_ip="192.168.10.200" op start interval="0s" timeout="30s" op stop interval="0s" timeout="20s" op promote interval="0s" timeout="50s" op demote interval="0s" timeout="50s" op monitor interval="10s" timeout="20s" 5 primitive ovnip ocf:heartbeat:IPaddr2 params ip="192.168.10.200" cidr_netmask="24" op start interval="0s" timeout="20s" op stop interval="0s" timeout="20s" op monitor interval="10s" timeout="20s" 6 ms ovndb_servers-master ovndb_servers meta notify="true" 7 colocation colocation-ovndb_servers-master-ovnip-INFINITY inf: ovndb_servers-master:Started ovnip:Master 8 order order-ovnip-ovndb_servers-master-mandatory inf: ovnip:start ovndb_servers-master:start 9 property stonith-enabled=false 10 #property no-quorum-policy=ignore
自动加载配置：
1 crm configure save ./ovndb-ha.conf 2 crm configure load replace ./ovndb-ha.conf 3 crm erase
*注意节点先启动 openvswitch.service！
crm_mon 显示各节点的状态
* 注 ovn-northd service HA：
1 #/usr/lib/ocf/resource.d/ovn/ovndb-servers 2 MANAGE_NORTHD_DEFAULT="yes"
参考文档：
http://docs.openvswitch.org/en/latest/topics/ovsdb-replication/
http://blog.csdn.net/rzhzhz/article/details/7104135
https://magedu.org/2016/04/23/corosync-pacemaker-mariadb/
https://mail.openvswitch.org/pipermail/ovs-dev/2016-October/324024.html
时间： 2024-12-18 06:48:07