第一部分:集群中事务决策各层简介
· Messageing Layer:心跳信息传递层
· ha_aware :集群事务决策软件。自己能够利用底层心跳信息传递层的功能,调用他的api完成事务决策的软件。
· DC:Designated Coordinator:(选定的协调员)为了防止主节点挂掉之后从节点争抢。有多从节点推举产生。
· CRM: (董事长)Cluster Resources Manager:负责做出决策。高可用集群中,任何资源都不应该自行启动,而是用CRM管理启动与否。
· LRM:(总经理)local resources Manager;让CRM的决策落实执行。真正管理本地资源,让本地资源启动停止,和状态监控。
· RA:resource agent :RA能够接受CRM的调度用于实现在节点上对某一个资源进行管理的工具,这个工具通常就是脚本。任何一个资源配置都需要依赖一个脚本或者程序。(需要接受四个参数{start|stop|status|restart} status:输出状态只能是running和stopped
failover:失效转移,故障转移
failback:失效转回,故障转回
每个层次中所用到的软件
· Messaging Layer:
heartbeat v1,v2,v3
corosync(openAIS):可用性事务委员会:定义开放的工业标准,为了能够让其理念让大家熟悉,openAIS推出了一个样例模版。corosync
cman(红帽):集群管理器。
· CRM:集群资源管理器:只要接口和信息层兼容,就可以独立使用。
heartbeat v1:haresour:配置文件 ;只是一个配置接口,
heartbeat v2:crm:(各节点都运行进程crmd;端口5566,客户端crmsh(shell)用户体验差):由于其不易配置,有人提供了接口heartbeat-GUI
heartbeat v3:heartbeat+pacemaker+ cluster-glue(黏合器):
pacemaker:(独立成了一个项目pacemaker)
配置接口:第三方开放配置工具,
CLI:命令行工具,crm(suse),pcs,---python语言研发
GUI:hawk(web界面),LCMC(窗口界面),pacemaker-mgmt
cman + rgmanager:
resource group manager:Failover Domanin:资源组管理器
配置接口:
RHCS:RedHat Cluster Suite
配置接口:Conga(完全生命周期的配置接口)安装,配置,资源,启动
· RA的类型。
heartbeat legacy:heartbeat的传统类型
LSB:脚本/etc/rc.d/init.d/*
OCF:Open Cluster Framework:开放集群框架
provider:pacemaker:资源代理脚本的组织
linbit :提供的资源代理
STONITH:节点隔离
· keepalived:命令。轻量级,跟前边的风格不同。借助vrrp协议完成ip地址资源流转,并利用自己内部的实现脚本调用的接口完成高可用功能。vrrp:虚拟路由冗余协议。
应用场景:
keepalived+ipvs
keepalived+haproxy
· RHEL OR CentOS高可用集群解决方案:
红帽5:
自带:RHCS(camn+rgmanager)
选用第三方:corosync+pacemaker,heartbeat(v1或v2),keepalived
红帽6:
自带:RHCS(cman+rgmanager)
corosync+rgmanager
cman+pacemaker
heartbeat v3 + pacemaker
keepalived
应用方法:
做前端负载均衡的高可用:keepalived
做大规模的高可用集群:corosync或(cman+pacemaker)支持多大100个节点
使用共享存储的时候,为了加快速度,存储元数据会保存的内存中,如果有两个集群节点同时对数据写操作,会导致文件系统崩溃。
实现资源隔离的方式:
可能需要借助硬件;1、硬件芯片(需要认证机制)。2、切断电源交换机。3、ssh
两个节点的集群是个特殊集群,断开的时候无法决策,此时可以使用ping node判断。
仲裁设备:
1 ping node ping group
2 qdisk:(红帽)选择一个硬盘
备用节点获取权限的时候为了放至其他集群访问资源,造成文件系统崩溃。需要让其他集群节点彻底kill掉。
第二部分:具体配置布置
1、确保两个节点主机名和解析,时间一至。时间可以使用ntp服务器来统一,请自行安装。在两台node上做相同配置
[root@node1 ~]# ntpdate 172.16.0.1 17 Apr 22:12:41 ntpdate[2079]: adjust time server 172.16.0.1 offset -0.001233 sec [root@node1 ~]# cat /etc/sysconfig/network NETWORKING=yes HOSTNAME=node1.nyist.com [root@node1 ~]# uname -n node1.nyist.com [root@node1 ~]# vim /etc/hosts 172.16.20.31 node1.nyist.com node1 172.16.20.32 node2.nyist.com node2 [root@node1 ~]# ping node1 PING node1.syist.com (172.16.20.31) 56(84) bytes of data. 64 bytes from node1.syist.com (172.16.20.31): icmp_seq=1 ttl=64 time=0.052 ms 64 bytes from node1.syist.com (172.16.20.31): icmp_seq=2 ttl=64 time=0.0
2、配置两节点实现基于ssh的密钥认证:目的是为了方便使用ssh命令管理对方节点。简化操作,当然,你要是有精力一次又一次的输入秘密验证,那此步骤也可以省略。(两个节点同时做一下操作)
[root@node1 ~]# ssh-keygen -t rsa -P ‘’ [root@node1 ~]# ls .ssh/ authorized_keys id_rsa id_rsa.pub [root@node1 ~]# ssh-copy-id -i .ssh/id_rsa.pub root@node2.nyist.com [root@node1 ~]# ssh node2 Last login: Thu Apr 17 22:24:51 2014 from 172.16.20.55 [root@node2 ~]#
3、Ok开始安装程序吧、(双节的都要安装)
[root@node1 heartbeat2]# yum install perl-TimeDate PyXML libnet net-snmp-libs
首先用yum先解决一些依赖。然后用rpm安装,注意,安装heartbeat的时候不要用yum,因为用yum安装会因为版本问题替换掉heartbeat所依赖的一下包。 [root@node1 heartbeat2]# rpm -ivh heartbeat-2.1.4-12.el6.x86_64.rpm heartbeat-pils-2.1.4-12.el6.x86_64.rpm heartbeat-stonith-2.1.4-12.el6.x86_64.rpm
4、安装好了让我们来配置吧。安装好heartbeat后,并没有生成配置文件,但是heartbeat提供了配置文件样本,所以需要我们将其copy过来。注意authkey的访问权限需要是600,因为它是跟安全相关的。
[root@node1 ha.d]# cp /usr/share/doc/heartbeat-2.1.4/authkeys /etc/ha.d [root@node1 ha.d]# cp /usr/share/doc/heartbeat-2.1.4/ha.cf /etc/ha.d [root@node1 ha.d]# cp /usr/share/doc/heartbeat-2.1.4/haresources /etc/ha.d
5、修改配置文件参数时其适用你的要求
[root@node1 ha.d]# vim authkeys auth 2 #1 crc 2 sha1 redhat #3 md5 Hello! [root@node1 ha.d]# chmod 600 authkeys [root@node1 ha.d]# vim ha.cf--------主配置文件 #debugfile /var/log/ha-debug------调试的时候开启用 File to write other messages to # logfile /var/log/ha-log---------------日志记录位置 #logfacility local0--------------也可以local7收日志 keepalive 1500ms-------------------心跳检测时间间隔 # # deadtime: how long-to-declare-host-dead? deadtime 6--------------------确认死亡时间 warntime 3-------------------节点掉线警告 # serial serialportname ...---串行心跳线时候使用的 #serial /dev/ttyS0 # Linux #serial /dev/cuaa0 # FreeBSD #serial /dev/cuad0 # FreeBSD 6.x #serial /dev/cua/a # Solaris mcast eth0 225.0.100.1 694 1 0---------多播方式通知 auto_failback on-------------------是否实现自动转回。 #stonith baytech /etc/ha.d/conf/stonith.baytech-------stonith设备,有设备了可以启用 node node1.nyist.com-------------------标识节点个数 node node2.nyist.com #debug 1------------------------------------调试级别 #compression bz2-------------压缩 #compression_threshold 2
分别给两个节点提供两个简单的验证页面然后验证httpd是否正常服务。切记不能让服务开机自动启动
[root@node1 ha.d]# curl node1 <h1>172.16.20.31@@@node1</h1> [root@node1 ha.d]# curl node2 curl: (7) couldn‘t connect to host [root@node1 ha.d]# curl node2 <h1>172.16.20.32@@@node2</h1> [root@node1 ha.d]# chkconfig httpd off [root@node1 ha.d]# ssh node2 ‘chkconfig httpd off‘
6、定义资源管理器开始提供服务。
[root@node1 ha.d]# vim haresources node1.nyist.com 172.16.20.100/16/eth0 httpd 定义虚拟IP:172.16.20.100
(定义资源:1.ip,2.httpd服务)
Ip:会自动去/etc/ha.d/resource.d目录下找IP定义的脚本定义IP
定义的Httpd服务也会去次目录还有/etc/rc.d/init.d/目录下找相应的服务
[root@node1 ha.d]# ls /etc/ha.d/resource.d/
apache IPaddr OCF
AudibleAlarm IPaddr2 portblock
db2 IPsrcaddr Raid1
Delay IPv6addr SendArp
Filesystem LinuxSCSI ServeRAID
hto-mapfuncs LVM WAS
ICP LVSSyncDaemonSwap WinPopup
ids MailTo Xinetd
Cp到节点二
[root@node1 ha.d]# scp ha.cf authkeys haresources root@node2:/etc/ha.d
7、启动服务
[root@node1 ~]# service heartbeat start logd is already running Starting High-Availability services: 2014/04/17_23:46:35 INFO: Resource is stopped 2014/04/17_23:46:35 INFO: Resource is stopped Done. [root@node1 ~]# ssh node2 ‘service heartbeat start‘ [root@node1 ha.d]# ifconfig eth0 Link encap:Ethernet HWaddr 00:0C:29:02:06:8E inet addr:172.16.20.31 Bcast:172.16.255.255 Mask:255.255.0.0 inet6 addr: fe80::20c:29ff:fe02:68e/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:33997 errors:0 dropped:0 overruns:0 frame:0 TX packets:6394 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:24361024 (23.2 MiB) TX bytes:777306 (759.0 KiB) eth0:0 Link encap:Ethernet HWaddr 00:0C:29:02:06:8E inet addr:172.16.20.100 Bcast:172.16.255.255 Mask:255.255.0.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:78 errors:0 dropped:0 overruns:0 frame:0 TX packets:78 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:9562 (9.3 KiB) TX bytes:9562 (9.3 KiB)
8、查看日志可以看出,有一个节点尝试连接,时间超时后自动KILL掉
[root@node1 ~]# tail /var/log/ha-log -f .heartbeat[2600]: 2014/04/18_18:41:12 info: ************************** heartbeat[2600]: 2014/04/18_18:41:12 info: Configuration validated. Starting heartbeat 2.1.4 heartbeat[2601]: 2014/04/18_18:41:12 info: heartbeat: version 2.1.4 heartbeat[2601]: 2014/04/18_18:41:12 info: Heartbeat generation: 1397817331 heartbeat[2601]: 2014/04/18_18:41:12 info: glib: UDP multicast heartbeat started for group 228.15.100.1 port 694 interface eth0 (ttl=1 loop=0) heartbeat[2601]: 2014/04/18_18:41:12 info: glib: ping heartbeat started. heartbeat[2601]: 2014/04/18_18:41:12 info: G_main_add_TriggerHandler: Added signal manual handler heartbeat[2601]: 2014/04/18_18:41:12 info: G_main_add_TriggerHandler: Added signal manual handler heartbeat[2601]: 2014/04/18_18:41:12 info: G_main_add_SignalHandler: Added signal handler for signal 17 heartbeat[2601]: 2014/04/18_18:41:12 info: Local status now set to: ‘up‘ heartbeat[2601]: 2014/04/18_18:41:13 info: Link 172.16.0.1:172.16.0.1 up. heartbeat[2601]: 2014/04/18_18:41:13 info: Status update for node 172.16.0.1: status ping heartbeat[2601]: 2014/04/18_18:41:31 WARN: node node2.nyist.com: is dead heartbeat[2601]: 2014/04/18_18:41:31 info: Comm_now_up(): updating status to active heartbeat[2601]: 2014/04/18_18:41:31 info: Local status now set to: ‘active‘ heartbeat[2601]: 2014/04/18_18:41:31 WARN: No STONITH device configured. heartbeat[2601]: 2014/04/18_18:41:31 WARN: Shared disks are not protected. heartbeat[2601]: 2014/04/18_18:41:31 info: Resources being acquired from node2.nyist.com. harc[2613]:2014/04/18_18:41:31 info: Running /etc/ha.d/rc.d/status status mach_down[2644]:2014/04/18_18:41:31 info: /usr/share/heartbeat/mach_down: nice_failback: foreign resources acquired mach_down[2644]:2014/04/18_18:41:31 info: mach_down takeover complete for node node2.nyist.com. heartbeat[2601]: 2014/04/18_18:41:31 info: mach_down takeover complete. heartbeat[2601]: 2014/04/18_18:41:31 info: Initial resource acquisition complete (mach_down) IPaddr[2687]:2014/04/18_18:41:31 INFO: Resource is stopped heartbeat[2614]: 2014/04/18_18:41:31 info: Local Resource acquisition completed. harc[2747]:2014/04/18_18:41:31 info: Running /etc/ha.d/rc.d/ip-request-resp ip-request-resp ip-request-resp[2747]:2014/04/18_18:41:31 received ip-request-resp 172.16.20.100/16/eth0 OK yes ResourceManager[2766]:2014/04/18_18:41:31 info: Acquiring resource group: node1.nyist.com 172.16.20.100/16/eth0 httpd IPaddr[2792]:2014/04/18_18:41:31 INFO: Resource is stopped ResourceManager[2766]:2014/04/18_18:41:31 info: Running /etc/ha.d/resource.d/IPaddr 172.16.20.100/16/eth0 start IPaddr[2889]:2014/04/18_18:41:31 INFO: Using calculated netmask for 172.16.20.100: 255.255.0.0 IPaddr[2889]:2014/04/18_18:41:31 INFO: eval ifconfig eth0:0 172.16.20.100 netmask 255.255.0.0 broadcast 172.16.255.255 IPaddr[2860]:2014/04/18_18:41:31 INFO: Success ResourceManager[2766]:2014/04/18_18:41:31 info: Running /etc/init.d/httpd start heartbeat[2601]: 2014/04/18_18:41:41 info: Local Resource acquisition completed. (none) heartbeat[2601]: 2014/04/18_18:41:41 info: local resource transition completed. heartbeat[2601]: 2014/04/18_18:43:42 info: Link node2.nyist.com:eth0 up. heartbeat[2601]: 2014/04/18_18:43:42 info: Status update for node node2.nyist.com: status init heartbeat[2601]: 2014/04/18_18:43:42 info: Status update for node node2.nyist.com: status up harc[3034]:2014/04/18_18:43:42 info: Running /etc/ha.d/rc.d/status status harc[3050]:2014/04/18_18:43:42 info: Running /etc/ha.d/rc.d/status status heartbeat[2601]: 2014/04/18_18:43:43 info: Status update for node node2.nyist.com: status active harc[3065]:2014/04/18_18:43:43 info: Running /etc/ha.d/rc.d/status status heartbeat[2601]: 2014/04/18_18:43:44 info: remote resource transition completed. heartbeat[2601]: 2014/04/18_18:43:44 info: node1.nyist.com wants to go standby [foreign] heartbeat[2601]: 2014/04/18_18:43:44 info: standby: node2.nyist.com can take our foreign resources heartbeat[3080]: 2014/04/18_18:43:44 info: give up foreign HA resources (standby). heartbeat[3080]: 2014/04/18_18:43:45 info: foreign HA resource release completed (standby). heartbeat[2601]: 2014/04/18_18:43:45 info: Local standby process completed [foreign]. heartbeat[2601]: 2014/04/18_18:43:45 WARN: 1 lost packet(s) for [node2.nyist.com] [11:13] heartbeat[2601]: 2014/04/18_18:43:45 info: remote resource transition completed. heartbeat[2601]: 2014/04/18_18:43:45 info: No pkts missing from node2.nyist.com! heartbeat[2601]: 2014/04/18_18:43:45 info: Other node completed standby takeover of foreign resources. Ourselves [root@node2 ha.d]# tail /var/log/ha-log heartbeat[17427]: 2014/04/18_18:43:41 info: ************************** heartbeat[17427]: 2014/04/18_18:43:41 info: Configuration validated. Starting heartbeat 2.1.4 heartbeat[17428]: 2014/04/18_18:43:41 info: heartbeat: version 2.1.4 heartbeat[17428]: 2014/04/18_18:43:41 info: Heartbeat generation: 1397749464 heartbeat[17428]: 2014/04/18_18:43:41 info: glib: UDP multicast heartbeat started for group 228.15.100.1 port 694 interface eth0 (ttl=1 loop=0) heartbeat[17428]: 2014/04/18_18:43:41 info: glib: ping heartbeat started. heartbeat[17428]: 2014/04/18_18:43:41 info: G_main_add_TriggerHandler: Added signal manual handler heartbeat[17428]: 2014/04/18_18:43:41 info: G_main_add_TriggerHandler: Added signal manual handler heartbeat[17428]: 2014/04/18_18:43:41 info: G_main_add_SignalHandler: Added signal handler for signal 17 heartbeat[17428]: 2014/04/18_18:43:42 info: Local status now set to: ‘up‘ heartbeat[17428]: 2014/04/18_18:43:42 info: Link 172.16.0.1:172.16.0.1 up. heartbeat[17428]: 2014/04/18_18:43:42 info: Status update for node 172.16.0.1: status ping heartbeat[17428]: 2014/04/18_18:43:43 info: Link node1.nyist.com:eth0 up. heartbeat[17428]: 2014/04/18_18:43:43 info: Status update for node node1.nyist.com: status active harc[17440]:2014/04/18_18:43:43 info: Running /etc/ha.d/rc.d/status status heartbeat[17428]: 2014/04/18_18:43:43 info: Comm_now_up(): updating status to active heartbeat[17428]: 2014/04/18_18:43:43 info: Local status now set to: ‘active‘ heartbeat[17428]: 2014/04/18_18:43:44 info: remote resource transition completed. heartbeat[17428]: 2014/04/18_18:43:44 info: remote resource transition completed. heartbeat[17428]: 2014/04/18_18:43:44 info: Local Resource acquisition completed. (none) heartbeat[17428]: 2014/04/18_18:43:44 info: node1.nyist.com wants to go standby [foreign] heartbeat[17428]: 2014/04/18_18:43:45 info: standby: acquire [foreign] resources from node1.nyist.com heartbeat[17458]: 2014/04/18_18:43:45 info: acquire local HA resources (standby). heartbeat[17458]: 2014/04/18_18:43:45 info: local HA resource acquisition completed (standby). heartbeat[17428]: 2014/04/18_18:43:45 info: Standby resource acquisition done [foreign]. heartbeat[17428]: 2014/04/18_18:43:45 info: Initial resource acquisition complete (auto_failback) heartbeat[17428]: 2014/04/18_18:43:45 info: remote resource transition completed.
9、用浏览器验证故障专业效果。
关闭node1节点的时候转移到node2上。重新打开node1后,转移到了node1.因为我们设置了资源优先级。
使用nfs创建共享存储:让web使用(假如不允许同时挂载)
创建提供NFS的服务器
[root@My2 ~]# mkdir /www/htdoc -p [root@My2 ~]# setfacl -m u:apache:rwx /www/htdoc/ [root@My2 ~]# vim /etc/exports /www/htdoc 172.16.0.0/16 (rw) [root@My2 ~]# vim /www/htdoc/index.html <h1>from NFS server</h1> [root@My2 ~]# servie nfs start
去各节点编辑资源:
[root@node1 ha.d]# vim haresources node1.nyist.com 172.16.20.100/16/eth0 Filesystem::172.16.20.32:/www/htdoc::/var/www/html::nfs httpd 定义了三个资源有先后顺序之分 172.16.20.100/16/eth0 172.16.20.32:/www/htdoc httpd
重启heartbeat服务即可见成果,测试,停掉一台node后,不仅vip回收了,而且挂载的NFS也自动卸载掉了,放置对数据资源的占有,
故障转移的日志信息
IPaddr[22187]:2014/04/18_22:10:40 INFO: Resource is stopped ResourceManager[22161]:2014/04/18_22:10:40 info: Running /etc/ha.d/resource.d/IPaddr 172.16.20.100/16/eth0 start IPaddr[22284]:2014/04/18_22:10:40 INFO: Using calculated netmask for 172.16.20.100: 255.255.0.0 IPaddr[22284]:2014/04/18_22:10:40 INFO: eval ifconfig eth0:0 172.16.20.100 netmask 255.255.0.0 broadcast 172.16.255.255 IPaddr[22255]:2014/04/18_22:10:40 INFO: Success Filesystem[22385]:2014/04/18_22:10:40 INFO: Resource is stopped ResourceManager[22161]:2014/04/18_22:10:40 info: Running /etc/ha.d/resource.d/Filesystem 172.16.20.62:/www/htdoc /var/www/html nfs start Filesystem[22463]:2014/04/18_22:10:41 INFO: Running start for 172.16.20.62:/www/htdoc on /var/www/html Filesystem[22452]:2014/04/18_22:10:41 INFO: Success ResourceManager[22161]:2014/04/18_22:10:41 info: Running /etc/init.d/httpd start mach_down[22136]:2014/04/18_22:10:41 info: /usr/share/heartbeat/mach_down: nice_failback: foreign resources acquired mach_down[22136]:2014/04/18_22:10:41 info: mach_down takeover complete for node node1.nyist.com. heartbeat[22097]: 2014/04/18_22:10:41 info: mach_down takeover complete. heartbeat[22097]: 2014/04/18_22:10:41 info: Initial resource acquisition complete (mach_down) heartbeat[22097]: 2014/04/18_22:10:51 info: Local Resource acquisition completed. (none) heartbeat[22097]: 2014/04/18_22:10:51 info: local resource transition completed.
测试页面
本文出自 “Sysinit” 博客,请务必保留此出处http://xuqimin.blog.51cto.com/3832456/1398127