顺其自然 2015-11-15
Heartbeat是一个基于Linux开源的,被广泛使用的高可用集群系统。我们可以基于Heartbeat构建web高可用服务环境。本文在CentOS 6.5下做了一个简单示例,并对其日志进行了初步分析,供大家参考。
有关Heartbeat的相关知识,可以参考:
一、配置host解析及网络
###主机名配置,与/etc/hosts中的解析两者配置保持一致
[root@orasrv1 ~]# more /etc/sysconfig/network
NETWORKING=yes
HOSTNAME=orasrv1.xlk.com
[root@orasrv1 ~]# more /etc/hosts
127.0.0.1 localhost.localdomain localhost
#public ip
192.168.21.10 orasrv1.xlk.com orasrv1 ###node 1
192.168.21.13 orasrv2.xlk.com orasrv2 ###node 2
#private ip
192.168.144.128 orasrv1-priv.xlk.com orasrv1-priv
192.168.144.129 orasrv2-priv.xlk.com orasrv2-priv
[root@orasrv1 ~]# scp /etc/hosts orasrv2:/etc/hosts
[root@orasrv1 ~]# ifconfig |grep eth -A1
eth0 Link encap:Ethernet HWaddr 00:0C:29:35:31:E5 #eth0对外提供服务
inet addr:192.168.21.10 Bcast:192.168.21.255 Mask:255.255.255.0
--
eth1 Link encap:Ethernet HWaddr 00:0C:29:35:31:EF #eth1用作心跳网络
inet addr:192.168.144.128 Bcast:192.168.144.255 Mask:255.255.255.0
二、配置等效性
###节点1
# ssh-keygen -t rsa -f ~/.ssh/id_rsa -P ''
# ssh-copy-id -i ./.ssh/id_rsa.pub [email protected]
# ssh orasrv2 date;ssh 192.168.21.13 date;
###节点2
# ssh-keygen -t rsa -f ~/.ssh/id_rsa -P ''
# ssh-copy-id -i ./.ssh/id_rsa.pub [email protected]
# ssh orasrv1 date;ssh 192.168.21.10 date;
三、配置时间同步
此处略。可以参考:配置Linux 11G R2 RAC NTP服务
四、安装httpd
节点1:orasrv1
# yum install httpd
# service httpd start ###启动服务
# echo "orasrv1.xlk.com" >>/var/www/html/index.html
###打开浏览器看看能否正常访问,http://192.168.21.10/index.html
###如能正常访问,关闭服务,关闭自启动
# service httpd stop
# chkconfig httpd off
节点2:orasrv2
# yum install httpd
# service httpd start ###启动服务
# echo "orasrv2.xlk.com" >>/var/www/html/index.html
###打开浏览器看看能否正常访问,http://192.168.21.13/index.html
###如能正常访问,关闭服务,关闭自启动
# service httpd stop
# chkconfig httpd off
五、配置heartbeat
1、配置认证方式
[root@orasrv1 ~]# vi /etc/ha.d/authkeys |grep -v ^#
auth 1
1 md5 74a6a503b5bec9373b22bb630608df41 ###使用md5验证
[root@orasrv1 ~]# chmod 600 /etc/ha.d/authkeys
2、配置ha.cf
[root@orasrv1 ~]# vi /etc/ha.d/ha.cf
logfile /var/log/ha-log
keepalive 2
deadtime 15
warntime 10
initdead 60
udpport 694
bcast eth1
auto_failback off
node orasrv1.xlk.com
node orasrv2.xlk.com
ping 192.168.21.254
#respawn hacluster /usr/lib64/heartbeat/ipfail
3、配置haresources
[root@orasrv1 ~]# vi /etc/ha.d/haresources
orasrv1.xlk.com IPaddr::192.168.21.1/24/eth0 httpd
4、同步到第二节点
# scp -p /etc/ha.d/{ha.cf,authkeys,haresources} 192.168.21.13:/etc/ha.d/
六、验证测试
1、节点1启动heartbeat
###备注:本样例并未使用到共享文件系统,如配置drbd,或者nfs等。
[root@orasrv1 ~]# service heartbeat start
Starting High-Availability services:
2015/11/06_11:46:51 INFO: Resource is stopped ###此处提示资源处于停止状态
Done.
###以下为基于节点1上的日志分析
[root@orasrv1 ~]# tail -fn 50 /var/log/ha-log
heartbeat[4373]: 2015/11/06_11:46:51 info: Version 2 support: false
heartbeat[4373]: 2015/11/06_11:46:51 WARN: Logging daemon is disabled --enabling logging daemon is recommended
heartbeat[4373]: 2015/11/06_11:46:51 info: **************************
heartbeat[4373]: 2015/11/06_11:46:51 info: Configuration validated. Starting heartbeat 2.1.4
heartbeat[4374]: 2015/11/06_11:46:51 info: heartbeat: version 2.1.4
heartbeat[4374]: 2015/11/06_11:46:52 info: Heartbeat generation: 1446714158
heartbeat[4374]: 2015/11/06_11:46:52 info: glib: UDP Broadcast heartbeat started on port 694 (694) interface eth0
heartbeat[4374]: 2015/11/06_11:46:52 info: glib: UDP Broadcast heartbeat closed on port 694 interface eth0 - Status: 1
heartbeat[4374]: 2015/11/06_11:46:52 info: glib: ping heartbeat started.
heartbeat[4374]: 2015/11/06_11:46:52 info: G_main_add_TriggerHandler: Added signal manual handler
heartbeat[4374]: 2015/11/06_11:46:52 info: G_main_add_TriggerHandler: Added signal manual handler
heartbeat[4374]: 2015/11/06_11:46:52 info: G_main_add_SignalHandler: Added signal handler for signal 17
heartbeat[4374]: 2015/11/06_11:46:52 info: Local status now set to: 'up'
heartbeat[4374]: 2015/11/06_11:46:52 info: Link 192.168.21.254:192.168.21.254 up.
heartbeat[4374]: 2015/11/06_11:46:52 info: Status update for node 192.168.21.254: status ping
heartbeat[4374]: 2015/11/06_11:46:52 info: Link orasrv1.xlk.com:eth0 up.
heartbeat[4374]: 2015/11/06_11:47:52 WARN: node orasrv2.xlk.com: is dead
###上面的日志可以看出节点1上的heartbeat已经成功启动,此时尚未开始分配资源
###紧接着下面的提示节点2处于死亡状态,其时间间隔为60s(46:52s~47:52s),即我们配置文件中的initdead
###也就是说等待1分钟还没有探测到对方心跳,此时认为节点2死亡,注意,这里的死亡不等同于deadtime
###下面开始资源分配
heartbeat[4374]: 2015/11/06_11:47:52 info: Comm_now_up(): updating status to active
heartbeat[4374]: 2015/11/06_11:47:52 info: Local status now set to: 'active'
heartbeat[4374]: 2015/11/06_11:47:52 WARN: No STONITH device configured.
heartbeat[4374]: 2015/11/06_11:47:52 WARN: Shared disks are not protected.
heartbeat[4374]: 2015/11/06_11:47:52 info: Resources being acquired from orasrv2.xlk.com.
harc[4385]: 2015/11/06_11:47:52 info: Running /etc/ha.d/rc.d/status status
mach_down[4419]: 2015/11/06_11:47:52 info: /usr/share/heartbeat/mach_down: nice_failback: foreign resources acquired
mach_down[4419]: 2015/11/06_11:47:52 info: mach_down takeover complete for node orasrv2.xlk.com.
heartbeat[4374]: 2015/11/06_11:47:52 info: Initial resource acquisition complete (T_RESOURCES(us))
heartbeat[4374]: 2015/11/06_11:47:52 info: mach_down takeover complete.
IPaddr[4457]: 2015/11/06_11:47:52 INFO: Resource is stopped
heartbeat[4386]: 2015/11/06_11:47:52 info: Local Resource acquisition completed.
harc[4518]: 2015/11/06_11:47:52 info: Running /etc/ha.d/rc.d/ip-request-resp ip-request-resp
ip-request-resp[4518]: 2015/11/06_11:47:52 received ip-request-resp IPaddr::192.168.21.1/24/eth0 OK yes
ResourceManager[4537]: 2015/11/06_11:47:52 info: Acquiring resource group: orasrv1.xlk.com IPaddr::192.168.21.1/24/eth0 httpd
IPaddr[4563]: 2015/11/06_11:47:52 INFO: Resource is stopped ###上一条日志获取资源组,下一条日志启动资源
ResourceManager[4537]: 2015/11/06_11:47:52 info: Running /etc/ha.d/resource.d/IPaddr 192.168.21.1/24/eth0 start
IPaddr[4658]: 2015/11/06_11:47:53 INFO: Using calculated netmask for 192.168.21.1: 255.255.255.0
IPaddr[4658]: 2015/11/06_11:47:53 INFO: eval ifconfig eth0:0 192.168.21.1 netmask 255.255.255.0 broadcast 192.168.21.255
IPaddr[4629]: 2015/11/06_11:47:53 INFO: Success ###vip 分配成功
ResourceManager[4537]: 2015/11/06_11:47:53 info: Running /etc/init.d/httpd start ###开启httpd服务
heartbeat[4374]: 2015/11/06_11:48:03 info: Local Resource acquisition completed. (none)
heartbeat[4374]: 2015/11/06_11:48:03 info: local resource transition completed.
###验证VIP分配,如下,eth0:0上已获得VIP
[root@orasrv1 ~]# ifconfig |grep eth -A1
eth0 Link encap:Ethernet HWaddr 00:0C:29:35:31:E5
inet addr:192.168.21.10 Bcast:192.168.21.255 Mask:255.255.255.0
--
eth0:0 Link encap:Ethernet HWaddr 00:0C:29:35:31:E5
inet addr:192.168.21.1 Bcast:192.168.21.255 Mask:255.255.255.0
--
eth1 Link encap:Ethernet HWaddr 00:0C:29:35:31:EF
inet addr:192.168.144.128 Bcast:192.168.144.255 Mask:255.255.255.0
###验证httpd服务,如下,此时透过VIP访问到主页,从orasrv1,即节点1返回数据
http://192.168.21.1/index.html
orasrv1.xlk.com
2、节点2启动heartbeat
[root@orasrv2 ~]# service heartbeat start
Starting High-Availability services:
2015/11/06_11:50:22 INFO: Resource is stopped
Done.
###下面的日志为节点1,继续跟踪节点1的日志
###从下面的日志可以得知orasrv2.xlk.com: status up以及remote resource transition completed
heartbeat[4374]: 2015/11/06_11:50:23 info: Link orasrv2.xlk.com:eth0 up.
heartbeat[4374]: 2015/11/06_11:50:23 info: Status update for node orasrv2.xlk.com: status init
heartbeat[4374]: 2015/11/06_11:50:23 info: Status update for node orasrv2.xlk.com: status up
harc[4809]: 2015/11/06_11:50:23 info: Running /etc/ha.d/rc.d/status status
harc[4825]: 2015/11/06_11:50:23 info: Running /etc/ha.d/rc.d/status status
heartbeat[4374]: 2015/11/06_11:50:25 info: Status update for node orasrv2.xlk.com: status active
harc[4840]: 2015/11/06_11:50:25 info: Running /etc/ha.d/rc.d/status status
heartbeat[4374]: 2015/11/06_11:50:25 info: remote resource transition completed.
3、节点1停止heartbeat
###通过下面的命令来停止节点1上的heartbeat
[root@orasrv1 ~]# service heartbeat stop
Stopping High-Availability services:
Done.
###此时继续观察节点1的日志变化情况
heartbeat[4374]: 2015/11/06_11:53:13 info: Heartbeat shutdown in progress. (4374)
heartbeat[4873]: 2015/11/06_11:53:13 info: Giving up all HA resources. ###节点1开始放弃资源
ResourceManager[4886]: 2015/11/06_11:53:14 info: Releasing resource group: orasrv1.xlk.com IPaddr::192.168.21.1/24/eth0 httpd
ResourceManager[4886]: 2015/11/06_11:53:14 info: Running /etc/init.d/httpd stop ###停止httpd服务
ResourceManager[4886]: 2015/11/06_11:53:14 info: Running /etc/ha.d/resource.d/IPaddr 192.168.21.1/24/eth0 stop
IPaddr[4975]: 2015/11/06_11:53:14 INFO: ifconfig eth0:0 down
IPaddr[4946]: 2015/11/06_11:53:14 INFO: Success
heartbeat[4873]: 2015/11/06_11:53:14 info: All HA resources relinquished.
###所有的HA资源在节点1放弃完毕,下面开始kill heartbeat相关进程
heartbeat[4374]: 2015/11/06_11:53:15 WARN: 1 lost packet(s) for [orasrv2.xlk.com] [95:97]
heartbeat[4374]: 2015/11/06_11:53:15 info: No pkts missing from orasrv2.xlk.com!
heartbeat[4374]: 2015/11/06_11:53:16 info: killing HBREAD process 4381 with signal 15
heartbeat[4374]: 2015/11/06_11:53:16 info: killing HBFIFO process 4377 with signal 15
heartbeat[4374]: 2015/11/06_11:53:16 info: killing HBWRITE process 4378 with signal 15
heartbeat[4374]: 2015/11/06_11:53:16 info: killing HBREAD process 4379 with signal 15
heartbeat[4374]: 2015/11/06_11:53:16 info: killing HBWRITE process 4380 with signal 15
heartbeat[4374]: 2015/11/06_11:53:16 info: Core process 4378 exited. 5 remaining
heartbeat[4374]: 2015/11/06_11:53:16 info: Core process 4379 exited. 4 remaining
heartbeat[4374]: 2015/11/06_11:53:16 info: Core process 4381 exited. 3 remaining
heartbeat[4374]: 2015/11/06_11:53:16 info: Core process 4377 exited. 2 remaining
heartbeat[4374]: 2015/11/06_11:53:16 info: Core process 4380 exited. 1 remaining
heartbeat[4374]: 2015/11/06_11:53:16 info: orasrv1.xlk.com Heartbeat shutdown complete.
###观察节点2,此时也获取VIP
[root@orasrv2 ~]# ifconfig |grep eth -A1
eth0 Link encap:Ethernet HWaddr 00:0C:29:F1:2E:E4 #Author : Leshami
inet addr:192.168.21.13 Bcast:192.168.21.255 Mask:255.255.255.0
--
eth0:0 Link encap:Ethernet HWaddr 00:0C:29:F1:2E:E4 #Blog : http://blog.csdn.net/leshami
inet addr:192.168.21.1 Bcast:192.168.21.255 Mask:255.255.255.0
--
eth1 Link encap:Ethernet HWaddr 00:0C:29:F1:2E:EE
inet addr:192.168.144.129 Bcast:192.168.144.255 Mask:255.255.255.0
###验证httpd服务,如下,此时透过VIP访问到主页,从orasrv2,即节点2返回数据
http://192.168.21.1/index.html
orasrv2.xlk.com
4、使用hb_standby切换
###下面通过hb_standby将服务切换回节点1,切换前启动节点1上的heartbeat
[root@orasrv2 ~]# /usr/share/heartbeat/hb_standby
2015/11/06_11:56:05 Going standby [all].
###继续跟踪节点1的日志
heartbeat[5114]: 2015/11/06_11:55:23 info: Version 2 support: false
heartbeat[5114]: 2015/11/06_11:55:23 WARN: Logging daemon is disabled --enabling logging daemon is recommended
heartbeat[5114]: 2015/11/06_11:55:23 info: **************************
heartbeat[5114]: 2015/11/06_11:55:23 info: Configuration validated. Starting heartbeat 2.1.4
heartbeat[5115]: 2015/11/06_11:55:23 info: heartbeat: version 2.1.4
heartbeat[5115]: 2015/11/06_11:55:23 info: Heartbeat generation: 1446714159
heartbeat[5115]: 2015/11/06_11:55:23 info: glib: UDP Broadcast heartbeat started on port 694 (694) interface eth0
heartbeat[5115]: 2015/11/06_11:55:23 info: glib: UDP Broadcast heartbeat closed on port 694 interface eth0 - Status: 1
heartbeat[5115]: 2015/11/06_11:55:23 info: glib: ping heartbeat started.
heartbeat[5115]: 2015/11/06_11:55:23 info: G_main_add_TriggerHandler: Added signal manual handler
heartbeat[5115]: 2015/11/06_11:55:23 info: G_main_add_TriggerHandler: Added signal manual handler
heartbeat[5115]: 2015/11/06_11:55:23 info: G_main_add_SignalHandler: Added signal handler for signal 17
heartbeat[5115]: 2015/11/06_11:55:23 info: Local status now set to: 'up'
heartbeat[5115]: 2015/11/06_11:55:23 info: Link 192.168.21.254:192.168.21.254 up.
heartbeat[5115]: 2015/11/06_11:55:23 info: Status update for node 192.168.21.254: status ping
heartbeat[5115]: 2015/11/06_11:55:23 info: Link orasrv1.xlk.com:eth0 up.
heartbeat[5115]: 2015/11/06_11:55:25 info: Link orasrv2.xlk.com:eth0 up.
heartbeat[5115]: 2015/11/06_11:55:25 info: Status update for node orasrv2.xlk.com: status active
harc[5124]: 2015/11/06_11:55:25 info: Running /etc/ha.d/rc.d/status status
heartbeat[5115]: 2015/11/06_11:55:25 info: Comm_now_up(): updating status to active
heartbeat[5115]: 2015/11/06_11:55:25 info: Local status now set to: 'active'
heartbeat[5115]: 2015/11/06_11:55:26 info: remote resource transition completed.
heartbeat[5115]: 2015/11/06_11:55:26 info: remote resource transition completed.
heartbeat[5115]: 2015/11/06_11:55:26 info: Local Resource acquisition completed. (none)
heartbeat[5115]: 2015/11/06_11:55:26 info: Initial resource acquisition complete (T_RESOURCES(them))
###上面的日志表面,此时节点1已经启动完毕,并且2节点均处于活动状态
###下面是开启hb_standby之后输出的日志信息,所有的资源被节点1接管
heartbeat[5115]: 2015/11/06_11:56:06 info: orasrv2.xlk.com wants to go standby [all]
heartbeat[5115]: 2015/11/06_11:56:06 info: standby: acquire [all] resources from orasrv2.xlk.com
heartbeat[5143]: 2015/11/06_11:56:06 info: acquire all HA resources (standby).
ResourceManager[5156]: 2015/11/06_11:56:06 info: Acquiring resource group: orasrv1.xlk.com IPaddr::192.168.21.1/24/eth0 httpd
IPaddr[5182]: 2015/11/06_11:56:06 INFO: Resource is stopped
ResourceManager[5156]: 2015/11/06_11:56:06 info: Running /etc/ha.d/resource.d/IPaddr 192.168.21.1/24/eth0 start
IPaddr[5277]: 2015/11/06_11:56:06 INFO: Using calculated netmask for 192.168.21.1: 255.255.255.0
IPaddr[5277]: 2015/11/06_11:56:06 INFO: eval ifconfig eth0:0 192.168.21.1 netmask 255.255.255.0 broadcast 192.168.21.255
IPaddr[5248]: 2015/11/06_11:56:06 INFO: Success
ResourceManager[5156]: 2015/11/06_11:56:06 info: Running /etc/init.d/httpd start
heartbeat[5143]: 2015/11/06_11:56:07 info: all HA resource acquisition completed (standby).
heartbeat[5115]: 2015/11/06_11:56:07 info: Standby resource acquisition done [all].
heartbeat[5115]: 2015/11/06_11:56:07 info: remote resource transition completed.
七、小结
a、对于集群上节点之间的切换实质上是资源的接管,通常包括VIP分配回收,磁盘卸载挂载,以及服务的启停等。
b、对于备用节点,其相应的共享磁盘(非严格意义上)不可读写,服务处于停止状态。
环境说明:本环境由两台mysql 数据库和heartbeat 组成,一台的ip 为192.168.10.197,一台为192.168.10.198,对外提供服务的vip 为192.168.10.200