suse vcs 双机配置研究

    

1，双机所含资源（可从main.cf）

vi /etc/VRTSvcs/conf/config/main.cf

include "types.cf"

 

cluster dbcluster (

        UserNames = { admin = ajkCjeJgkFkkIskEjh }

        Administrators = { admin }

        )

 

system GDGZ-VPMN-SV03A-DBSVR (

        )

 

system GDGZ-VPMN-SV03B-DBSVR (

        )

 

//这个是snmp开放AMOS监控所需的group

group ClusterService (

        SystemList = { GDGZ-VPMN-SV03A-DBSVR = 0, GDGZ-VPMN-SV03B-DBSVR = 0 }

        AutoStartList = { GDGZ-VPMN-SV03A-DBSVR, GDGZ-VPMN-SV03B-DBSVR }

        )

 

        NIC hanic (

                Enabled = 0

                Device @GDGZ-VPMN-SV03A-DBSVR = bond0

                Device @GDGZ-VPMN-SV03B-DBSVR = bond0

                )

 

        NotifierMngr ntfr (

                SnmpConsoles = { "172.28.1.35" = Information }

                )

ntfr requires hanic

 

//这个是连接IOD的agent/DB/Datasync/Fileanalyse资源组

group dbcluster (

        SystemList = { GDGZ-VPMN-SV03A-DBSVR = 1, GDGZ-VPMN-SV03B-DBSVR = 2 }

        AutoStartList = { GDGZ-VPMN-SV03A-DBSVR, GDGZ-VPMN-SV03B-DBSVR }

        )

  Application RawRes (

                User = root

                StartProgram = "/home/cluster/raw/rawstart.sh"

                StopProgram = "/home/cluster/raw/rawstop.sh"

                CleanProgram = "/home/cluster/raw/rawstop.sh"

                MonitorProgram = "/home/cluster/raw/rawmonitor.sh"

                )

 

        Application WebsmsAPP (

                User = root

                StartProgram = "/home/cluster/app/start.sh"

                StopProgram = "/home/cluster/app/stop.sh"

                CleanProgram = "/home/cluster/app/stop.sh"

                MonitorProgram = "/home/cluster/app/monitor.sh"

                )

 

        Application WebsmsDB (

                User = root

                StartProgram = "/home/cluster/ora/orastart.sh"

                StopProgram = "/home/cluster/ora/orastop.sh"

                MonitorProgram = "/home/cluster/ora/oramonitor.sh"

                )

 

        Application WebsmsVG (

                User = root

                StartProgram = "/home/cluster/vg/vgstart.sh"

                StopProgram = "/home/cluster/vg/vgstop.sh"

                CleanProgram = "/home/cluster/vg/vgclean.sh"

                MonitorProgram = "/home/cluster/vg/vgmonitor.sh"

                )

 

        IP serviceip (

                Device = bond0

                Address = "X.X.X.X"

                NetMask = "X.X.X.X"

                )

 

Mount data0 (

                MountPoint = "/websmsshare"

                BlockDevice = "/dev/mapper/filevg-lv_websms"

                FSType = ext3

                FsckOpt = "-y"

                )

 

        Mount data1 (

                MountPoint = "/dataSync"

                BlockDevice = "/dev/mapper/filevg-lv_datasync"

                FSType = ext3

                FsckOpt = "-y"

                )

 

        Mount data2 (

                MountPoint = "/back"

                BlockDevice = "/dev/mapper/filevg-lv_back"

                FSType = ext3

                FsckOpt = "-y"

                )

 

        NIC SmcNic (

                Device = bond0

                )

RawRes requires WebsmsVG

        WebsmsAPP requires WebsmsDB

        WebsmsDB requires RawRes

        WebsmsVG requires serviceip

        data0 requires WebsmsVG

        data1 requires WebsmsVG

        data2 requires WebsmsVG

        serviceip requires SmcNic

 

 

2，             双机脚本路径/home/cluster

GDGZ-VPMN-SV03A-DBSVR:/home/cluster # ll

total 24

drwxr-x--- 2 root root 4096 Apr 12 16:28 app

drwxr-xr-x 2 root root 4096 Apr  9 17:41 lv

drwxr-xr-x 2 root root 4096 Apr 12 16:30 ora

drwxr-xr-x 2 root root 4096 Jan  7 14:43 raw

drwxr-xr-x 2 root root 4096 Apr 12 15:05 vg

 

 

lv下有一个binddbraw.sh脚本，该脚本定义了3个运行参数

start:

对每一个系统lv执行如下操作

raw  /dev/raw/raw23   /dev/datavg/lv_spfile

chown oracle:oinstall /dev/raw/raw23

 

 

stop:

对每一个系统lv执行如下操作

raw  /dev/raw/raw23   0 0

 

 

restart:

先执行stop,再执行start

 

 

Vg下有如下脚本

GDGZ-VPMN-SV03A-DBSVR:/home/cluster/vg # ll

total 20

-rwxr-xr-x 1 root root   81 Mar 23 15:51 vg_config.inc

-rwxr-xr-x 1 root root 1400 Jun  9  2009 vgclean.sh

-rwxr-xr-x 1 root root 2297 Jun  9  2009 vgmonitor.sh

-rwxr-xr-x 1 root root 1532 Jun  9  2009 vgstart.sh

-rwxr-xr-x 1 root root 1400 Jun  9  2009 vgstop.sh

 

其中vg_config.inc定义了VCS系统识别的2个VG

TOTAL_VGS=2

NAME_OF_VG1=datavg

NAME_OF_VG2=filevg

 

vgstart.sh执行激活VG操作

vgchange -ay ${VG_NAME} 2>&1

 

vgstop.sh执行去激活VG操作

vgchange -an ${VG_NAME} 2>&1

 

vgclean.sh和vgstop.sh内容一样

GDGZ-VPMN-SV03A-DBSVR:/home/cluster/vg # diff ./vgclean.sh ./vgstop.sh

GDGZ-VPMN-SV03A-DBSVR:/home/cluster/vg #

这也是为什么很多资源的clean动作直接引用stop动作脚本的原因！

 

vgmonitor.sh通过分析vgscan/lvscan的结果来判断vg的状态！

 

raw下有如下脚本：

GDGZ-VPMN-SV03A-DBSVR:/home/cluster/raw # ll

total 16

-rwxr-xr-x 1 root root 1498 Dec 25 11:24 raw_config.inc

-rwxr-xr-x 1 root root  839 Jun  9  2009 rawmonitor.sh

-rwxr-xr-x 1 root root   54 Jun  9  2009 rawstart.sh

-rwxr-xr-x 1 root root   54 Jun  9  2009 rawstop.sh

 

其中raw_config.inc定义了VCS监控的raw，具体如下：

 

#DB2 dataBase Cluster Scripter

 

DB2_VGNAME=datavg

 

#raw device count

DB2_LV_COUNT=30

 

#raw name (corresponding with the raw devices list)

DB2_MOUNT_RAW_NAME_1=/dev/raw/raw23

DB2_MOUNT_RAW_NAME_2=/dev/raw/raw24

DB2_MOUNT_RAW_NAME_3=/dev/raw/raw25

…

 

rawstart.sh就是执行如下命令：

/home/cluster/lv/binddbraw.sh start

 

rawstop.sh就是执行如下命令：

/home/cluster/lv/binddbraw.sh stop

 

rawmonitor.sh就是通过分析raw -qa的结果来判断raw的状态

 

 

ora下有如下脚本：

GDGZ-VPMN-SV03A-DBSVR:/home/cluster/ora # ll

total 12

-rwxr-xr-x 1 root root 500 Apr 12 16:06 oramonitor.sh

-rwxr-xr-x 1 root root 207 Jun  9  2009 orastart.sh

-rwxr-xr-x 1 root root 204 Jun  9  2009 orastop.sh

 

其中orastart.sh即执行如下两条命令

su - oracle -c "/home/oracle/product/11.1/bin/dbstart"

su - oracle -c "/home/oracle/product/11.1/bin/lsnrctl start"

 

orastop.sh即执行如下两条命令

su - oracle -c "/home/oracle/product/11.1/bin/dbshut"

su - oracle -c "/home/oracle/product/11.1/bin/lsnrctl stop"

 

oramonitor.sh通过分析ps -fu oracle的结果来判断oracle核心进程ora_pmon ora_ckpt ora_smon ora_dbw ora_lgw tnslsnr是否退出！

 

app下有如下脚本：

GDGZ-VPMN-SV03A-DBSVR:/home/cluster/app # ll

total 36

-rwx--x--- 1 root root  3693 Apr  1 10:33 monitor.sh

-rwxr-x--- 1 root root  1856 Apr 12 16:18 start.sh

-rwx--x--- 1 root root  2704 Apr  1 10:33 stop.sh

 

其中start.sh中定制了到IOD1/IOD2/IOD3和各区域userproxy/ccm的主机路由！这个很重要，否则不会从bond0:0(即以浮动IP 172.28.1.35出去)，而对端防火墙却只放通了172.28.1.35（10.244.70.11影射）！

route add -host 10.244.74.67 gw 172.28.1.1 dev bond0:0

route add -host 172.10.15.141 gw 172.28.1.1 dev bond0:0

route add -host 172.10.19.135 gw 172.28.1.1 dev bond0:0

route add -host 172.10.12.233 gw 172.28.1.1 dev bond0:0

route add -host 172.20.252.150 gw 172.28.1.1 dev bond0:0

route add -host 172.10.11.150 gw 172.28.1.1 dev bond0:0

route add -host 172.23.252.25 gw 172.28.1.1 dev bond0:0

route add -host 10.243.190.135 gw 172.28.1.1 dev bond0:0

route add -host 10.243.225.135 gw 172.28.1.1 dev bond0:0

后续还要添加4个AO点的主机路由！

 

另外还有如下3条启动应用资源的命令

su - websms -c "(cd "$WEBSMSPATH"/agent/bin; nohup sh smscenter.sh &)"

su - websms -c "(cd "$WEBSMSPATH"/agent/bin; nohup sh datasync.sh &)"

su - $UOMUSER -c "(cd "$UOMPATH"/bin; fileanalyse &)"

 

stop.sh其实就是ps –fu 出来上面的对应进程后，再kill -9掉这个进程，具体如下：

ps -fu $WEBSMSUSER | grep $PROC_DATASYNC |grep -v grep | awk '{print $2}'

kill -9 $pid

 

monitor.sh进程会对上面的3个进程自动拉起！！！即上面的3个进程如果退出，会被Monitor拉起来，而不是直接切换！！！这是由于DB双机上部署这些进程，DB优先级高，相对稳定，应用相对不稳定，如果应用异常就触发切换，对系统的影响更大，故采用了这种方式！

 

顺别加一句：对连接CCM的2个节点的进程，采用了root定义定时任务的方式监控：

5,10,15,20,25,30,35,40,45,50,55,59 * * * * sh /home/websms/mymonitor.sh

 

 

 

 

 

 

 

 

附录：

Datavg下的lv:

GDGZ-VPMN-SV03A-DBSVR:/dev/datavg # ll |awk '{print($9,$10,$11)}'

 

lv_ctl1 -> /dev/mapper/datavg-lv_ctl1

lv_ctl2 -> /dev/mapper/datavg-lv_ctl2

lv_ctl3 -> /dev/mapper/datavg-lv_ctl3

lv_data41 -> /dev/mapper/datavg-lv_data41

lv_data42 -> /dev/mapper/datavg-lv_data42

lv_data43 -> /dev/mapper/datavg-lv_data43

lv_data44 -> /dev/mapper/datavg-lv_data44

lv_data45 -> /dev/mapper/datavg-lv_data45

lv_data46 -> /dev/mapper/datavg-lv_data46

lv_data47 -> /dev/mapper/datavg-lv_data47

lv_data48 -> /dev/mapper/datavg-lv_data48

lv_data49 -> /dev/mapper/datavg-lv_data49

lv_data50 -> /dev/mapper/datavg-lv_data50

lv_data51 -> /dev/mapper/datavg-lv_data51

lv_data52 -> /dev/mapper/datavg-lv_data52

lv_data53 -> /dev/mapper/datavg-lv_data53

lv_data54 -> /dev/mapper/datavg-lv_data54

lv_data55 -> /dev/mapper/datavg-lv_data55

lv_data56 -> /dev/mapper/datavg-lv_data56

lv_data57 -> /dev/mapper/datavg-lv_data57

lv_data58 -> /dev/mapper/datavg-lv_data58

lv_data59 -> /dev/mapper/datavg-lv_data59

lv_data60 -> /dev/mapper/datavg-lv_data60

lv_data61 -> /dev/mapper/datavg-lv_data61

lv_data62 -> /dev/mapper/datavg-lv_data62

lv_data63 -> /dev/mapper/datavg-lv_data63

lv_data64 -> /dev/mapper/datavg-lv_data64

lv_data65 -> /dev/mapper/datavg-lv_data65

lv_redo1 -> /dev/mapper/datavg-lv_redo1

lv_redo2 -> /dev/mapper/datavg-lv_redo2

lv_redo3 -> /dev/mapper/datavg-lv_redo3

lv_spfile -> /dev/mapper/datavg-lv_spfile

lv_sysaux -> /dev/mapper/datavg-lv_sysaux

lv_system -> /dev/mapper/datavg-lv_system

lv_temp -> /dev/mapper/datavg-lv_temp

lv_undotbs -> /dev/mapper/datavg-lv_undotbs

lv_users -> /dev/mapper/datavg-lv_users

 

 

filevg下的lv:

GDGZ-VPMN-SV03A-DBSVR:/dev/filevg # ll |awk '{print($9,$10,$11)}'

 

lv_back -> /dev/mapper/filevg-lv_back

lv_datasync -> /dev/mapper/filevg-lv_datasync

lv_websms -> /dev/mapper/filevg-lv_websms

 

 

 

回到上面的binddbraw.sh

 

raw  /dev/raw/raw23   /dev/datavg/lv_spfile

                raw  /dev/raw/raw24   /dev/datavg/lv_ctl1

                raw  /dev/raw/raw25   /dev/datavg/lv_ctl2

                raw  /dev/raw/raw26   /dev/datavg/lv_ctl3

                raw  /dev/raw/raw27   /dev/datavg/lv_sysaux

                raw  /dev/raw/raw28   /dev/datavg/lv_system

                raw  /dev/raw/raw29   /dev/datavg/lv_temp

                raw  /dev/raw/raw30   /dev/datavg/lv_undotbs

                raw  /dev/raw/raw31   /dev/datavg/lv_users

                raw  /dev/raw/raw32   /dev/datavg/lv_redo1

                raw  /dev/raw/raw33   /dev/datavg/lv_redo2

                raw  /dev/raw/raw34   /dev/datavg/lv_redo3

                raw  /dev/raw/raw41   /dev/datavg/lv_data41

                raw  /dev/raw/raw42   /dev/datavg/lv_data42

                raw  /dev/raw/raw43   /dev/datavg/lv_data43

                raw  /dev/raw/raw44   /dev/datavg/lv_data44

                raw  /dev/raw/raw45   /dev/datavg/lv_data45

                raw  /dev/raw/raw46   /dev/datavg/lv_data46

                raw  /dev/raw/raw47   /dev/datavg/lv_data47

                raw  /dev/raw/raw48   /dev/datavg/lv_data48

                raw  /dev/raw/raw49   /dev/datavg/lv_data49

                raw  /dev/raw/raw50   /dev/datavg/lv_data50

                raw  /dev/raw/raw51   /dev/datavg/lv_data51

                raw  /dev/raw/raw52   /dev/datavg/lv_data52

                raw  /dev/raw/raw53   /dev/datavg/lv_data53

                raw  /dev/raw/raw54   /dev/datavg/lv_data54

                raw  /dev/raw/raw55   /dev/datavg/lv_data55

                raw  /dev/raw/raw56   /dev/datavg/lv_data56

                raw  /dev/raw/raw57   /dev/datavg/lv_data57

                raw  /dev/raw/raw58   /dev/datavg/lv_data58

                raw  /dev/raw/raw59   /dev/datavg/lv_data59

                raw  /dev/raw/raw60   /dev/datavg/lv_data60

                raw  /dev/raw/raw61   /dev/datavg/lv_data61

                raw  /dev/raw/raw62   /dev/datavg/lv_data62

                raw  /dev/raw/raw63   /dev/datavg/lv_data63

                raw  /dev/raw/raw64   /dev/datavg/lv_data64

                raw  /dev/raw/raw65   /dev/datavg/lv_data65

可知，实际就是绑定datavg下的lv/raw

 

 

 

了解了上面的原理：

假如双机异常，应即方案就是尽快在一台单机上启动DB，然后启动连接IOD的smscenter！最后启动同步开户文件的datasync和fileanalyse!

 

启动DB方法如下：

1，             在一台机器上尽快恢复浮动IP，可以使用资源最小方法。仅保留浮动IP资源！

2，             在获得浮动IP的机器上激活VG

3，             在获得浮动IP的机器上绑定raw

4，             启动DB（）