DB2 HADR操作手册

1．HACMP配置

Hosts文件主机，备机/etc/hosts如下：

主机：10.60.3.1(BootIP)

备机：10.60.3.2(BootIP)

/usr/es/sbin/cluster/etc/rhosts文件如下：

1.1配置HACMP脚本

在配置HACMP的过程中设定了两个资源组，分别为appa和appb，其中appa的

脚本为：

Start Script：/usr/es/sbin/cluster/script/

Stop Script： /usr/es/sbin/cluster/script/

资源组appb的脚本如下：

Start Script：/usr/es/sbin/cluster/script/

Stop Script：/usr/es/sbin/cluster/script/

1.2配置HACMP监控

在HACMP的监控中设定了两个监控事件分别为mona,monb，其中mona的脚本为：

/usr/es/sbin/cluster/script/

Cleanup Method为：/usr/es/sbin/cluster/script/

Restart Method为：/usr/es/sbin/cluster/script/

其中时间间隔以及稳定时间建议值如图所示：

monb的脚本为：

/usr/es/sbin/cluster/script/

其中时间间隔以及稳定时间建议值如图所示：

2．HADR配置

2.1HADR准备工作

1. 安装DB2，更新补丁，版本和主服务器一致

2. 创建相同的实例erpdbprd，服务端口：50000

/usr/opt/db2_08_01/instance/db2icrt -a server -w 64 -p db2c_erpdbprd -u

erpfusr erpdbprd

3. 更新DBM参数：db2 update dbm cfg using TP_MON_NAME CICS

4. 设置DB2的环境变量

db2set DB2_PINNED_BP=YES

db2set AUTOSTART=NO

db2set DB2_HASH_JOIN=ON

db2set DB2COUNTRY=86

db2set DB2COMM=TCPIP

db2set DB2CODEPAGE=1386

db2set DB2AUTOSTART=NO

db2set DB2_SKIPINSERTED=ON

db2set DB2_EVALUNCOMMITTED=ON

db2set DB2_HADR_BUF_SIZE= N*LOGBUFSZ (N>2)

5. 恢复DB2数据库，并处于Rollforward Pending状态

db2 restore db dberp from /dberpbackup to /dberp

db2 rollforward db erpdb to end of logs

通过db2 get db cfg for dberp|grep -i rollforward查看数据库状态，结果是

database(rollforward pending)

2.2HADR配置工作

1. 在erpdb_prd和erpdb_stb上配置HADR服务和侦听端口

用vi编辑/etc/services文件（需要切换到root用户），加入下面两行：

DB2_HADR_1 55001/tcp

DB2_HADR_2 55001/tcp

2. 在ERPDB_PRD上修改主数据库（ERPDB_PRD - DBERP）的配置参数：

UPDATE DB CFG FOR DBERP USING LOGINDEXBUILD ON

UPDATE DB CFG FOR DBERP USING INDEXREC RESTART

UPDATE DB CFG FOR DBERP USING HADR_LOCAL_HOST erpdb_prd

UPDATE DB CFG FOR DBERP USING HADR_LOCAL_SVC DB2_HADR_1

UPDATE DB CFG FOR DBERP USING HADR_REMOTE_HOST erpdb_stb

UPDATE DB CFG FOR DBERP USING HADR_REMOTE_SVC DB2_HADR_2

UPDATE DB CFG FOR DBERP USING HADR_REMOTE_INST erpdbprd

UPDATE DB CFG FOR DBERP USING HADR_SYNCMODE NEARSYNC

UPDATE DB CFG FOR DBERP USING HADR_TIMEOUT 25

3. 在erpdb_stb上修改备用数据库（erpdb_stb - DBERP）的配置参数：

UPDATE DB CFG FOR DBERP USING LOGINDEXBUILD ON

UPDATE DB CFG FOR DBERP USING INDEXREC RESTART

UPDATE DB CFG FOR DBERP USING HADR_LOCAL_HOST erpdb_stb

UPDATE DB CFG FOR DBERP USING HADR_LOCAL_SVC DB2_HADR_2

UPDATE DB CFG FOR DBERP USING HADR_REMOTE_HOST erpdb_prd

UPDATE DB CFG FOR DBERP USING HADR_REMOTE_SVC DB2_HADR_1

UPDATE DB CFG FOR DBERP USING HADR_REMOTE_INST erpdbprd

UPDATE DB CFG FOR DBERP USING HADR_SYNCMODE NEARSYNC

UPDATE DB CFG FOR DBERP USING HADR_TIMEOUT 25

2.3HADR启动

1． 停止应用

2． 启动备机HADR

#su – erpdbadm

$db2start

$db2 start hadr on db dberp as standby

3． 检查HADR是否是standby状态

$db2pd –hadr –db dberp

4． 启动主机HADR

$db2start

$db2 start hadr on db dberp as primary

5． 检查HADR是否是peer状态，建议在peer状态后才可以启动HACMP

$db2pd –hadr –db dberp

6． 启动备机HACMP

在erpdb_stb机器上，执行＃ smitty clstart

7． 启动主机HACMP

在erpdb_prd机器上，执行＃ smitty clstart

检查主机HACMP的状态，看serviceIP是否已经工作

2.4 HADR停止

在停止DB2 HADR之前，需要保证Hadr Primary运行在主机之上，同时Hadr

Standby运行在备机之上。如下图：

Hadr Primary运行在主机之上：

Hadr Standby运行在备机之上：

然后，在主机和备机之上分别运行smitty clstop的命令：

命令成功完成后，检查双机的状态：

则表示HACMP正常停止，并且HADR仍在工作，然后在Hadr Standby和Primary

上分别输入：

db2 deactivate db dberp

db2 stop hadr on db dberp

3．Hadr滚动升级过程

3.1DB2滚动升级（备份服务器erpdb_stb软件升级）

1. 升级erpdb_stb上的DB2，安装DB2的FIXPACK

#su – erpdbadm

$db2 deactivate db dberp

$db2 stop hadr on db dberp

$db2 force applications all

$db2 terminate

$db2 stop

$db2licd -end

$exit

#su – dasusr1

$db2admin stop

$exit

#/usr/opt/db2_08_01/bin/db2fmcu –d

#/usr/opt/db2_08_01/bin/db2fm –D –i dausr1

#/usr/sbin/slibclean

#cd /mnt/db2ptf

#./installFixPak –a -y

#/usr/opt/db2_08_01/instance/dasupdt dasusr1

#/usr/opt/db2_08_01/instance/db2iupdt erpdbadm

#su – erpdbadm

$db2start

$db2 START HADR ON DATABASE DBERP AS STANDBY

$db2pd –hadr –db dberp

检查数据库的HADR能够同步数据，处于Peer的状态。

3.2HADR切换过程（备份服务器接管主服务器）

停止应用

在进行HADR的切换之前必须要满足两个条件：

1． 主机和备机的Hadr必须处于Peer的状态，并且PrimaryFile和StandByFile

相同

2． 主机和备机的HACMP状态处于可工作状态（lssrc –g cluster）

以上两个条件缺一不可。

主机上停止hacmp：smitty clstop(takeover)，通过HACMP的takeover接管，将

DB2服务切换到HADR备机，主机HADR服务停止

在备机察看serviceIP是否已经工作，如图，ServiceIP已经切换至备机

察看备机Hadr的状态为Primary disconnected

3.3主服务器DB2升级

1. 升级erpdb_prd的DB2，安装DB2的FIXPACK

#su – erpdbadm

$db2 force applications all

$db2 terminate

$db2 stop

$db2licd -end

$exit

#su – dasusr1

$db2admin stop

$exit

#/usr/opt/db2_08_01/bin/db2fmcu –d

#/usr/opt/db2_08_01/bin/db2fm –D –i dasur1

#/usr/sbin/slibclean

#cd /mnt/db2ptf

#./installFixPak –a -y

#/usr/opt/db2_08_01/instance/dasupdt dasusr1

#/usr/opt/db2_08_01/instance/db2iupdt erpdbadm

#su – erpdbadm

$db2start

$db2 ACTIVATE DATABASE DBERP

$db2 START HADR ON DATABASE DBERP AS STANDBY

$db2pd –hadr –db dberp，HADR处于Peer状态。

3.4主服务器启动HACMP

在erpdb_prd机器上，执行＃ smitty clstart

3.5切换回主服务器进行正常的生产(可选)

在进行HADR的切换之前必须要满足两个条件：

1． 主机和备机的Hadr必须处于Peer的状态，并且PrimaryFile和StandByFile

相同

2． 主机和备机的HACMP状态处于可工作状态（lssrc –g cluster）

以上两个条件缺一不可。

在备份服务器发出：smitty clstop的命令，如下，选择Takeover的选项。

在主服务器检查serviceIP是否已经正常工作：

在主机上，检查Hadr的状态：Primary Peer

4．故障发生后的恢复方法

在故障发生后，需要进行Hadr的恢复工作，故障大致可以分为以下几类应用：

1． 主机系统故障

2． 主机应用故障

3． 备机系统故障

4． 备机应用故障

在以上4类的故障中，Hadr会通过故障切换来保证应用的正常生产。我们需要做的

工作不外乎以下的几个步骤：

1． 恢复现有的服务器能够正常工作。

2． 从现有的正常生产的系统进行online的备份，然后将备份在恢复后的服务器上

进行数据库的restore。

3． 启动restore后的数据库至hadr standby状态，察看主备机的Hadr状态是否

peer。

4． 检查现有的系统的Hacmp是否已经正常工作，否则，正常启动HACMP。

5． 仿照正常的滚动升级的步骤，进行Hadr的切换（如果有需要的话）。

5. 附录：HADR在HACMP中的脚本

#!/bin/ksh

route add 0 10.60.2.254

set -x

LOGFILE=/tmp/

banner appastart

echo "---start at time `date`---" >>${LOGFILE}

echo "`banner appastart`" >>${LOGFILE}

hostname=`hostname`

echo "---Start `hostname` begin at `date`---" >>${LOGFILE}

if [[ $hostname = "erpdb_prd" ]]

then

#Check if instance is started on 10.60.4.1, if not then start it

echo "Check if instance is started on 10.60.4.1, if not then start

it" >>${LOGFILE}

instancestatus=`su - erpdbadm -c db2gcf -s|grep -i state |awk '{print $4}'`

if [[ $instancestatus = "Available" ]]

then

echo "--------db2 is started------ " >>${LOGFILE}

else

su - erpdbadm -c db2start

echo "---------db2start at time `date`-------" >>${LOGFILE}

if [ $? -ne 0 ]

then

echo "Start db2 instance failure " >>${LOGFILE}

exit 0

fi

fi

#check if standby is alive

echo "check if standby is alive" >>${LOGFILE}

ping -c 1 10.60.4.2 > /dev/null

if [ $? -eq 0 ]

then

#10.60.4.2 is alive

echo "#10.60.4.2 is alive" >>${LOGFILE}

instancestatus=`rsh 10.60.4.2 su - erpdbadm -c db2gcf -s|grep -i state |awk

'{print $4}'`

if [[ $instancestatus = "Available" ]]

then

checkStandby=`rsh 10.60.4.2 ps -ef|grep db2hadrs|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# Standby is started, then start primary on 10.60.4.1

echo "scenario (3):Standby is started, then start primary on

10.60.4.1" >>${LOGFILE}

su - erpdbadm -c "db2 start hadr on database dberp as primary"

su - erpdbadm -c "db2 activate database dberp"

su - erpdbadm -c "db2 connect to dberp"

su - erpdbadm -c "db2 bind /usr/opt/db2_08_01/bnd/@

blocking all grant public"

su - erpdbadm -c "db2 bind /usr/opt/db2_08_01/bnd/@

blocking all grant public"

su - erpdbadm -c "db2 bind /usr/opt/db2_08_01/bnd/

blocking all grant public"

su - erpdbadm -c "db2rbind dberp -l all"

su - erpdbadm -c "db2 connect reset"

#Start CICS

#restart cics region

echo "----CICS start ... at time `date` ------" >>${LOGFILE}

sleep 5

/usr/es/sbin/cluster/script/

fi

#If 10.60.4.2 is primary , then takeover it

echo "#If 10.60.4.2 is primary , then takeover it" >>${LOGFILE}

checkStandby=`rsh 10.60.4.2 ps -ef|grep db2hadrp|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# Standby is started, then start primary on 10.60.4.1

echo "scenario (4): Standby is not started, then start standby on

10.60.4.1" >>${LOGFILE}

su - erpdbadm -c "db2 start hadr on database dberp as standby"

# wait for peer status

while :

do

hadrStatus=`su - erpdbadm -c db2pd -db dberp -hadr|grep Peer|awk '{print

$2}'`

echo $hadrStatus >> /tmp/

if [[ $hadrStatus = "Peer" ]]

then

break

fi

sleep 5

done

echo "scenario (4): Standby is started, then takeover it on

10.60.4.1" >>${LOGFILE}

su - erpdbadm -c "db2 takeover hadr on database dberp "

su - erpdbadm -c "db2 activate database dberp"

su - erpdbadm -c "db2 connect to dberp"

su - erpdbadm -c "db2 bind /usr/opt/db2_08_01/bnd/@

blocking all grant public"

su - erpdbadm -c "db2 bind /usr/opt/db2_08_01/bnd/@

blocking all grant public"

su - erpdbadm -c "db2 bind /usr/opt/db2_08_01/bnd/

blocking all grant public"

su - erpdbadm -c "db2rbind dberp -l all"

su - erpdbadm -c "db2 connect reset"

#Start CICS

#restart cics region

echo "----CICS start ... at time `date` ------" >>${LOGFILE}

sleep 5

/usr/es/sbin/cluster/script/

fi

#check hadr standby is running on 10.60.4.1,then takeover hadr on 10.60.4.1

echo "#check hadr standby is running on 10.60.4.1,then takeover hadr on

10.60.4.1" >>${LOGFILE}

checkStandby=`ps -ef|grep db2hadrs|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# Standby is started on 10.60.4.1, then takeover hadr on 10.60.4.1

echo "scenario (5): Standby is started on 10.60.4.1, then takeover hadr by

force on 10.60.4.1" >>${LOGFILE}

su - erpdbadm -c "db2 takeover hadr on database dberp by force"

su - erpdbadm -c "db2 activate database dberp"

#Start CICS

#restart cics region

echo "----CICS start ... at time `date` ------" >>${LOGFILE}

sleep 5

/usr/es/sbin/cluster/script/

fi

else

#Instance is not started on 10.60.4.2 , so needn't start HADR on 10.60.4.1 or

takeover hadr on 10.60.4.1

echo "Instance is not started on 10.60.4.2, so needn't start HADR on

10.60.4.1 or takeover hadr on 10.60.4.1" >>${LOGFILE}

#check hadr standby is running on 10.60.4.1,then takeover hadr on 10.60.4.1

checkStandby=`ps -ef|grep db2hadrs|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# Standby is started on 10.60.4.1, then takeover hadr on 10.60.4.1

echo "scenario (2): Standby is started on 10.60.4.1 , then takeover hadr by

force on 10.60.4.1" >>${LOGFILE}

su - erpdbadm -c "db2 takeover hadr on database dberp by force"

su - erpdbadm -c "db2 activate database dberp"

#Start CICS

#restart cics region

echo "----CICS start ... at time `date` ------" >>${LOGFILE}

sleep 5

/usr/es/sbin/cluster/script/

fi

#end check hadr standby is running on 10.60.4.1

exit 0

fi

else

#10.60.4.2 is not alive, so needn't start HADR on 10.60.4.1 or takeover hadr on

10.60.4.1

echo "10.60.4.2 is not alive, so needn't start HADR on 10.60.4.1 or takeover

hadr on 10.60.4.1" >>${LOGFILE}

#Check hadr standby is running on 10.60.4.1, then takeover hadr on 10.60.4.1

checkStandby=`ps -ef|grep db2hadrs|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# Standby is started on 10.60.4.1, then takeover hadr on 10.60.4.1

echo "Scenario (1): Standby is started on 10.60.4.1, then takeover hadr by

force on 10.60.4.1" >>${LOGFILE}

su - erpdbadm -c "db2 takeover hadr on database dberp by force"

su - erpdbadm -c "db2 activate database dberp"

#Start CICS

#restart cics region

echo "----CICS start ... at time `date` ------" >>${LOGFILE}

sleep 5

/usr/es/sbin/cluster/script/

fi

#end check hadr standby is running on 10.60.4.1

exit 0

fi

else

#Hostname is 10.60.4.2

#Check if instance is started on 10.60.4.2, if not then start it

echo "---------run on 10.60.4.2 at time `date`----------" >>${LOGFILE}

echo "Check if instance is started on 10.60.4.2, if not then start it"

instancestatus=`su - erpdbadm -c db2gcf -s|grep -i state |awk '{print $4}'`

if [[ $instancestatus = "Available" ]]

then

echo "----db2 is started----" >>${LOGFILE}

else

su - erpdbadm -c db2start

echo "---------db2start at time `date`-------" >>${LOGFILE}

if [ $? -ne 0 ]

then

echo "Start db2 instance failure " >>${LOGFILE}

exit 0

fi

fi

#If database on 10.60.4.2 is not started as standby, then start it as standby

checkStandby=`ps -ef|grep db2hadrs|grep DBERP|wc -l`

if [ $checkStandby -eq 0 ]

then

su - erpdbadm -c db2 start hadr on database dberp as standby

echo "---------start hadr as standby at time `date`-------" >>${LOGFILE}

fi

#Start 10.60.4.1 on 10.60.4.2

ping -c 1 10.60.4.1 > /dev/null

if [ $? -eq 0 ]

then

rsh 10.60.4.1 su - erpdbadm -c db2gcf -s

instancestatus=`rsh 10.60.4.1 su - erpdbadm -c db2gcf -s|grep -i state |awk

'{print $4}'`

if [[ $instancestatus = "Available" ]]

then

checkStandby=`rsh 10.60.4.1 ps -ef|grep db2hadrp|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# primary on 10.60.4.1 is fine, so change standby to primary on 10.60.4.2

echo "scenario (9): primary on 10.60.4.1 is fine, so change standby to

primary on 10.60.4.2" >>${LOGFILE}

su - erpdbadm -c db2 takeover hadr on database dberp

else

echo "scenario (8): primary on 10.60.4.1 is not fine, so takeover by force on

10.60.4.2 by force" >>${LOGFILE}

su - erpdbadm -c db2 takeover hadr on database dberp by force

fi

else

#intance on 10.60.4.1 is not alive, so takeover HADR by force

echo "scenario (7): intance on 10.60.4.1 is not alive, so takeover HADR by

force" >>${LOGFILE}

su - erpdbadm -c db2 takeover hadr on database dberp by force

fi

else

#10.60.4.1 is not alive, so takeover HADR by force

echo "scenario (6): 10.60.4.1 is not alive, so takeover HADR by

force" >>${LOGFILE}

su - erpdbadm -c db2 takeover hadr on database dberp by force

su - erpdbadm -c "db2 activate database dberp"

fi

#Start CICS

#restart cics region

echo "----CICS start ... at time `date` ------" >>${LOGFILE}

sleep 5

/usr/es/sbin/cluster/script/

fi

echo "--- finished on `hostname` at `date`---" >>${LOGFILE}

exit 0

#!/bin/ksh

set -x

banner appastop

LOGFILE=/tmp/

hostname=`hostname`

echo "--- start on `hostname` at `date`---" >>${LOGFILE}

#Stop CICS

#/usr/es/sbin/cluster/script/stopCICS

if [[ $hostname = "erpdb_prd" ]]

then

#check if standby is alive

ping -c 1 10.60.4.2 > /dev/null

if [ $? -eq 0 ]

then

#10.60.4.2 is alive

instancestatus=`rsh 10.60.4.2 su - erpdbadm -c db2gcf -s|grep -i state |awk

'{print $4}'`

if [[ $instancestatus = "Available" ]]

then

checkStandby=`rsh 10.60.4.2 ps -ef|grep db2hadrs|grep DBERP|wc -l`

if [ $checkStandby -eq 1 ]

then

# Standby is started, do nothing on primary

echo "scenario (2): Standby is started, do nothing on

primary" >>${LOGFILE}

exit 0

else

echo "scenario (3): stop db2 at time `date`" >>${LOGFILE}

su - erpdbadm -c db2gcf -k

fi

else

#Instance is not started on 10.60.4.2 , so needn't to consider takeover

echo "scenario (1): stop db2 at time `date`" >>${LOGFILE}

su - erpdbadm -c db2gcf -k

fi

else

#10.60.4.2 is not alive, so needn't to consider takeover

echo "scenario (4): stop db2 at time `date`" >>${LOGFILE}

su - erpdbadm -c db2gcf -k

fi

else

#Stop 10.60.4.1 on 10.60.4.2

#Do nothing on 10.60.4.2

echo "scenario (5): standby is not ok,nothing to do" >>${LOGFILE}

echo ok

#kill db2 process on 10.60.4.2

#su - erpdbadm -c db2gcf -k

fi

echo "--- finished on `hostname` at `date`---" >>${LOGFILE}

echo " " >>${LOGFILE}

exit 0

#!/bin/ksh

route add 0 10.60.2.254

set -x

banner appbstart

LOGFILE=/tmp/

hostname=`hostname`

echo "--- start on `hostname` at `date`---" >>${LOGFILE}

if [[ $hostname = "erpdb_stb" ]]

then

#Check if instance is started on 10.60.4.2, if not then start it

instancestatus=`su - erpdbadm -c db2gcf -s|grep -i state |awk '{print $4}'`

if [[ $instancestatus = "Available" ]]

then

echo "----db2 is started----" >>${LOGFILE}

else

echo "-------start db2 at time `date`----" >>${LOGFILE}

su - erpdbadm -c db2start

if [ $? -ne 0 ]

then

echo "Start db2 instance failure " >>${LOGFILE}

exit 1

fi

fi

echo "-----scenario (1): start hadr as standby at time `date`-----" >>${LOGFILE}

su - erpdbadm -c db2 start hadr on database dberp as standby

else

#Start 10.60.4.2 on 10.60.4.1

echo "-----scenario (2): nothing to do at time `date`-----" >>${LOGFILE}

#Do nothing on 10.60.4.1

exit 0

fi

echo "--- finished on `hostname` at `date`---" >>${LOGFILE}

echo " " >>${LOGFILE}

exit 0

#!/bin/ksh

set -x

banner appbstop

LOGFILE=/tmp/

hostname=`hostname`

echo "--- start on `hostname` at `date`---" >>${LOGFILE}

echo "--- finished on `hostname` at `date`---" >>${LOGFILE}

echo " " >>${LOGFILE}

exit 0

#! /bin/ksh

LOGFILE=/tmp/

echo " run at `date`">>${LOGFILE}

#Check instance status

su - erpdbadm -c db2gcf -s -i erpdbadm

rc=$?

if [ $rc -ne 0 ]

then

echo "DB2 instance error">>${LOGFILE}

exit $rc

fi

#Check database status

su - erpdbadm -c db2 connect to dberp

rc=$?

if [ $rc -ne 0 ]

then

echo "DB2 database error">>${LOGFILE}

exit $rc

fi

su - erpdbadm -c db2 disconnect current

#Check HADR status, write to logfile

echo "HADR status:`su - erpdbadm -c db2pd -db dberp -hadr|grep Peer|awk

'{print $2}'`" >>${LOGFILE}

echo "return code:$rc">>${LOGFILE}

exit $rc

#! /bin/ksh

su - erpdbadm -c db2gcf -s -i erpdbadm

流程图

Appastart

程序开始

判断机器名

是否为

Erpdb_prd

Y

判断db2启

动

N

启动db2

Y

N

程序结束

判断db2启

动是否成功

Y

判断stb机

是否存活

Y

判断stb数

据库实例是

否启动

Y

判断stb是

否hadrs

(4)

N

Y

(3)

启动成

primary并

且启动应用

程序结束

N

N

App接管

判断prd是

否hadrs

Y

(2)

Takeover

by force并

且启动应用

Sys接管

N

(6)

程序结束N

判断db2启

动是否成功

Y

判断stb是

否hadrs

Y

判断prd是

否存活

Y

判断prd数

据库实例是

否启动

Y

判断prd是

否hadrp

(9)

Y

程序结束Takeover

正常启

动或升

级

N

程序结束

判断stb是

否hadrp

Y

(4)

启动成

standby并

且等待peer

N

正常启

动或升

级

Takeover

并且启动应

用

程序结束

启动应用

N

(7)

N

App接管

N

启动成

hadrs

N

判断db2启

动

N

启动db2

Y

N

App接管

Takeover

by force

Takeover

by force

Sys接管

判断prd是

否hadrs

Y

(1)

Takeover

by force并

且启动应用

(8)

Takeover

by force

N

判断prd是

否hadrs

Y

(5)

Takeover

by force并

且启动应用

N

App接管

程序结束