Implementing ServiceGuard on HP_UX 11v3 for oracle apps R12

[Barakati]

Hi,

We are planning to implement ServiceGuard on HP_UX 11v3 for oracle E- Business Suite R12.
I would like to know the possibilities of doing that with oracle apps. I appreciate if you could direct me to proper documentaion or white paper or any resource i should refer before planing.

Kindly suggest, Thanks in advance.

[Steven E. Protter]

Shalom,

Like any server daemon, you can install the binaries into both nodes of a server, and create an active passive configuration, running on one of two nodes.

Data needs to be on shared storage.

http://docs.hp.com/en/6033/HPServiceguardClusterConfig_WP.pdf

That document is not specific to any OS, but will work.

It is also possible to buy oracle add on scripts and such. This is merely a shortcut.

SEP

[Vinod Kumar Yadav]

Hi,

For Oracle E- Business or any ERP like SAP
you can install 2 Node Cluster with Active Passive Configuration for running the DB & APP.

You Can install the Oracle Binaries in the local disk.For connecting the Oracle Database.

Create the 2 pkg one for DB & one for App.

The DB & App will used the External Storage.

Thanks
Vinod

[Barakati]

Hi, Thanks for your replies.

We already have our ERP R12 running on shared drive. I mean we have installed everything on shared drive on one server(s1).
In this case we have to get ready the another node(s2) to implement SG.

How shall i go in this scenario.
Kindly let me know if it is not clear.

Thanks in advance.

[Basheer_2]

hi Barakati,

If everything is running in node1 and you want to create a cluster of node1 and another node2.

follow these steps.


Both nodes
Root mirror

# dont allow the os to auto activate the vgs
# (vg00 is auto activated)
vi /etc/lvmrc
VG_AUTO_ACTIVATE=0
customer_vgactivation
{
# If there are any LOCAL vgs, activate here
/usr/sbin/vgchange -a y /dev/vglocal-not-cluster-vg
}

# let the cluster auto start after reboots
vi /etc/rc.config.d/cmcluster
AUTOSTART_CMCLD=1

NTP config /etc/ntp.conf
Primary and secondary time sources
Nptq â p
/sbin/init.d/xntpd start/stop/status
Xntpq
/etc/rc.config.d/netdaemons â TIME_SERVER=1

.rhosts or cmclnodelist (/etc/cmcluster)



Node 1
-------
0) Create LUN, ioscan, insf -e
1) pvcreate -f /dev/rdks/c1t1d0 ( Assume this is a SAN LUN vdisk)
2) mkdir /vg01
3) mknod /dev/vg01/group c 64 0x010000 (0x## this number must match in node2, step 2 under Node 2)
4) vgcreate /dev/vg01 /dev/dsk/c1t1d0
5) lvcreate -L 2048 -n lvora /dev/vg01
6) newfs -F vxfs /dev/vg01/rlvora
7) mkdir /oracle
8) mount -F vxfs /dev/vg01/lvora /oracle
9) vgexport -v -s -p -m /tmp/vg01.map /dev/vg01
10) scp /tmp/vg01.map node2:/tmp

Node 2
------
1) mkdir /dev/vg01
2) mknod /dev/vg01/group c 64 0x010000 (0x## This number must be same as Step 3 of Node 1)
3) vgimport -v -s -m /tmp/vg01.map /dev/vg01
4) mkdir /oracle
5) vgchange -a r /dev/vg01
6) mount â F vxfs â o ro /dev/vg01/lvora /oracle
7) vgcfgbackup /dev/vg01
8) vgchange -a n /dev/vg01


node1
------
lan0 heartbeat 192.168.1.2
lan1 primary A.B.C.101
lan2 standby

node2
------
lan0 heartbeat 192.168.1.2
lan1 primary A.B.C.102
lan2 standby

add lan0 and lan1 to /etc/hosts


in Node 1 and Node 2 (for ftp/rcp) cmclnodelist is Identicall on all nodes
--------------------------------------------------------------------------
#Custer configuration steps
#1)
vi /etc/cmcluster/cmclnodelist (create in node1 and node2)
node1 root
node2 root

chmod 444 cmclnodelist
rcp cmclnodelist node2:$PWD

in Node 1
---------
#2) Create cluster config file
cd /etc/cmcluster
cmquerycl -v -C cmclconfig.ascii -n node1 -n node2


vi cmclconfig.ascii # change ( minimum it needs these 2)
CLUSTER_NAME psoft
NODE_TIMEOUT 2000000 # 2 seconds, good to have 2-3 secs

# change other parameters
# Add the vgs
VOLUME-GROUPS
/dev/vg01

#3) check for errors
cmcheckconf -v -C /etc/cmcluster/cmclconfig.ascii

#4) compile cmclconfig.ascii to binary and distribute the binary file to all the other cluter nodes
cmapplyconf -v -C /etc/cmcluster/cmclconfig.ascii

#5) Start the cluster and view
cmruncl â v (first time start, 100% node attendance is required. For other times to start cluster in one node cmruncl â v â n node1
cmviewcl


now a basic cluster is up.
to halt
cmhaltl -v


now start creating the packages.
Each Package needs 2 files 1) pkg conf file and 2) pkg script control file

create pkg conf file
mkdir /etc/cmcluster/jdeprod
cd /etc/cmcluster/jdepro
cmmakepkg -v -p jdeprod.conf

vi jdeprod.conf
PACKAGE_NAME jdeprod
PACKAGE_TYPE FAILOVER
FAILOVER_POLICY CONFIGURED NODE
FAILBACK_POLICY MANUAL
NODE_NAME node1
NODE_NAME node2
AUTO_RUN YES
LOCAL_LAN_FAILOVER_ALLOWED YES
RUN_SCRIPT /etc/cmcluster/clustername/pkgname/jdeprod.ctl start
RUN_SCRIPT_TIMEOUT NO_TIMEOUT
HALT_SCRIPT /etc/cmcluster/clustername/pkgname/jdeprod.ctl stop
HALT_SCRIPT_TIMEOUT NO_TIMEOUT
SERVICE_NAME orapromon
SERVICE_HALT_TIMEOUT 120
SERVICE_NAME jdepromon
SERVICE_HALT_TIMEOUT 120
SUBNET 204.187.168.0


create pkg control script
cd /etc/cmcluster/jdepro
cmmakepkg -v -s jdepro.cntl

add all the vgs, lvs, fs and mount points used by the pkg
VG[0]=vg01
VG{1}=vg02

Still working on to finish the pkg conf and pkg control script.

[Barakati]

HI,

I have one more doubt as a DBA.
Do i need to install oracle binaries on node2?

Because we already installed whole db+apps on shared hard disk between node1 and node2.
If not needed how node2 will run oracle?

Thanks

[Vinod Kumar Yadav]

Hi,


The above MCSG Cluster Installation & pkg configuration are correct.Folloe the steps.

And Oracle Binaries need to be install on both node correct.



Thanks

[Vinod Kumar Yadav]

Hi,


Please find the attached MCSG Cluster Installation Step by Step.


Thanks