General Help Needed

[cheesytime]

I have a HP-UX 11.11 box running on a HP 9000 hardware.

I would like to setup a two node cluster (hardware will be the same on both nodes).

The latest version of ServiceGuard that would run on a HP-UX 11.11 is 11.16 right?

The main node is running oracle and has a bunch of disk mounted off an HP XP Storage.

Could someone please tell me what needs to be done in order to create the cluster?

I have been reading some stuff regarding ServiceGuard, but because this is my first cluster and the the box (which would be the first node) is already on a production environment im afraid I might break things.

Is there any guide that tells you step-by-step what needs to be done in order to create an oracle cluster with ServiceGuard?

Help on this matter would be strongly appreciated.

[Steven E. Protter]

Shalom,

The latest version of ServiceGuard that would run on a HP-UX 11.11 is 11.16 right?

Yes.

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

Its a great doc. I am using it now.

Basic principle.

Each package needs its own volume group on shared storage which it activates at start up.

SEP

[cheesytime]

Steven:

Thanks a lot for the quick response.

Do you have any document that would explain how to set it up step by step on a UX box?

if the doc covers an oracle cluster, amazing.

I cant find any on the net, at least not a very intuitive one.

[Mel Burslan]

Andrew,

There is no simple. step-by-step, how to create a cluster cookbook, other than the document SEP has pointed in the previous post. Clusters and the application packages running on them are so site specific, there is no easy way outline the process in such a document. Every site should have such document but it has to be created by the cluster's sysadmin as the cluster gets built.

My suggestion is to initially set up clusters on a totally non-production hardware, then shift your production instance into the newly formed cluster, leaving the standalone production system behind. The reason I am saying this, before you can certify a cluster, you will need to perform several tests regarding fail-over capabilities of your application, and some of these tests should involve, abruptly powering off the server. And since the results of this activity can be quite unpredictable, ending up taking your production system down (due to a tie breaking requirement or some other reason). You really do not want to do that, do you ?

I know it is a bit of a hassle and extra cost of finding hardware to setup brand new cluster but it is the best approach. What you can do is, you can match your new server to be clustered with the running prod system, match by an old clunker as the initial cluster setup. Once you are sure everything is in order, you can take a weekend downtime to install and test your cluster on the existing prod machine. As you might or might not know, serviceguard can run on different types and classes of hardware as long as the OS is the same on both machines.

HTH

[Mel Burslan]

On a second note, I just realized you are setting up an oracle cluster. There is an add-on product (your eyes should roll $$$ signs like Donald Duck's rich uncle, right about now) called SGeOracle (I am not 100% sure about the name) which is a set of scripts and maybe executables, that makes integration of Oracle into a serviceguard package easy(er).If you purchase this Oracle integration package, generally an HP consultant comes onsite to set it up. Then they support the SG together with Oracle running on it. So, if one day Oracle spits out an error message related to OS, HP helps you, instead of telling to go pound sand by saying, "you need to ask Oracle about it".

Consider licensing this integration package. It is expensive, but after all, if you are (actually your company is) spending big bucks to protect a business critical asset like a database, I think adding a few thousand dollars to this expanse every year is well worth the money.

[Ganesan R]

Hi,

http://docs.hp.com/en/B3936-90135/B3936-90135.pdf

[cheesytime]

Thanks Gen, the following document looks amazing:

http://docs.hp.com/en/B3936-90135/B3936-90135.pdf

Anyawys, where do I get ServiceGuard, on which media set / CD is it located?

I know we have purchased it, I just dont know where it might be.

[Ganesan R]

Hi,

It should be there in your application CD. However you can check with swlist command once you mounted your cd.

#swlist -d @ /cdrom

This command will show the products available on mount point /cdrom.

[Mel Burslan]

It has been a while I have gotten application *CD*s instead of DVD but if I am not mistaken SG was on CD 4 (out of 5 CD set).

[Vinod Kumar Yadav]

Hi,

Please check the MC/Service Guard is installed.

root # swlist | grep -i service
B3935DA A.11.09 MC / Service Guard
PHNE_23697 1.0 NTP timeservices upgrade plus utilities
PHSS_22683 1.0 MC/ServiceGuard and SG-OPS Edition A.11.09
PHSS_25499 1.0 MC/ServiceGuard and SG-OPS Edition A.11.09.

root # ps -ef | grep -i cmcld
root 2222 1 0 May 17 ? 17:57 /usr/lbin/cmcld -j
root 5322 4142 0 12:03:27 pts/ta 0:00 grep -i cmcld

And if MC/ServiceGuard License is not installed install the Software CODE Word trough sam.

Than Install & Configure the Cluster Step by Step.

Node #1
-------

# strings /etc/lvmtab
# pvcreate /dev/rdsk/c10t0d5
# vgcreate /dev/dsk/c10t0d5 vg01
# mkdir /dev/vg01
# mknod /dev/vg01/group c 64 0x010000
# vgchange -a n vg01
# vgexport -v -p -s -m /tmp/vg01.map vg01
# rcp /tmp/vg01.map node2:/tmp/vg01.map
Node #2
-------

# mkdir /dev/vg01
# mknod /dev/vg01/group c 64 0x010000
# vgimport -v -s -m /tmp/vg01.map vg01
# vgchange -a n vg01

Node #1
-------

# cd /etc/cmcluster

# cmquerycl -C /etc/cmcluster/cmclconfig.ascii -n node1 -n node2

# vi cmclconfig.ascii

# cmcheckconf -C /etc/cmcluster/cmclconfig.ascii

# cmapplyconf -C /etc/cmcluster/cmclconfig.ascii

# rcp /etc/cmcluster/cmclconfig.ascii node2:/etc/cmcluster/cmclconfig.ascii

# cmviewcl

# cmruncl

# cmviewcl


===============================================================================
How to make package for cluster

# cd /etc/cmcluster

# mkdir db_pkg as_pkg

# cmmakepkg -p /etc/cmcluster/db_pkg/dbcrmis.conf
# cmmakepkg -s /etc/cmcluster/db_pkg/dbcrmis.cntl


# cmmakepkg -p /etc/cmcluster/as_pkg/ascrmis.conf
# cmmakepkg -s /etc/cmcluster/as_pkg/ascrmis.cntl


# vi /etc/cmcluster/db_pkg/dbcrmis.conf

# vi /etc/cmcluster/db_pkg/dbcrmis.cntl


# vi /etc/cmcluster/as_pkg/ascrmis.conf

# vi /etc/cmcluster/as_pkg/ascrmis.cntl


# cmcheckconf -v -C /etc/cmcluster/cmconfig.ascii -P /etc/cmcluster/db_pkg/dbcrmis.conf

# cmapplyconf -v -C /etc/cmcluster/cmconfig.ascii -P /etc/cmcluster/db_pkg/dbcrmis.conf

# cmcheckconf -v -C /etc/cmcluster/cmconfig.ascii -P /etc/cmcluster/as_pkg/ascrmis.conf

# cmapplyconf -v -C /etc/cmcluster/cmconfig.ascii -P /etc/cmcluster/as_pkg/ascrmis.conf


Node#2
------

# mkdir /etc/cmcluster/as_pkg db_pkg


Node#1
------
# rcp /etc/cmcluster/db_pkg/dbcrmis.* node2:/etc/cmcluster/db_pkg/dbcrmis.*

# rcp /etc/cmcluster/as_pkg/ascrmis.* node2:/etc/cmcluster/as_pkg/ascrmis.*



# cmviewcl

# cmruncl -v

# cmviewcl

# cmrunpkg dbcrmis

# cmrunpkg ascrmis

# cmviewcl

Thanks