cancel
Showing results for 
Search instead for 
Did you mean: 

ECC6 in HP- UX cluster load sharing environment

Former Member
0 Kudos

Hi All,

We are planning to install ECC6 in HP- UX cluster load sharing environment.

The cluster will be Active-Passive i.e. at any point of time in one Node both oracle and SAP will run and

in other Node, dialog instance will be running.

Did anybody try this environment? If yes, can you please clarify the following doubts?

1) How will the fail occur? I.e. consider in node A both oracle and SAP is running and in node B,

dialog instance is running. If node A fails, will oracle, SAP (central instance) and dialog instance will run in Node B?

2) What will be the starting procedure?

Regards,

Sriram.S

Accepted Solutions (0)

Answers (3)

Answers (3)

audunlea_hansen
Active Participant
0 Kudos

We are running HP ServiceGuard with SAP extention om RHEL5.2, but I don't think it's much different.

The installation use 2 packages, DB and CI. When we tested the HA-solution, we ended up with setting the CI-package depending on the DB package was up and running. This works fine for us.

Regards

Audun

Former Member
0 Kudos

Thanks Michael.

But what will be the start up procedure?

Regards,

Sriram.S

Former Member
0 Kudos

But what will be the start up procedure?

I am not completely sure, what you mean, but if you want to know how you start the cluster:

root# cmruncl

To start the dbci package, this mounts the filesystems, starts the database and the central instance:

root# cmrunpkg SIDdbci

The package start of the dialog instance will be included in the DB/CI package as well.

Best regards, Michael

Former Member
0 Kudos

Hi Sriram

We use HP Service Guard on HP-UX for clusters. There is an extension called [HP Serviceguard Extension for SAP|http://h71028.www7.hp.com/enterprise/cache/54664-0-0-225-121.html].

Normally a DB/CI Package runs on the primary server, and a dialog package on the failover node. If the primary fails the DI packages is stopped and the DB/CI package is switched to the failover node.

You will need to share the disks for the database and central instance between the nodes.

Regards Michael