Cluster setup

[bchinnat]

Hi,

I am very new to the RHEL cluster. We have the requirement to setup RHEL 2 node cluster setup. Cluster setup is fine but I like to know how to configure applications in cluster ware. Any suggestion would be much appreciated..

[S.M.Athar]

Which application you want to use for clustering? and what is the RHEL version? If you are using the customized application then should use your own start/stop script. Place the script in /usr/share/cluster.
First add the IP resource and go the script resource to give the path of your customize script.
After creating the resources now you have to create the cluster service and add these two resource in it.
Thanks
Athar

[bchinnat]

Thanks S.M.A

The application would be weblogic, but can I have any document for allocating resources..

Thanks in advance

[S.M.Athar]

For documentation plz check
http://docs.redhat.com/docs/en-US/index.html

Regards
Athar

[Matti_Kurkela]

There probably won't be a general One True Way to install Weblogic on a RedHat cluster, because both Weblogic and RedHat Cluster are very flexible.

There are many ways to install Weblogic: you should first plan a Weblogic configuration that fulfills the requirements of your Java application(s), then plan a RedHat Cluster configuration that fulfills the requirements of both Weblogic itself and the application(s) it's supposed to run.

Once planning is completed, you can begin setting it up in the reverse order: first the RedHat Cluster configuration, then install Weblogic on the cluster resources, then application(s) on top of Weblogic.

The simplest setup would be to create one clustered VG, with one or more LVs with filesystems for Weblogic binaries and data, one service IP address and one script for starting/stopping Weblogic.

In RedHat cluster terminology, this would be a single cluster service, with one script resource, one IP resource and one or more filesystem resources.

If you want the cluster to monitor that the Weblogic keeps running (and restart it if it fails, and to trigger a failover to the other node if it keeps failing), you may have to write your own resource script, using one of the existing resource scripts as a template.

If you use CLVM locking (as is the default on RHEL 5 Cluster), there is no need to define LVM VG resources. If you don't want to use CLVM, see RedHat Knowledge Base article DOC-3068 (RedHat Network access required):

https://access.redhat.com/kb/docs/DOC-3068
(If you're familiar with HP Serviceguard on Linux, this alternative LVM locking scheme should be familiar to you.)

MK