3+ Node CentOS5 Cluster + SoftSAN: Difference between revisions

From Alteeve Wiki
Jump to navigation Jump to search
(Created page with '{{ht_header}} This is a new How-To and is not finished. Until this warning is removed, consider this document very flawed and unsuitable for use. = Overview = The goal of this…')
 
mNo edit summary
Line 1: Line 1:
{{ht_header}}
{{howto_header}}


This is a new How-To and is not finished. Until this warning is removed, consider this document very flawed and unsuitable for use.
This is a new How-To and is not finished. Until this warning is removed, consider this document very flawed and unsuitable for use.

Revision as of 05:01, 13 October 2009

 AN!Wiki :: How To :: 3+ Node CentOS5 Cluster + SoftSAN

This is a new How-To and is not finished. Until this warning is removed, consider this document very flawed and unsuitable for use.

Overview

The goal of this How-To is to build a 3+ node cluster where each node's disk space comes from a home-brew SAN slice.

Specifically, this is an extension to the 2-Node CentOS5 Cluster How To. In practice, the SAN would be a stand-alone 2-node cluster to provide high availability of the SAN to the main cluster, but to keep things simple, and to reduce the hardware needed for you to follow along, it will be a simple disk in this How-To.

The cluster itself will comprise of four nodes. I will start with three node and then show how to add in the fourth to simulate growing your cluster as your resource requirements grow.

What You Should Know

I will try to assume as little as possible in this How-To. However, a certain base-level of knowledge is expected.

Specifically, you will need to know;

  • The bash command line.
  • Your way around CentOS or Red Hat.
  • Basic RPM package management.
  • An understanding of TCP/IP networking and network configuration under CentOS
  • Data storage under Linux (fdisk, md, LVM, etc)

The SAN

The SAN will be the centralized data store all cluster nodes will use. The idea is that a node can come or go and another node can slide into a dead node's place easily by simply accessing the dead node's data on the SAN. It also simplifies allocating various amounts of disk space to nodes as their storage needs change over time.

KISSing the SAN

As said earlier, you would probably want to build a SAN server that has as few single points of failure as possible. A reasonable and reasonably inexpensive approach would be a 2-node cluster with each node having two dedicated NICs to two seperate switches dedicated to data traffic.

For the case of this paper though, I will use a single server with a single '/dev/sdX' device. You should be able to simply replace '/dev/sdX' with '/dev/drbdX' in practice. I recommend though, for your first attempt, keep it simple. Get things working and then "do it right".

Remember, "Keep It Simple, Silly"!

OS Install

Start with a basic CentOS install. If you'd like to mimic the install used for this paper, here is my 'anaconda-ks.cfg', renamed as 'san01.ks' for you to use as a kickstart script. Before you can use it though, edit it to set your hostname and to use your password hash.

The main thing to remember in your OS install is to keep it simple. I generally start with as minimal install as possible and install packages I need via RPM as I go.


 

Any questions, feedback, advice, complaints or meanderings are welcome.
Alteeve's Niche! Enterprise Support:
Alteeve Support
Community Support
© Alteeve's Niche! Inc. 1997-2024   Anvil! "Intelligent Availability®" Platform
legal stuff: All info is provided "As-Is". Do not use anything here unless you are willing and able to take responsibility for your own actions.