Redhat HA Cluster Functionality : Creating Separate Cluster using same set of Nodes

There was a project requirement of us to create two clusters using the same set of nodes . We are using
REDHAT Cluster Suite and it does not support creating different clusters using same set of nodes.
Initiated a thread at  Redhat mailing list for Linux HA Clusters on how to achieve the same and an
interesting discussion followed. So, thought of creating a blog article based on the discussion which goes 
below . Original discussion is available here


==================================================================
==   Original Query raised by me at Redhat mailing list for Linux HA Clusters.  ==
==================================================================
Hi All ,

Need clarifications whether it is possible to create two different cluster using the same
set of nodes. 

Looks like Redhat Cluster Suite does not support creating different clusters
using the same nodes. I am getting the following error while building the second 
cluster using the same nodes using luci interface .  


====
The following errors occurred:

    * Host system3.example.com is already a member of the cluster named "ClusterA"

    * Host system4.example.com is already a member of the cluster named "ClusterA"
===


My query is that does Redhat Cluster Suite allows in any way to create two different clusters using same nodes.  If not , any reason for not allowing  this feature?.

===========================================================
===                                                Response 1                                                === 
===========================================================

It is not possible, no. A node must be in one cluster only.
May I ask why you're trying to do this?
===========================================================
===                        Response 2                        ====                        ===
===========================================================
Thanks  for clarifying....
I am trying to create the following setup or rather will say I have been asked to do so . 
=====================================================
Cluster Name: ClusterA 
Node1: system1.example.com   
Priority:1 in Failover Domain
Node2: system2.example.com   
Priority:2  in Failover Domain
File System Resource : /data1 - An ext3 file system
 ===== 
Cluster Name : ClusterB 
Node1: system1.example.com 
Priority:2  in Failover Domain
Node2: system2.example.com  
Priority:1  in Failover Domain
File System Resource : /data2 - An ext3 file system
===========================================================
What I will achieve with this scenario , is that both the nodes will always be in active mode
as one node is having higher priority in Failover Domain  in one cluster and the other node 
has higher priority in the other  cluster as shown above.This means that both the file 
system resource will always be available in either of the nodes. And if a node goes down  ' suppose system1.example.com ' which is active in 'ClusterA' , cluster, the file system resource /data1 on the cluster will be mounted on system2.example.com which is already having /data2 mounted on  'ClusterB'  
So , based on the above architecture , we are achieving the following 
1) Both the mount points will be always available and 
2) Both the nodes will be utilized as both the nodes will be in active mode  in either of the cluster  .
Will be great if you have some inputs to achieve the same . 
===========================================================
===                        Response 3                         ====                       === 
===========================================================
Just one thought: did you intentionally decide against using GFS and favour ext3?
Regards,

===========================================================
===                         Response 4                        ====                        === 
===========================================================
Nothing intentional here . Was initially trying to check whether it is possible to create
two different clusters using same nodes and chosen ext3 for this. Is it not going to be 
a useful feature if we allow same nodes to be used in different clusters ? 
===========================================================
===                          Response 5                        ===                        === 
===========================================================
 Is it not going to be a useful feature if we allow same nodes to be used in different
       clusters ?
Even if we leave aside the fact that a lot of code would need to be rewritten and retested, 
it's a feature that would benefit very few users, at the cost of making the management of
the cluster extremely more complicated for the user and you gain nothing in HA terms. 
At the end of the day it is cheaper to buy an extra node and have two separate clusters.


 
===========================================================
===                          Response 6                      ====                          === 
===========================================================
Why not having ClusterA with two failover domains? You do not need two  clusters - just two failover domains
===========================================================
===                          Response 7                       ====                         === 
===========================================================
> Why not having ClusterA with two failover domains? You do not need two > clusters - just two failover domains. Tried as suggested by kaloyan by creating two failover
 domains instead of trying to create  two clusters. One Node (e.g system1 ) has
 higher priority in one  failover domain and the other node (system2 ) has higher priority
 in the other failover domain. With this both the nodes are in active state which we are 
 trying to achieve . Thanks All for all the replies and suggestions. 

Comments

Popular posts from this blog

PSSH : Parallel SSH to execute commands on a number of hosts

How to add check_http as a service in Nagios Monitoring using NRPE

Configuring Nagios to monitor services using NRPE