Solved

Cluster query

Posted on 2008-11-01
9
261 Views
Last Modified: 2008-11-02
Hi

Let's say I have an Active-Passive Windows 2003 cluster. ServerA and ServerB are the physical boxes, ServerV is the virtual server that can reside on either.

If I make changes to the registry of ServerV, do I need to make changes to ServerA and ServerB as well, or are these changes made on the Quorom so therefore there is only a need to make changes on ServerV?

Or do I need to fail over ServerV to the passive node, make the reg. changes to the former active node, then fail over again and then make changes to the other one?

Hope someone can help...
0
Comment
Question by:bruce_77
  • 4
  • 3
  • 2
9 Comments
 
LVL 83

Expert Comment

by:oBdA
ID: 22857566
"servers" in a cluster are just *network* *names*, as the name of the cluster resource indicates; in other words: these "servers" have no registry.
When accessing such a network name, you'll obviously end up on the physical node owning the resource.
0
 
LVL 8

Expert Comment

by:smilerz
ID: 22857591
Typically, when you make changes you have to make them to each physical node.  The exception to this is if you are making changes to a 'cluster aware' application - then the application will typically take care of making the changes 'portable.'

If you are making changes directly to the registry though, you are going to make changes in both places.  You shouldn't need to fail over to accomplish this though.
0
 
LVL 2

Author Comment

by:bruce_77
ID: 22858428
Thanks guys - I think I am just trying to get my head round clusters really.

So, if I make a reg. key change, then it has to be done on the two physical servers. What happens if I make it only on one - will the changes only take effect when the virtual server is on the node with the changes made?

And what if I make the changes on the virtual server itself rather than the two physical boxes?

0
 
LVL 8

Expert Comment

by:smilerz
ID: 22858750
If you only make the change in one place if the virtual resources failover the change will not exist on the second node.
The virtual server doesn't really exist there are just two physical servers with some shared resources. Though there are some shared settings, you would have to provide specific examples on what you are needing to change to decide if changing it in one place is sufficient.
0
Comprehensive Backup Solutions for Microsoft

Acronis protects the complete Microsoft technology stack: Windows Server, Windows PC, laptop and Surface data; Microsoft business applications; Microsoft Hyper-V; Azure VMs; Microsoft Windows Server 2016; Microsoft Exchange 2016 and SQL Server 2016.

 
LVL 83

Assisted Solution

by:oBdA
oBdA earned 500 total points
ID: 22859295
Again, as I said above: there is no "virtual server itself". The network name resource is exactly that: a network name that the node that the resource curently owns will react to; nothing more. Every access to this name will be handled by the physical node owning the resource. That's why you need the shared storage when building a cluster: to have a drive where information can be stored that will "wander" from owner to owner.
A cluster is not about virtualizing hardware and then installing an independent OS into this virtualized hardware; this are two completely different concepts if virtual servers. Cluster resources are *very* dependent on the node OS, because it's the *node* that handles requests to the virtual network names.
0
 
LVL 2

Author Comment

by:bruce_77
ID: 22860858
Ok, I think I get it :)

Many thanks! Just before I assign the points, I hope you don't mind if I just make sure I have things clear in my head.

So, in my situation, ServerV is not some "floating" virtual (I understand this has nothing to do with virtual in the sense of Vmware) server that can reside on ServerA and ServerB. ServerB is a network name that can point to either ServerA or ServerB at any point in time? Therefore, any changes should really be made on ServerA or ServerB, not ServerV, unless they are for applications that can port the changes over (i.e. cluster aware).

The quorom is responsible for deciding which node is active and for handling fail over situations? For some reason, I had the assumption that the quorom actually held information relating to the "virtual" server (e.g. settings). This is incorrect?

Thanks again,
0
 
LVL 2

Author Comment

by:bruce_77
ID: 22860863
Sorry, typo in my above post

"ServerB is a network name that can point to either ServerA or ServerB at any point in time"

should read

"ServerV is a network name that can point to either ServerA or ServerB at any point in time"
0
 
LVL 83

Accepted Solution

by:
oBdA earned 500 total points
ID: 22861152
"ServerV" is a network name generated by the cluster; it will be pointing to the one single node that owns the resource group that the network name resource is in. So it's not a random and/or distributed access like in NLB; the resource group can only be owned by one node at a time, and this is the physical node that will handle the requests.
So if you need to make a registry change that is required for accessing ServerV successfully, this change has to be made on all nodes that can own the resource group.
The quorum holds the cluster database, and in case of communication problems of the cluster nodes amongst each other, the node which owns the quorum resource group is the one that will actually "be" the cluster; the other nodes won't bring resources online until they can communicate with the node owning the quorum again.
Server Clusters Technical Reference
http://technet.microsoft.com/en-us/library/cc759014.aspx
0
 
LVL 2

Author Comment

by:bruce_77
ID: 22863572
Thanks guys, excellently answered!
0

Featured Post

Zoho SalesIQ

Hassle-free live chat software re-imagined for business growth. 2 users, always free.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

A quick step-by-step overview of installing and configuring Carbonite Server Backup.
Learn about cloud computing and its benefits for small business owners.
A short film showing how OnPage and Connectwise integration works.
With Secure Portal Encryption, the recipient is sent a link to their email address directing them to the email laundry delivery page. From there, the recipient will be required to enter a user name and password to enter the page. Once the recipient …

932 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

11 Experts available now in Live!

Get 1:1 Help Now