Hi all,
We are currently in the process of building a multi-site (stretched?) failover cluster for our general purpose file server. We will be moving away from a NAS appliance which has been causing us a number of issues.
Failover clusters are somewhat new to me, especially stretched ones, but I have done a lot of reading and looked into the MVA series as well as some of the new Ignite videos (Thanks Ned and Elden).
I have a few questions and am hoping someone in this channel will be able to assist.
- We will have 2 sites, with 2 nodes in each site. At this point in time, we may not have a 3rd site (another DC / cloud) available to place a witness file share in. If we don't I'm curious as to whether we should:
- Still deploy a file share witness in the primary site?
- Just have 2 nodes in each site without a witness, but define a primary DC for quorum split (is this available in 2012 R2?)
2. We will be using manual failover to the second site, so we will never want the DR site to automatically bring the cluster resource online. I believe we can do this by removing the votes from the servers in the second site, is there anything else we need to do? Is there any documentation for managing this process? Say for example that we failover to the DR site, do we also need to go in and allow both nodes in the second site to now have a vote, in case there is a problem with one of the 2 nodes in DR site while the resource is being hosted in that site? Are there any technet articles or blog posts around managing this process?
3. We use a number of DNS aliases to connect to our file shares. I've seen Elden's post (http://blogs.msdn.com/b/clustering/archive/2012/04/08/10291792.aspx) that details how to do this for a failover cluster, with a caveat that NTLM will be used instead of Kerberos. A person has commented saying that if the SPN is set correctly for the file server resource name, then kerberos will be used. Is anyone else able to confirm this?
Cheers, Matt.