
Signature
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP
Sorry, Geoff but I really don't get what you mean. As I sad, I've done
this. There could be issues reffering to AD/DNS depending on if the
netbiosname for the virtual server is an object i AD or not.
Maybe I've been a little short on info for you.
Here is the timeline:
The origninal virtual server (call it VS1) is running on it's present
hardware (called OS1 and OS2). There is two more groups in the mscs,
the cluster itself (called C1) and dtc (called D1).
Install 2 new servers with new name (NS1 and NS2).
Create a new cluster (C2)
Create a new msdtc (D2).
Verify that the ckuster is working.
Then comes the part where it's time to change who is hosting VS1.
Create login output by running sp_help_revlogin and job output by
scripting out the jobs you need to move.
Take down the virtual server cluster (VS1) resources, exept disk. That
will free the netbiosname and ipaddress.
Install a virtual server on cluster C2 with the name VS1. This is no
problem with the steps above done. When this install is done with
servicepack and pathes we have a running virtual server in the cluster
C2 with the name VS1.
After verification that VS1 can run om both new clusternodes it's time
for user data. Here I'll script in the logins created by
sp_help_revlogin, and script in the jobs to msdb. Then I'll copy the
userdatabases (or change the zoning in our SAN). Then attach, and we
are up and running. time for user verfication.
/Peter
>You should not be able to create a virtual server name that is the same as
>the existing virtual server name. AD gets confused as to who you mean and
>DNS really gets out of whack.
>I disagree, I've done this and reused the virtual server name with
> other cluster, latest in the same AD just a month ago. The differens
[quoted text clipped - 20 lines]
>>scripted so all you are doing is executing scripts, not clicking GUI
>>screens.
Geoff N. Hiten - 14 May 2008 14:34 GMT
OK. I see what you are doing. That process will work.
Most folks try and run BOTH virtual servers at the same time to do the data
synch. That won't work for reasons you have undoubtably discovered.
Again, I usually simplify the process using DNS records to point the old
name to the new server. This is especially easy if the name is already
abstracted via DNS to begin with. This has the advantage of allowing me to
have both servers up and running. I can keep my cutover downtime to 15-30
minutes this way with the right setup.

Signature
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP
> Sorry, Geoff but I really don't get what you mean. As I sad, I've done
> this. There could be issues reffering to AD/DNS depending on if the
[quoted text clipped - 64 lines]
>>>scripted so all you are doing is executing scripts, not clicking GUI
>>>screens.
Peter Lindberg - 14 May 2008 15:33 GMT
First, thanks for your wellwritten replys.
I agree, it would be perfect to have both the new and the old one up
at the same time. In this case I talked to the application vendor and
they have problems see this be done at all. They more or less said
that "we don't support it if you change VS name".
About my earlier question, can you point me to a Microsoft
article/document that said it's not supported to move systemdatabases
(between architecture)? I can't find it!
I
>OK. I see what you are doing. That process will work.
>
[quoted text clipped - 6 lines]
>have both servers up and running. I can keep my cutover downtime to 15-30
>minutes this way with the right setup.