Carlisle, D Renee wrote:
> When we upgraded everything on our master server, including hardware, os, 
> foundation suite and NBU, we built a new environment with a temporary name, 
> migrated over and tried to change to the original master server name.  In 
> effect, changing the master server name with NBU on the box.  We worked with 
> Veritas for awhile, even though it was technically unsupported.  They were 
> gracious enough to really spend time with us trying to get it to work and it 
> never did.  We ended up having to completely uninstall and reinstall NBU to 
> get the name set up correctly.  Unless you want to completely 
> reinstall...don't change the name. It really is embedded far too deep to ever 
> remove. 
> 
> 

I actually pulled this one off in my current production environment, and as a 
way to pull an exact snapshot of the catalog at a point-in-time onto my 
workstation for testing and verification. I concur it is rather hairy if you 
know what you're doing, and disasterous if you don't.

You need to:

*) snapshot the filesystem, rsync, ufsdump, or some other cold backup scheme.

*) do a complete hostname change, including DNS/NIS/LDAP and locally. If 
solaris, don't forget /etc/nodename.

*) redo all the devices, if the server is the voldb host. This includes 
cleaning 
out volmgr/database and volmgr/misc of all the lock/defs files, and possibly 
.namespace*.

*) redo the *.conf files, globally.

*) move all the media assigned to the host to the new entity.

*) move all the images assigned to the host to the new entity.

*) munge the voldb in a fashion I can't remember right at the moment -- it may 
have been as simple as using the vmdb_* utils. I know it didn't involve playing 
with the header with dd(1), like when merging voldb's in the early days of SSO.

*) consistency-check the images db.

*) Pray.

It's not something I'd suggest for the faint-at-heart, and it really helps if 
you're changing a single-server environment, especially if you're going to be 
adding-in servers, as we did.

For obvious reasons, I'm not ever going to post a recipe for public consumption 
anywhere, but anyone who followed all of that should be able to recreate the 
steps in a test lab.

Cheers,

jf