Keith Gunderson wrote:
> ----- Forwarded Message ----
> From: Keith Gunderson <krgkg AT yahoo DOT com>
> To: bacula-users AT lists.sourceforge DOT net
> Sent: Tuesday, March 31, 2009 11:55:10 AM
> Subject: [Bacula-users] Changing file-daemon hosts while keeping
> existing jobs, filesets without doing FULL backup
>
>
>
> Hi There,
>
> I've been unsubscribed to bacula-users for a number of months...  no
> issues is good, right?  But my apologies if this has been asked a lot. 
> I didn't find the answer by searching.
>

I just did a some testing and figured it out for myself.  Nothing that follows is recommended (by me) unless you're comfortable with db-admin tasks, sql, and probably C so that you can look into bacula source code to confirm your suspicions about how things work.  But actually, doing what follows was pretty easy.


> We're running bacula 2.2.8 on Debian.  I'd like to replace our main
> fileserver without having to do a full backup of the jobs immediately
> after the switch.

Just to clarify, our fileserver is just computer running bacula-fd.

>
> We have about 25 TB of data on some FC disk arrays.  Our filesystems
> are xfs and we use LVM for some nice flexibilty.  We'll unplug our old
> server from the FC switch and plug our new server in.  Mount the
> filesystems and go from there.  

This worked really well.  Copying the /etc/lvm configuration from old to new is what's needed.

>
> What is my best bet for getting incrementals to work the next evening?
>
> 1) Alter the client table.  (I.e. add a new client entry, immediately
> stop bacula, manually change the two client entries in database table
> so that the new server has the old servers clientID.) 2) Alter the
> jobs table (i.e. alter the clientID entry for every job row so that
> the new servers clientID is listed where the old server used to be.)
> 3) Buy new tapes dude, you're screwed.  If so, why?  
>
> Thanks!
> Keith

Option 2 was the path I took. After backing up the bacula config and db to my hearts content, I configured the new client and ran a simple backup.  Then I looked into the client table and saw what I wanted.  Then I ran:
update job set clientid = 14 where clientid = 2; (Your query may be different and involve more in the where clause depending on your setup.)

Then I updated the jobs & jobdefs so that the new fileserver replace the old fileserver. 

Now its like the old filesystems are on a new server, and I can do incremental backups immediately.  If you want to do this, make sure you test it all yourself.

Thanks!
Keith