On Thu, 2005-09-01 at 14:58, Ian Batten wrote:
> Imagine I'm backing up a machine.  The backup is taking place as part  
> of an autostarted group.  The pool that is selected on the basis of  
> the hostname, group and level is disk based.   When the group  
> completes, an incremental backup is taken of the indexes, which goes  
> to tape.  We now have the savesets on disk, with their indexes on tape.
> 
> At some later point, either based on the age of the saveset or the  
> capacity of the disk volume, the savesets are staged to tape.   The  
> savesets that end up on tape have the same ssid as those that  
> originally went to disk, and therefore reference the same items in  
> the indexes, which has gone to tape.  However, only the media  
> database will know that the savesets are now on tape.   By  
> observation, no metadata goes to tape at this point.  The clones that  
> were on disk are then deleted.
> 
> So the timeline is:
> 
> Perform backup to disk.
> 
> Write indexes to tape.
> 
> Write savesets to tape.
> 
> Delete savesets from disk.
> 
> What would happen if my media database were to be damaged after the  
> savesets have been deleted from disk?  The only thing that knows that  
> the savesets are on tape is the media database, and I only have one  
> copy of that.  Which just got destroyed.  The only way to find those  
> savesets again is to get the ssid from the copy of the media database  
> which was written to tape after the original save to disk, and then  
> trawl through all the media that could possibly have been near a tape  
> drive in the intervening period looking for savesets.
> 
> I think that the media database should be backed up to tape after the  
> savesets have moved from disk to tape, but before they are deleted  
> from their source device.
> 
> Or am I missing something?
> 

No you're not missing anything, but there are very many instances when the 
media database on disk is newer than the backup of it on tape. Running any 
backup that doesn't include the bootstrap will create this situation, and any 
time that a backup is in progress the same could be said. So what is the 
solution? The only way you could get round this is to backup the bootstrap more 
often. There is nothing at all to prevent you doing this. I don't think most 
people would see this as a problem, but it all depends on your level of 
paranoia.

To sign off this list, send email to listserv AT listserv.temple DOT edu and 
type "signoff networker" in the
body of the email. Please write to networker-request AT listserv.temple DOT edu 
if you have any problems
wit this list. You can access the archives at 
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER