On 2012-10-15 15:15, Michael Leone wrote:

Michael, I thought I read somewhere in some older documentation -- and
I'm going to do an absolutely wretched job explaining this -- but when a
new server is set up, the old CFIs from the previous server are still
property of that older server, not the clients, nor the new server. So
to recover older client indexes backed up by another server, there's a
somewhat obfuscated procedure that I *recall* having read that outlined
the solution wherein one wants to be able to recover those later down
the road to a different backup server than was originally used to back
them up -- something about the new server forever inheriting those save
sets (option 1) versus simply seeing the old server as a client (option
2). But maybe I'm confusing that with a case wherein one only wants to
recover the local data (non indexes) backed up by the previous server?
In other words, only the data that gets backed up when the server backs
itself up? But if by chance it's a matter of inheritance on those
indexes then this gets into issues of assigning the same clientid of the
old server to the new server. Or some *other* workaround, and that's
where my memory is very vague. Is it possible that this is the issue? Or
is it instead the case that if this had been the culprit then your
procedures would not have run period as opposed to running but simply
not producing the desired results?

I have done the recommended procedure - cretae a new client with the sme
client id that is on the tape; scan into the media db; run report showing
order of the volumes involved in the saveset I want; then scan again using
"scanner -i -S<ssid>". And then change the browse, retention, and
expiration dates. And those savesets are still browsable.

Did you mean 'not browsable.' ?

Regardless, I will be very curious to see what the magic pill is that
resolves this or what the hangup was that caused everything to go awry.
But let's be clear. You're not saying that your procedures overtly
failed or produced any error messages or warnings, correct? In other
words, you had no reason to believe that anything didn't work correctly
until you noticed later that you were unable to browse the CFI for the
information that should have been rebuilt by the scanner command and
added to the CFI, right? You did check the logs to see if any unusual
messages were issued aside from what you could see on the CLI?

Exactly.

What *may* have happened is that those CFIs were removed by the command
that removes the oldest NW indexes (I'm really rushed right now, so I
can't check the exact command). But if that command was run, then it's
possible that these would have been one of the indexes removed, as these
would have been one of the oldest indexes.

Yep! That could certainly have happened.

Just a related question here: what if you're scanning a whole tape, and
there's spanning save sets, but they don't all continue onto the same
subsequent tapes? What does one do in that case? You have to break it
down by individual save sets?

Scanner is supposed to be smart enough to see the ssid being scanned, and
match it with the ssid it finds in the media db, and therefore just add
the incoming info to the saveset.

I think.

:-)

--
George Sinclair
Voice: (301) 713-3284 x210
- The preceding message is personal and does not reflect any official or 
unofficial position of the United States Department of Commerce -
- Any opinions expressed in this message are NOT those of the US Govt. -