The reason I wrote this up was that it happened 
to me a couple of times.  The first time I found out 
about it was one morning, on a 3 tape parallel backup.
I noticed the backup had not finished, I did a RECALL 
and it wanted another tape.  "O.K. this is expected to
happen once in a while."   So I put in the tape that 
it wanted. 

Methinks it was happy, until a minute later it says
at the console it wanted a tape in the 2nd drive.
O.K. I says again to myself, here's another.

A minute later it wants a tape in the 3rd drive!
Finally after putting in 3 new tapes, it RoadRunner
starts writing.  

It wrote for a total of 5 minutes, then the job
completed.

The part that really hurt was it made my tape filing
system unusual.  Instead of the tapes being marked
1/3, 2/3 and 3/3; I had to mark them 1/2 of 1/3;
1/2 of 2/3, 1/3 of 3/3 the 2/2 of 1/3 then 2/2 of
2/3 and finally 2/2 of 3/3!.


Worst case was 3 drives,
it wanted another tape, so I did a recall, and
it wanted a tape on

BT
NNNN
Tracy Johnson
MSI Schaevitz Sensors 

> -----Original Message-----
> From: John Clogg [mailto:[log in to unmask]]
> Sent: Monday, May 06, 2002 2:21 PM
> To: [log in to unmask]
> Subject: Re: [HP3000-L] Thinking about a new tape drive
> 
> 
> Tracy's description of RoadRunner's behavior does not match 
> my experience
> with the product.  In fact it is quite rare for multiple 
> volumes to fill up
> at the same time.  We used RoadRunner for a number of years in an
> environment with two tape drives.  Three-tape backups were 
> quite common.
> 
> I believe the way it works is that RoadRunner/Backpack 
> determines at the
> outset which files will go to which drives.  If all the files 
> destined for
> one drive can fit on one volume, it won't ask for another one 
> on that drive,
> even if the other drives require additional tapes.  In our 
> case, we were
> using the highest level of data compression, so the total 
> amount of data
> written to each device was difficult to predict, often 
> resulting in uneven
> distribution of the load among drives.  In a situation where 
> the volume is
> more predictable, I suppose the behavior Tracy describes is 
> more likely.
> 
> -----Original Message-----
> From: Johnson, Tracy [mailto:[log in to unmask]]
> Sent: Monday, May 06, 2002 11:05 AM
> To: [log in to unmask]
> Subject: Re: Thinking about a new tape drive
> 
> 
> The only drawback to performing the Parallel method, is that you must
> monitor of how much tape is used over time.
> 
> This is because when it goes *over-the-edge* (at least in
> BackPack/RoadRunner) so it needs to write to more than the 
> number of tapes
> specified, you're required to double the number of tapes used.
> 
> i.e,
> 
> If you write parallel to 2 tapes and the job wants more, even 
> if there is
> only one more byte to go, you must insert 2 more tapes.
> 
> If you write parallel to 3 tapes and the job wants more, even 
> if there is
> only one more byte to go, you must insert 3 more tapes.
> 
> <snip>
> 
> * To join/leave the list, search archives, change list settings, *
> * etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
> 

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *