Yes, for backup as in using flashcopy to backup a set of drives to
another, you are correct.

However, my comments was for the receiving drives.

In this case, I'm flashing a Linux drive (virtual address) 150 to
(virtual address) 1150.
To restore back, I take down the Linux machine and then flash 1150 to
150.
If CP isn' aware of what happened, any MDC contents for 150, should no
longer be trusted as the DS6800 hardware function, refreshed 150 with
the contents of 1150.

Going a little deeper, I assume that the DS6800 will invalidate the
cache in the DS6800 controller also.

Tom Duerbusch
THD Consulting

>>> [EMAIL PROTECTED] 12/6/2005 11:56 AM >>>
I don't understand Alan's reply.  Isn't the mdisk cache always a R/O 
cache.  Aren't all dasd writes done to the dasd and not to the MDC? 
Isn't 
FLASHCOPY going to reflect the real contents of the dasd at the time of
the 
copy.  I know that there can be differences between what is actually 
written on a mdisk directory and what the writer of the mdisk has in
memory 
until the disk is released, but that isn't a MDC consideration, I don't
think.
Jim

At 10:26 AM 12/6/2005, you wrote:
>On Monday, 12/05/2005 at 11:36 CST, Tom Duerbusch
><[EMAIL PROTECTED]> wrote:
> > I have been using flashcopy for some interm backups that I restore
from,
> > also using flashcopy.
> >
> > So, if I have been using device 150, and I do a flashcopy from 1150
to
> > 150, what happens to the MDC contents for 150?
>
>If you use the CP FLASHCOPY command or a guest initiates a "virtual"
>flashcopy I/O, CP will handle the cache.
>
>If you go behind CP's back to do the flashcopy from the disk
controller
>console then you are in trouble.  The controller does not notify the
host
>of "background" activity on a volume and the cache will not be valid. 
So,
>if you use the controller console be sure to SET MDC FLUSH <vdev>
before
>you start reading/writing the minidisk again.
>
>Obviously it is far better to initiate a flashcopy from CP.
>
>Alan Altmark
>z/VM Development
>IBM Endicott

Jim Bohnsack
Cornell Univ.
(607) 255-1760

Reply via email to