Thanks to all for the information this will help prepare backup and
recovery processes moving forward:

First, we do not count on Full pack backups for our restores normally we
use FDRUPSTREAM to backup the guests file systems on a daily and weekly
basis, however the guests are still running when this is done other than
the Data Base backups where the guest is shutdown and a full backup is
taken of the Oracle data base using FDRUPSTREAM. Since we just started
adding the z/VM and z/Linux environment as part of the DR testing we are
still feeling things out. For instance this will be the first time that
we will be taking FDRUPSTREAM with us. The last DR test which was the
first one we restored one guest all ECKD DASD from FULL DFDSS backups
running on z/OS. 

Second, for this exercise there is no LVM involved these are just ECKD
devices. So synchronization will be less of an issue for these guests
and for this test.

Thank You,

Terry Martin
Lockheed Martin - Citic
z/OS and z/VM Performance Tuning and Operating Systems Support
Office - 443 348-2102
Cell - 443 632-4191


-----Original Message-----
From: The IBM z/VM Operating System [mailto:[email protected]] On
Behalf Of David Boyes
Sent: Monday, August 16, 2010 10:51 AM
To: [email protected]
Subject: Re: How to copy a disk using a z/Linux guest

On 8/16/10 10:37 AM, "Macioce, Larry" <[email protected]>
wrote:

> What about Bacula? Dr Boyes could speak could speak on this point
better
> And you always want to shutdown the guest to get a good backup

Long term, setting up Bacula would probably be a Good Thing, but
probably
doesn't help much in this scenario, other than to give you a convenient
way
to dump and restore the contents of the filesystem onto a new disk that
doesn't have the missing VTOC problem.

The question in my mind is whether the last cylinder of the pack is
actually
in use and/or whether the filesystem could be resized online to be 180K
smaller (so as to free up the last cylinder). DDR is smart enough to be
able
to relocate the disk image (eg, restore it to cyl+1 on another disk),
but
that won't fix up any internal references to the filesystem size or LVM
magic that might be there, and you will get Weird Failures that you
really
don't want to try to track down. If it can't be resized or is part of a
LVM,
then you're SOL on that solution.

Ultimately, if you want to keep using DFDSS (and I'll point out that you
have been EXTREMELY lucky if you haven't had failures of your backups)
the
solution is to allocate a new pack with a proper VTOC in cyl 0, and
follow
the instructions on linuxvm.org to migrate the filesystem from the
current
disk to the new disk. Short of that, you'll have to use DDR for those
two
packs (if you use CMSDDR, you can store those two disks as CMS files on
one
of the other packs, and then restore them back to real DASD from there).
That's probably the only way you'll be able to do this quickly enough
for
your DR test. Larry's suggestion of Bacula is a better long term
solution,
but that takes a bit of planning you probably don't have time to do now.

-- db

Reply via email to