On the 13/5 I posted a question (original posting listed at the end of this
mail)
which was, in essence, the following:
I have a disk group with 9 disks (loads of volumes), 8 are active and have
valid
sub disks on, one is inactive (no subdisks). I want to remove the one
unused disk from the disk group. The problem being that this one disk has
the only lsm group configuration copy on it! Can I add a config copy to an
existing disk without reinitialising it?
Now I already pretty much knew the answer (add another disk(s) with config
copies) but thought I should ask anyhow. I received replies from the
following:
Eliahu, Ronny [Ronny.Eliahu_at_disney.com]
gguethlein_at_giantofmaryland.com
Hergert, John (PSC) [extern.John.Hergert_at_vw.com]
Davis [Davis_at_Tessco.Com]
Robert Mulley [robert_at_gnsconsulting.com.au]
Most people suggested deporting the disk group. However this is not
possible as I'm not removing the last disk in the group, just the last disk
with a configuration copy.
I have attached replies from Robert Mulley and John Hergert as they are the
most relevant in this case. I was luck in this case as I had two available
slots in another cabinet which I slotted two wide 4gb disks in, initialised
them with 2 configuration copies each of the appropriate size and added to
the disk group. At this point I could remove the original disk (with a
config copy on it).
Many thanks - Tony
============================================================================
========
As far as I am aware the only way you can get another copy is to add
another disk with a private area. If at all possible I would remove on of
the other 8 disks and re-add it making sure that I included a priv section
of at least 1M. Otherwise add back on of the removed disks (ensuring priv
section) and copy the data from one of the other disks onto it etc...
If there is another way I would be keen to hear it. I've been stuck in
similar binds before.
Robert Mulley
Unix Administrator
GNS Consulting
============================================================================
============
I ran into this a few weeks ago myself. You're in a bit of bind right now.
The only way to create a config db is at the time the disk is added to lsm.
Basically right now you need to either add more disks to the disk group and
make sure they have config db on them or free up some of the ones you are
already using, remove them and re add them. DEQ recommends that there be 4
copies of the config db. Also, the one you do have is potentially small at
764 blocks. I would recommend at least 1024 blocks, since things get pretty
ugly if you do run out of space.
John Hergert
Volkswagen of America
============================================================================
=============
+-----------------------------------------------------------------+
| TONY MILLER - Systems Projects - VODAFONE LTD, Derby House, |
| Newbury Business Park, Newbury, Berkshire. |
+-------------+---------------------------------------------------+
| Phone | 01635-507687(local) |
| Work email | ANTHONY.MILLER_at_VF.VODAFONE.CO.UK |
| X.400 | G=ANTHONY; S=MILLER; C=GB; A=GOLD 400; P=VODAFONE |
| FAX | 01635-583856 |
+-------------+---------------------------------------------------+
Disclaimer: Opinions expressed in this mail are my own and do not
reflect the company view unless explicitly stated. The information
is provided on an 'as is' basis and no responsibility is accepted for
any system damage howsoever caused.
===================original
posting================================================
All...
Here is the situation. We have an 8400 as an oracle database server. DUNIX
V4.0d. This contains a number of ESA10000 cabs with 9 and 18gb disks (a cab
has only 18's or 9's, not both).
We are replacing the 9's with 18's. The DBA moves the data from an entire
cabinet, I then do the following:
a. Delete all the lsm volumes (this is raw storage, not file systems)
d. Once there are no subdisks remaining, delete the disk(s) from its disk
group
c. Do the business on the HSZ (the detail from this point onwards is not
important.
Once the new 18's come online, they are going into A DIFFERENT disk group
from where the 9's came from.
The problem is as follows. The disk group in question had 16 disks (HSZ
based stripe sets) - that's disks not volumes.
I have happily removed 7 (of 8) disks using "voldg -g ??? rmdisk rz????".
When attempting to remove the 8th disk (there are, at this point still 9
disks within the disk group), I get a failure to the effect that "cant
remove last disk group configuration copy".
I have checked ALL the 9 disks within the group for the existence (or not)
of configuration copies. The disk in question does indeed seem to have the
last configuration copy:
#voldisk list rze144 | grep -i configs
configs: count=1 len=726
#voldisk list rzc104 | grep -i configs
configs: count=0 len=0
#voldisk list rzc112 | grep -i configs
configs: count=0 len=0
#voldisk list rzd104 | grep -i configs
configs: count=0 len=0
#voldisk list rzd112 | grep -i configs
configs: count=0 len=0
#voldisk list rze104 | grep -i configs
configs: count=0 len=0
#voldisk list rze112 | grep -i configs
configs: count=0 len=0
#voldisk list rzf104 | grep -i configs
configs: count=0 len=0
#voldisk list rzf112 | grep -i configs
configs: count=0 len=0
The new 18's are going into a different disk group.
Question - Can I add a configuration copy to an EXISTING disk with EXISTING
data on it? If so how. I've checked the archives for 98/99 and cant see
any similar problems. The LSM reference manual doesn't mention it either.
Any suggestions would be appreciated.
Thanks - Tony
Received on Mon May 17 1999 - 06:56:10 NZST