disk issue in PVG

[Sreer]

Hi Gurus,

I have to replace a failed disk from my cluster node.
it is configured in a PVG.

all the details are attached.

Iam not sure whether the failded disk is already mirrored or stripped...

My PVG experience is well low....

Since i need to do in a production bOX ..could you pls guide me how to proceed on it?

What I guess is the data is already mirroed as well as in striped mode ? is it so?


If so can I go ahead with reducing the lv from the failed disk and reduce the failed disk from concerned VG?

ca01/root> swlist -l bundle | grep -i QP
ca01/root>


I understand the Qpacks are not installed in this server so the OLR facility and the pvchange may not work..


Since its a PVG related one " when good disks fail" is not helping me much..

Thnx for your help..


rgds
Sree

[Torsten.]

As said in your other thread - This still applies here:


When_Good_Disks_Go_Bad_WP
http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c01911837/c01911837.pdf

no matter if the disk is in PVG or not.

- deactivate (pvchange -a N ...)
- replace
- vgcfgrestore
- sync

all from active node.

[Sreer]

Hi Torsen ..

Thanks for help..

But still iam afraid ..mainly because OLR facility is not available in my server even though its 11.23.

So I may need to reduce the failed disk from mirror...iam getting confused here..
since many disks are invilved in striped way in this LV..if how can I reduse a particular disk from this LV by using lvreduce -m 0 lvname pvname?

will it affet any other mirrors lying in other disks?


Once again to remind yo that its striped onE!!!!

LE PV1 PE1 Status 1 PV2 PE2 Status 2
00000 /dev/dsk/c4t2d0 00000 current /dev/dsk/c5t2d0 00000 current
00001 /dev/dsk/c4t3d0 00000 current /dev/dsk/c5t3d0 00000 current
00002 /dev/dsk/c4t4d0 00000 current /dev/dsk/c5t4d0 00000 current
00003 /dev/dsk/c4t5d0 00000 current /dev/dsk/c5t5d0 00000 current
00004 /dev/dsk/c4t6d0 00000 current /dev/dsk/c5t6d0 00000 current
00005 /dev/dsk/c4t7d0 00000 current /dev/dsk/c5t7d0 00000 current
00006 /dev/dsk/c4t8d0 00000 current /dev/dsk/c5t8d0 00000 current
00007 /dev/dsk/c4t9d0 00000 current /dev/dsk/c5t9d0 00000 current
00008 /dev/dsk/c4t10d0 00000 current /dev/dsk/c5t10d0 00000 current
00009 /dev/dsk/c4t11d0 00000 stale /dev/dsk/c5t11d0 00000 current
00010 /dev/dsk/c4t12d0 00000 current /dev/dsk/c5t12d0 00000 current
00011 /dev/dsk/c4t13d0 00000 current /dev/dsk/c5t13d0 00000 current
00012 /dev/dsk/c4t14d0 00000 current /dev/dsk/c5t14d0 00000 current
00013 /dev/dsk/c4t2d0 00001 current /dev/dsk/c5t2d0 00001 current
00014 /dev/dsk/c4t3d0 00001 current /dev/dsk/c5t3d0 00001 current
00015 /dev/dsk/c4t4d0 00001 current /dev/dsk/c5t4d0 00001 current
00016 /dev/dsk/c4t5d0 00001 current /dev/dsk/c5t5d0 00001 current
00017 /dev/dsk/c4t6d0 00001 current /dev/dsk/c5t6d0 00001 current
00018 /dev/dsk/c4t7d0 00001 current /dev/dsk/c5t7d0 00001 current
00019 /dev/dsk/c4t8d0 00001 current /dev/dsk/c5t8d0 00001 current
00020 /dev/dsk/c4t9d0 00001 current /dev/dsk/c5t9d0 00001 current
00021 /dev/dsk/c4t10d0 00001 current /dev/dsk/c5t10d0 00001 current
00022 /dev/dsk/c4t11d0 00001 stale /dev/dsk/c5t11d0 00001 current
00023 /dev/dsk/c4t12d0 00001 current /dev/dsk/c5t12d0 00001 current
00024 /dev/dsk/c4t13d0 00001 current /dev/dsk/c5t13d0 00001 current
00025 /dev/dsk/c4t14d0 00001 current /dev/dsk/c5t14d0 00001 current
00026 /dev/dsk/c4t2d0 00002 current /dev/dsk/c5t2d0 00002 current

rgds
Sree

[Manix]

Hello Sree !!

of course the best way to do this is to
use "OLAR" ,pvchange , & vgcfgrestore as
per the standard process.

But if you need to break the PVG mirrors
then you may do as follows, to check the
disk in respective PVGs see /etc/lvmpvg

1) lvreduce -m 0
2) lvextend -m 1

Hope this help.

"when good disks go bad" is a must to read doc for any sys admin , do it when possible.

Thanks

Manix

[Torsten.]

You will get much deeper in trouble if you reduce anything due to the PE distribution - maybe, maybe not.


In theory the best solution would be to patch the other node, switch the package and do the replacement. I assume you did not patch the system for 4 years or more.


IMHO the "don't change a running system" is a bad advise. Compare this to your car, firmware and patches are the oil, brakes and tires, they really need maintenance from time to time ... ;-)



Since the disk is already in unavailable status, I would consider to just pull it, but this is up to you.

[Sreer]

Hi All,

Just updating the progress of this case .
I reached in workaround as:
reduce the failded disk from the affected LV.
its possible since the disk is dedicated for this LV only

so I may be using it as:

lvreduce lvname pvname

Hope it may reduce the disk from LV and data is already mirrored well.

I will update you once the task is completed.

will update the forum with out fail.

thnx & rgds

Sree