RAID-5 expansion / harddisk replacement (with bigger disks) w ithout total data-loss

Steve_Boley@Dell.com Steve_Boley at Dell.com
Tue Aug 26 16:59:01 CDT 2003


My suggestion would be to find a system to host the site temporarily while
you reconfigure the current system.
 
Doing it right the first time will save a lot of time and wasted effort in
case of the worst case happening.
 
This would give you some type of redundancy and backup plan in place as well
as be a clean migration to the
new array.  
 
You really shouldn't have a server that you seem to think is vital to your
organization, running that
doesn't have some kind of contingency plan or failover functions in place in
case of failure.
 
Think in terms of what if the arrays go down and what do you do then instead
of how do you keep it running.
Steve
-----Original Message-----
From: René Hoffmann [mailto:hoffmann at chamaeleon.de]
Sent: Tuesday, August 26, 2003 12:13 PM
To: Johnathan Conley
Cc: linux-poweredge at dell.com
Subject: Re: RAID-5 expansion / harddisk replacement (with bigger disks)
without total data-loss


That´s the problem - by various reasons we don´t have 
this second machine available.

Has anybody a detailed answer for my serious question
below ?

- P L E A S E -



Johnathan Conley wrote:


 

I have successfully used Symantec Ghost to move linux between different
systems /w different drive configurations.

 

Of course the system is down while you do this - but usually if you're 24x7
you have 2+ machines clustered/balanced and you can convert them 1 at a
time.

 

Good luck.

 

 

-----Original Message-----
From: René Hoffmann [ mailto:hoffmann at chamaeleon.de
<mailto:hoffmann at chamaeleon.de> ] 
Sent: Friday, August 22, 2003 1:00 PM
To: linux-poweredge at dell.com <mailto:linux-poweredge at dell.com> 
Subject: RAID-5 expansion / harddisk replacement (with bigger disks) without
total data-loss

 

Hi,

First of all thanks to everyone helping my colleague 
Nicolai in this particular topic.

The overall suggestion is: don't do it.

For your background information:

The system is an internet server running 24x7, 
so our goal is to reduce downtimes (btw: that's 
the primary function of RAID-redundancy, isn't it?)

On the other side the load beside office-times is
low; lower performance (e.g. caused by rebuilds)
is no problem.

It´s also ok to have at last two partitions:
one 9 GB and one 27 GB. 


In order to be able to maintain the system,  I now
would like to understand, what's the specific problem:

Is it risky:

1) to replace one 9GB disk with a 36GB ?

2) to replace more than one disk with a 36GB ?

3) to replace even the last disk with a 36GB ?

- if it´s safe to do that until now, we would 
  have a 9GB container on 36 GB disks -

4) to grow the container size ?

- if that´s the point: what could we do alternatively - 

5) create another container on the left 27GB unused space ? 



Looking foreword to your answers

Greetings




René Hoffmann
Leiter Technik
 
Chamaeleon Aktiengesellschaft
für innovative Netzlösungen
Robert-Bosch-Str. 12 / Haus IV
D-56410 Montabaur
Hotline:  +49 26 02 - 10 16 9 - 160
Zentrale: +49 26 02 - 10 16 9 - 0
Telefon:  +49 26 02 - 10 16 9 - 154
Fax:      +49 26 02 - 10 16 9 - 101
Email:  hoffmann at chamaeleon.de <mailto:hoffmann at chamaeleon.de> 
http://www.chamaeleon.de <http://www.chamaeleon.de> 






> Nicolai Schleifer < schleifer at chamaeleon.de
<mailto:schleifer at chamaeleon.de> > wrote:
> 
> > Hello,
> > 
> > I'd like to replace the four 9GB harddisks of container 1 in the example
> > below against 36GB harddisks without having a total data loss of that
> > container. The goal is to extend the overall size of that container from
"(4 time 9GB)-raid-5_overhead" to "(4 times 36GB)-raid-5_overhead".
> > 
> > I think it should be possible, by first replacing all the 9GB harddisks,
> > one by one (like if they were broken and needed to be replaced) and
after that by extending the container's size to the full extent that should
be possible by then (found hints about that, using afacli commands
"container reconfigure"/"container extend").
> > 
> > I'm currently reading this mailinglist about my planned expansion task,
> > to various different results. I can't find an example on the list which
> > matches my task. Any hints?
> > 
> > Example Configuration:
> > AFA0> container list
> > Executing: container list
> > Num          Total  Oth Chunk          Scsi   Partition
> > Label Type   Size   Ctr Size   Usage   B:ID:L Offset:Size
> > ----- ------ ------ --- ------ ------- ------ -------------
> >  0    Mirror 8.47GB            Open    0:00:0 64.0KB:8.47GB 
> >  /dev/sda             SYSTEM           0:01:0 64.0KB:8.47GB 
> > 
> >  1    RAID-5 25.4GB       64KB Open    0:02:0 64.0KB:8.47GB 
> >  /dev/sdb             DATA             0:03:0 64.0KB:8.47GB 
> >                                        0:04:0 64.0KB:8.47GB 
> >                                        0:05:0 64.0KB:8.47GB 
> > 
> > Maybe someone can tell me if it's possible at all. A recipe on how to do
it would even be greater.
> > 
> > Thank you! :)
> > 
> > 
> > Greetings Nicolai
> > -- 
> > Nicolai Schleifer
> > Chamaeleon Aktiengesellschaft
> > fur innovative Netzlosungen
> > Robert-Bosch-Str. 12 / Haus IV
> > D-56410 Montabaur
> > Hotline:  +49 26 02 - 10 16 9 - 160
> > Zentrale: +49 26 02 - 10 16 9 - 0
> > Fax:      +49 26 02 - 10 16 9 - 101
> > Funk:     +49 177 - 32 08 60 7
> > Email:  schleifer at chamaeleon.de <mailto:schleifer at chamaeleon.de> 
> >  http://www.chamaeleon.de <http://www.chamaeleon.de> 






 
 


-- 

René Hoffmann

Leiter Technik



Chamaeleon Aktiengesellschaft

für innovative Netzlösungen

Robert-Bosch-Str. 12 / Haus IV

D-56410 Montabaur

Hotline:  +49 26 02 - 10 16 9 - 160

Zentrale: +49 26 02 - 10 16 9 - 0

Telefon:  +49 26 02 - 10 16 9 - 154

Fax:      +49 26 02 - 10 16 9 - 101

Email:  hoffmann at chamaeleon.de <mailto:hoffmann at chamaeleon.de> 

http://www.chamaeleon.de <http://www.chamaeleon.de> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.us.dell.com/pipermail/linux-poweredge/attachments/20030826/8a9409a4/attachment.htm


More information about the Linux-PowerEdge mailing list