Dear Guy,


On Monday 10 January 2005 11:40 am, you wrote:

First, use mdadm, not the obsolete raidtools!

You could create 3 arrays, each with 1 disk missing, then add/remove the
extra disk where needed.  Your system and filesystems would be up and
running during the add and remove, and all other times, if you want.

The disks should be partitioned.  Partition type FD (RAID autodetect).
If you have data on these drives that you want to keep, please explain.

To create the arrays:
mdadm -C /dev/md0 -l1 -n2 /dev/hda1 missing
mdadm -C /dev/md1 -l1 -n2 /dev/hdc1 missing
mdadm -C /dev/md2 -l1 -n2 /dev/hde1 missing

Create your ext3 filesystems on /dev/md0 md1 and md2.

To add the extra disk:
mdadm /dev/md0 -a /dev/hdg1

A re-sync or re-build will start.  But your filesystem is still usable.

To remove the extra disk:
First fail it:
mdadm /dev/md0 -f /dev/hdg1
Then remove it.
mdadm /dev/md0 -r /dev/hdg1

Now you can add it to another array.

Just be sure to not fail a disk if the other disk in the array has failed!
Or, game over!

Guy

thank you very much, for your detailed guide to using mdadm.
I now understand it more. I will try to use it from now on. I must review how 
a  nice configuration file gets created by these actions so that I can keep 
track of what I did on the machine (similar to /etc/raidtab which is easy to 
review).

PS. I like the game over bit :) !

I have practiced last night with using raidsetfaulty and raidhotadd and 
raidstart and raidstop and thesee seem analogous to that.

I will work to convert over to mdadm.

Mitchell

-----Original Message-----
From: linux-raid-owner@vger.kernel.org
[mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Mitchell Laks
Sent: Monday, January 10, 2005 2:03 AM
To: linux-raid@vger.kernel.org
Subject: convert degraded raid1 drive to ext3 format

Hi,

Only a newbie could ask this question. Guess what, I really am!

I have 4 hard drives on my ssytem to store data. currently i use raid5.

I am seriously considering using raid1 instead of raid5.
But  I only need raid1 while data is being initially collected on the
server.
So  I would like to use raid1 (with ext3 format) initially and then convert
to standard ext3 drives.if this is possible.

i have 4 drives
/dev/hda, /dev/hdc, /dev/hde, /dev/hdg

What I would like to do is use
stage 1)     /hda and /hdg as  a raid initially and leave /dev/hdc,
/dev/hde

alone
then use
stage 2)    /hdc and /hdg as a raid and use "/hda as if it were a ext3
drive"
then use
stage 3)   /hde and  /hdg as a raid and use "/hda and /hdc as if there were
ext3 drives"

Now i see easily that I can run at stage3)

/dev/md0 --      /hda alone in degraded mode
/dev/md1 ----  /hdc alone in degraded mode
/dev/md2 ---    /hde and /hdg together.

if i do raidsetfaulty to /dev/hdg and raidhotremove /dev/hdg each time and
add
it to the new /dev/md1 and /dev/md2 as I need them.

But i would prefer REAL ext3 drives instead of degraded /dev/md0 and
/dev/md1
(i think???). In place of /dev/md0 and /dev/md1.

So my question is: I s it possible to "convert a degraded raid1  consisting
of
a single hard drive over to a standard ext3 format drive (ie /dev/md0 is
only
one drive so can it become ext3) ?" without doing something like rsync of
the
data on /dev/md0  to a fresh drive "the new /dev/hda" ???

Thanks,
Mitchell Laks


-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html