Good day :-)

I've 8 sata disk on raid 6 software (debian stable).

For purpose test I stopped 3 disks of the array.

Now I rebooted and I tring to re-assemble the array but without successful..

mdadm -A /dev/md0
mdadm: /dev/md0 assembled from 5 drives and 3 spares - not enough to
start the array.

cat /proc/mdstat

Personalities :
md0 : inactive sdb[0](S) sdi[10](S) sdk[9](S) sdj[8](S) sdf[4](S)
sde[3](S) sdd[2](S) sdc[1](S)
3907091968 blocks

unused devices: <none>

Consider the data there are on the raid, and there aren't any disks
failure, is there a way to re-assemble the same raid and maintan the data?

Thanks
Pol


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Wed, Jan 30, 2008 at 01:52:49PM +0100, Pol Hallen wrote:
> I've 8 sata disk on raid 6 software (debian stable).
>
> For purpose test I stopped 3 disks of the array.
^^^^
NOTE WELL: The md(4) man pages says that raid6 can handle the failure of
any 2 disks. You pulled 3. Good way to hose the array.
>
> Now I rebooted and I tring to re-assemble the array but without successful..
>
> mdadm -A /dev/md0
> mdadm: /dev/md0 assembled from 5 drives and 3 spares - not enough to
> start the array.
>
> cat /proc/mdstat
>
> Personalities :
> md0 : inactive sdb[0](S) sdi[10](S) sdk[9](S) sdj[8](S) sdf[4](S)
> sde[3](S) sdd[2](S) sdc[1](S)
> 3907091968 blocks
>
> unused devices: <none>
>
> Consider the data there are on the raid, and there aren't any disks
> failure, is there a way to re-assemble the same raid and maintan the data?

Considering that this is a test, there should be no concern over data.
You did a backup prior to this test?

You hosed the RAID6 array by pulling 3 disks. The spares only if a
spare disk can get syncd before a third disk died.

So, no, you have to make a new raid and restore from backup.

Doug.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

> NOTE WELL: The md(4) man pages says that raid6 can handle the failure of
> any 2 disks. You pulled 3. Good way to hose the array.
Is failure the same of disconnession of disk?

I thinked there is it so. Because if the disk crash the all data are lost, but
I if disconnect the disk there are date yet.

> Considering that this is a test, there should be no concern over data.
> You did a backup prior to this test?
Yes :-)

> You hosed the RAID6 array by pulling 3 disks. The spares only if a
> spare disk can get syncd before a third disk died.
> So, no, you have to make a new raid and restore from backup.
Is there another way to increase the security data?

Thanks
Pol


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

On Wed, Jan 30, 2008 at 05:36:35PM +0100, Pol Hallen wrote:
> > NOTE WELL: The md(4) man pages says that raid6 can handle the failure of
> > any 2 disks. You pulled 3. Good way to hose the array.
> Is failure the same of disconnession of disk?

As far as the raid system is, yes a missing disk is indistinguisable
from a failed disk that is non-responsive.

>
> I thinked there is it so. Because if the disk crash the all data are lost, but
> I if disconnect the disk there are date yet.
>

But how does the raid system know that the data is intact? Remember
that the parity data for the other disks is interspursed with the data
on each disk. Without the raid system adopting the disk, how do you
sort it out?

> > Considering that this is a test, there should be no concern over data.
> > You did a backup prior to this test?
> Yes :-)
>
> > You hosed the RAID6 array by pulling 3 disks. The spares only if a
> > spare disk can get syncd before a third disk died.
> > So, no, you have to make a new raid and restore from backup.
> Is there another way to increase the security data?

I always use raid1 or raid 10. With either, there is not parity data
floating around, just more copies of the data. Take raid1 for example.
You could have as many active disks spread out over many controllers as
you like until the risk of multiple disk failures is as low as the
cumulative risk to the box which contains the disks and controllers.

Normally, good backups, one two live copies plus a spare disk is
adequate. If it is not, then you don't have a "normal" threat tolerance
and you need to consider the risk of an individual box. I.e. totally
redundant boxes with real-time fail-over, Network Attached Storage,
iSCSI to remote boxes, etc.

Doug.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org