[gentoo-user] How to get raid

[gentoo-user] How to get raid

[Jeff Cranmer]

Hi all,

I have recently built a new system, running Gentoo on a Sabertooth 990FX
motherboard.  The board has a raid controller on which I'm running a
120GB solid state drive for the OS (Raid 0) and a set of three 1.5TB
drives which were previously running as a RAID5 array.

I can see the sda 120GB drive and have installed the operating system on
that.  I can't see one device for the three disk RAID5 array, even
though the RAID BIOS reports it as a healthy 3TB disk.  Instead I see
three separate devices, sdb, sdc and sdd

What do I need to do to mount the 3TB RAID disk?  I'm running genkernel,
and compiled it with genkernel --dmraid all.  It should already have
data on it, if I can only get gentoo to recognise it.

I can see the RAID controller when I use lspci

00:11.0 RAID bus controller: ATI Technologies Inc SB7x0,SB8x0,SB9x0 SATA
Controller [RAID5 mode] (rev 40)

One possible clue may be in dmesg, where I get the error
device-mapper: table: 253:0: raid45: unknown target type

Any assistance gratefully received.

Thanks

Jeff

Re: [gentoo-user] How to get raid

[Paul Hartman]

On 01/03/2012 08:57 PM, Jeff Cranmer wrote:
> device-mapper: table: 253:0: raid45: unknown target type

Maybe a dumb question, but is the raid45 module enabled in your kernel
config?

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Tue, 2012-01-03 at 22:21 -0600, Paul Hartman wrote:
> On 01/03/2012 08:57 PM, Jeff Cranmer wrote:
> > device-mapper: table: 253:0: raid45: unknown target type
> 
> Maybe a dumb question, but is the raid45 module enabled in your kernel
> config?
> 
genkernel --dmraid all
Not sure how to check those details in genkernel.

Re: [gentoo-user] How to get raid

[Alexander Puchmayr]

On Wednesday 04 January 2012 11:57:18 Jeff Cranmer wrote:
> Hi all,
> 
> I have recently built a new system, running Gentoo on a Sabertooth 990FX
> motherboard.  The board has a raid controller on which I'm running a
> 120GB solid state drive for the OS (Raid 0) and a set of three 1.5TB
> drives which were previously running as a RAID5 array.
> 
> I can see the sda 120GB drive and have installed the operating system on
> that.  I can't see one device for the three disk RAID5 array, even
> though the RAID BIOS reports it as a healthy 3TB disk.  Instead I see
> three separate devices, sdb, sdc and sdd
> 
> What do I need to do to mount the 3TB RAID disk?  I'm running genkernel,
> and compiled it with genkernel --dmraid all.  It should already have
> data on it, if I can only get gentoo to recognise it.
> 
> I can see the RAID controller when I use lspci
> 
> 00:11.0 RAID bus controller: ATI Technologies Inc SB7x0,SB8x0,SB9x0 SATA
> Controller [RAID5 mode] (rev 40)
> 
> One possible clue may be in dmesg, where I get the error
> device-mapper: table: 253:0: raid45: unknown target type
> 

The first question is: What type of raid are you using?

a) Software-Raid created with mdadm & co

b) Hardware-Controller based raid.

While in the first case you see all individual disks with their partitions and 
a /dev/mdX entry that actually contains the raid failsystem, the second one
shows only a /dev/sdX holding the final raid drive. 

Additionally, for the hardware based raid, you'll need a driver for the 
controller that supports the raid5. I think this is the configuration you're 
trying to run, since you mentioned that you created your raid in the RAID 
BIOS.

I'm not sure (I've never tried this) whether there is a driver for Linux 
supporting raid modes on board-embedded HW raid controllers.

    Alex

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

I was using a hardware-based 'fakeRAID'.  It used to work on my old
OpenSuse install, but that broke and I installed gentoo instead.  I
wasn't able to get that to work, and then the motherboard died, so I
built a new system and reused the 3-drive RAID5 array.
> 
> While in the first case you see all individual disks with their partitions and 
> a /dev/mdX entry that actually contains the raid failsystem, the second one
> shows only a /dev/sdX holding the final raid drive. 
> 
> Additionally, for the hardware based raid, you'll need a driver for the 
> controller that supports the raid5. I think this is the configuration you're 
> trying to run, since you mentioned that you created your raid in the RAID 
> BIOS.
> 
> I'm not sure (I've never tried this) whether there is a driver for Linux 
> supporting raid modes on board-embedded HW raid controllers.
> 
>     Alex
> 
> 
> 
> 
> 

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Dienstag, 3. Januar 2012, 21:57:18 schrieb Jeff Cranmer:
> Hi all,
> 
> I have recently built a new system, running Gentoo on a Sabertooth 990FX
> motherboard.  The board has a raid controller on which I'm running a
> 120GB solid state drive for the OS (Raid 0) and a set of three 1.5TB
> drives which were previously running as a RAID5 array.

no, it does not have a raid controller. It is bios raid. AKA fake raid. You 
will have less trouble if you stop using it.

google for mdadm. There are some very nice howto's.

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Wed, 2012-01-04 at 14:39 +0100, Volker Armin Hemmann wrote:
> Am Dienstag, 3. Januar 2012, 21:57:18 schrieb Jeff Cranmer:
> > Hi all,
> > 
> > I have recently built a new system, running Gentoo on a Sabertooth 990FX
> > motherboard.  The board has a raid controller on which I'm running a
> > 120GB solid state drive for the OS (Raid 0) and a set of three 1.5TB
> > drives which were previously running as a RAID5 array.
> 
> no, it does not have a raid controller. It is bios raid. AKA fake raid. You 
> will have less trouble if you stop using it.
> 
> google for mdadm. There are some very nice howto's.
> 

Not sure I'd agree with you about the howtos being nice.  They mostly
deal with trying to boot from a RAID array (don't want that, as I have
my OS on a non-RAID 120GB SSD).  They're also contradictory, with some
saying I need dmraid, and some saying not.  Most seem to make no more
than a passing nod towards genkernel.

So, given that from the links that I've found, here's my starting set of
questions.

In /etc/genkernel.conf, which options do I need to enable.
One guide suggested the following settings 
DMRAID="no"
MDADM="yes"
MDADM_CONFIG="/etc/mdadm.conf"
MDADM_VER="3.1.4"

If this is correct, does it matter that my mdadm version which I emerged
is 3.1.5?  The tarball in /var/cache/genkernel/src is
mdadm-3.1.4.tar.bz2
Should I copy mdadm-3.1.5.tar.bz2 from /etc/portage/distfiles into there
and rebuild genkernel.

Do I need the dodmraid option compiled into genkernel, or is that only
for fakeraid, or situations where I need to boot from a raid partition?

Do I need the dodmraid option set true in the grub.conf file, or is
'domdadm' more appropriate?

Jeff

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Mittwoch, 4. Januar 2012, 21:28:32 schrieb Jeff Cranmer:
> On Wed, 2012-01-04 at 14:39 +0100, Volker Armin Hemmann wrote:
> > Am Dienstag, 3. Januar 2012, 21:57:18 schrieb Jeff Cranmer:
> > > Hi all,
> > > 
> > > I have recently built a new system, running Gentoo on a Sabertooth 990FX
> > > motherboard.  The board has a raid controller on which I'm running a
> > > 120GB solid state drive for the OS (Raid 0) and a set of three 1.5TB
> > > drives which were previously running as a RAID5 array.
> > 
> > no, it does not have a raid controller. It is bios raid. AKA fake raid.
> > You
> > will have less trouble if you stop using it.
> > 
> > google for mdadm. There are some very nice howto's.
> 
> Not sure I'd agree with you about the howtos being nice.  They mostly
> deal with trying to boot from a RAID array (don't want that, as I have
> my OS on a non-RAID 120GB SSD).  They're also contradictory, with some
> saying I need dmraid, and some saying not.  Most seem to make no more
> than a passing nod towards genkernel.

the short one:

partition one disk with (c)fdisk. Use sfdisk to transfer the partition scheme 
to the other disks.

run mdadm --create /dev/md0 level=whatever you want --raid-
devices=thenumberofdevices /dev/sdXY /dev/sdZY ... 

mdadm --detail --scan >> /etc/mdadm.conf

done


> 
> So, given that from the links that I've found, here's my starting set of
> questions.
> 
> In /etc/genkernel.conf, which options do I need to enable.
> One guide suggested the following settings
> DMRAID="no"
> MDADM="yes"
> MDADM_CONFIG="/etc/mdadm.conf"
> MDADM_VER="3.1.4"

there is a reason why I never ever touch genkernel.

you should forget that crap. You don't need to copy around anything. If your 
root is not on some fancy setup, you don't need initramfs.

Just make a nice kernel, put it in /boot. Done.

grub.conf:
kernel /vmlinuz root=/dev/sda1 nmi_watchdog=0

and you are fine.

Have the raids assembled by a) kernel (in that case you have to tell mdadm 
that on creation time, man mdadm is your friend) or by mdadm init script.

Don't use fakeraid. Set bios to ahci and be done with this.

the relevant part of Kernel config for example:

   <*>   RAID support                                                                                   
│ │  
  │ │                               [*]     Autodetect RAID arrays during 
kernel boot                                                    │ │  
  │ │                               < >     Linear (append) mode                                                                         
│ │  
  │ │                               < >     RAID-0 (striping) mode                                                                       
│ │  
  │ │                               <*>     RAID-1 (mirroring) mode                                                                      
│ │  
  │ │                               < >     RAID-10 (mirrored striping) mode                                                             
│ │  
  │ │                               <*>     RAID-4/RAID-5/RAID-6 mode                                                                    
│ │  
  │ │                               [ ]       RAID-4/RAID-5/RAID-6 Multicore 
processing (EXPERIMENTAL)                                   │ │  
  │ │                               < >     Multipath I/O support                                                                        
│ │  
  │ │                               < >     Faulty test module for MD                                                                    
│ │  
  │ │                               < >   Device mapper support                                                                          
│ │  
  │ │                                                                                                                                    
│ │  
  │ │

as you can see no dm support in my kernel.

No look what I got...

cat /proc/mdstat 
Personalities : [raid1] [raid6] [raid5] [raid4] 
md5 : active raid1 sdg2[2] sdf2[1]
      830278202 blocks super 1.2 [2/2] [UU]
      
md4 : active raid1 sdf1[1] sdg1[2]
      146479542 blocks super 1.2 [2/2] [UU]
      
md124 : active raid1 sdc1[2] sdd1[1] sdb1[0]
      64128 blocks [3/3] [UUU]
      
md1 : active raid5 sdc3[2] sdd3[1] sdb3[0]
      78123904 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU]
      
md2 : active raid5 sdc5[2] sdd5[1] sdb5[0]
      39069824 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU]
      
md127 : active raid5 sdc6[2] sdd6[1] sdb6[0]
      843813888 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU]

the numbers where once nicely 0-4 but some update fucked that up. No big deal 
- I mount by UUID. Something I strongly recommend.

-- 
#163933

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Thu, 2012-01-05 at 04:01 +0100, Volker Armin Hemmann wrote:

> the short one:
> 
> partition one disk with (c)fdisk. Use sfdisk to transfer the partition scheme 
> to the other disks.
> 
> run mdadm --create /dev/md0 level=whatever you want --raid-
> devices=thenumberofdevices /dev/sdXY /dev/sdZY ... 
> 
> mdadm --detail --scan >> /etc/mdadm.conf
> 
> done
> 
> 
OK, but there is active data on the disks, so I don't want to partition
them.  They should already partitioned, and running fdisk will erase the
data.

If I run mdadm --create /dev/md0 level=5
--raid-devices=3 /dev/sdb /dev/sdc /dev/sdd, will that erase data
already on the disks?

Prior to running this command, there is no /dev/md entry.  Is this
correct?

Looking further by using fdisk, it appears that sdc has a linux
partition on sdc1 starting at sector 34, and a GPT partition of size 0+
at /dev/sdc4, sector 0.  Nothing else is on that disk (no sdc2 or sdc3).

sdd and sdb report invalid partition table flags and do not appear to
have active partitions.  Does this make sense?

Is it possible that I ordered the disks incorrectly when I installed
them, and by simply swapping disks b and c at the raid I can get things
to start making sense?  Is there an order to a set of RAID5 disks?  I
thought any two of three RAID5 disks could be recovered, regardless of
which one dies?
 
> there is a reason why I never ever touch genkernel.
> 
> you should forget that crap. You don't need to copy around anything. If your 
> root is not on some fancy setup, you don't need initramfs.
> 
> Just make a nice kernel, put it in /boot. Done.
> 
OK.  The OS disk is non-RAID (120GB SSD), so I don't need any fancy
options in my kernel. All the domdadm and dodmraid stuff is needed just
when your OS disk is raided.  Correct?

Thanks

Jeff

Re: [gentoo-user] How to get raid

[Hinnerk van Bruinehsen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 05.01.2012 04:45, Jeff Cranmer wrote:
> On Thu, 2012-01-05 at 04:01 +0100, Volker Armin Hemmann wrote:
> 
>> the short one:
>> 
>> partition one disk with (c)fdisk. Use sfdisk to transfer the
>> partition scheme to the other disks.
>> 
>> run mdadm --create /dev/md0 level=whatever you want --raid- 
>> devices=thenumberofdevices /dev/sdXY /dev/sdZY ...
>> 
>> mdadm --detail --scan >> /etc/mdadm.conf
>> 
>> done
>> 
>> 
> OK, but there is active data on the disks, so I don't want to
> partition them.  They should already partitioned, and running fdisk
> will erase the data.
> 
> If I run mdadm --create /dev/md0 level=5 --raid-devices=3 /dev/sdb
> /dev/sdc /dev/sdd, will that erase data already on the disks?
> 
> Prior to running this command, there is no /dev/md entry.  Is this 
> correct?
> 
> Looking further by using fdisk, it appears that sdc has a linux 
> partition on sdc1 starting at sector 34, and a GPT partition of
> size 0+ at /dev/sdc4, sector 0.  Nothing else is on that disk (no
> sdc2 or sdc3).
> 
> sdd and sdb report invalid partition table flags and do not appear
> to have active partitions.  Does this make sense?
> 
> Is it possible that I ordered the disks incorrectly when I
> installed them, and by simply swapping disks b and c at the raid I
> can get things to start making sense?  Is there an order to a set
> of RAID5 disks?  I thought any two of three RAID5 disks could be
> recovered, regardless of which one dies?
> 
>> there is a reason why I never ever touch genkernel.
>> 
>> you should forget that crap. You don't need to copy around
>> anything. If your root is not on some fancy setup, you don't need
>> initramfs.
>> 
>> Just make a nice kernel, put it in /boot. Done.
>> 
> OK.  The OS disk is non-RAID (120GB SSD), so I don't need any
> fancy options in my kernel. All the domdadm and dodmraid stuff is
> needed just when your OS disk is raided.  Correct?
> 
> Thanks
> 
> Jeff

If you used a hardware-based RAID before, you should do nothing with
mdadm or fdisk until you have a working copy of your data.

If I recall correctly, you said, you used that RAID-array on a
different mobo before. Then the mobo died and you want just to reuse
the array. Correct?

If that's correct you may be in serious trouble, because afaik there
ist no real standart in how to create a hardware RAID. If the old
RAID-controller/firmware isn't available anymore you could try to find
an identical one.
There may be even the possibility that through your tries with the new
controller/mobo the array is damaged right now.
That is - by the way - one very good reason to use a software-based
solution like mdadm: you aren't restricted to specific hardware...



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJPBV29AAoJEJwwOFaNFkYcdhYH/A6zUEe8AQGYR959DNYvfkIV
V4XRyP1QbVYNcC5hou3vtC8ey7SvZisOXh3JX7vo534ATtY+KW6hRIHu5xlDJe67
KuX1aa37fZ9ivhkpLaGGXOluDZIlf28L70jGV48FMd95TMWFmK4tO12CwTbmRy30
ckuyHFgLrJOsYcTIlrlB/DSsosklsQ3wyMJX5XbqUi7dJuae+h+yiphuPoAU99iX
FnO/QxhjfrX37Ch56ughvTSMKxRe6XDtECyIB/v3/2Dq1GH07FHONjIfJ8qEbqjK
bfa4W9XHygkWd/Wwfop//2hmz1bXJmGrVRd9iLzs9prnc/Cgv0yDuAPoUhFKgkw=
=Yrz+
-----END PGP SIGNATURE-----

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Mittwoch, 4. Januar 2012, 22:45:45 schrieb Jeff Cranmer:
> On Thu, 2012-01-05 at 04:01 +0100, Volker Armin Hemmann wrote:
> > the short one:
> > 
> > partition one disk with (c)fdisk. Use sfdisk to transfer the partition
> > scheme to the other disks.
> > 
> > run mdadm --create /dev/md0 level=whatever you want --raid-
> > devices=thenumberofdevices /dev/sdXY /dev/sdZY ...
> > 
> > mdadm --detail --scan >> /etc/mdadm.conf
> > 
> > done
> 
> OK, but there is active data on the disks, so I don't want to partition
> them.  They should already partitioned, and running fdisk will erase the
> data.

first rule:

always mount a scratch monkey

In your case: always backup data.

There is a way to preserve the data on one disk, create a raid5 with one disk 
missing, then copying the data onto the raid and add the disk. 

But that is high risk stuff.

> 
> If I run mdadm --create /dev/md0 level=5
> --raid-devices=3 /dev/sdb /dev/sdc /dev/sdd, will that erase data
> already on the disks?
> 
> Prior to running this command, there is no /dev/md entry.  Is this
> correct?

yes. You might have to create the nodes with mknod - my memory is sketchy 
there.


> Looking further by using fdisk, it appears that sdc has a linux
> partition on sdc1 starting at sector 34, and a GPT partition of size 0+
> at /dev/sdc4, sector 0.  Nothing else is on that disk (no sdc2 or sdc3).
> 
> sdd and sdb report invalid partition table flags and do not appear to
> have active partitions.  Does this make sense?

if you used fakeraid before, yes. But that means: without the original 
fakeraid everything on that disks is inaccessible... and you need to partition 
them.

> 
> Is it possible that I ordered the disks incorrectly when I installed
> them, and by simply swapping disks b and c at the raid I can get things
> to start making sense?  Is there an order to a set of RAID5 disks?  I
> thought any two of three RAID5 disks could be recovered, regardless of
> which one dies?

no.
First, the order of the disks is irrelevant, but the most important thing:

with Raid5 ONE disk out of an array might fail. No matter how many disks - two 
fail and everything is lost.

> 
> > there is a reason why I never ever touch genkernel.
> > 
> > you should forget that crap. You don't need to copy around anything. If
> > your root is not on some fancy setup, you don't need initramfs.
> > 
> > Just make a nice kernel, put it in /boot. Done.
> 
> OK.  The OS disk is non-RAID (120GB SSD), so I don't need any fancy
> options in my kernel. All the domdadm and dodmraid stuff is needed just
> when your OS disk is raided.  Correct?

yes


-- 
#163933

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Thu, 2012-01-05 at 11:22 +0100, Volker Armin Hemmann wrote:
> Am Mittwoch, 4. Januar 2012, 22:45:45 schrieb Jeff Cranmer:
> > On Thu, 2012-01-05 at 04:01 +0100, Volker Armin Hemmann wrote:
> > > the short one:
> > > 
> > > partition one disk with (c)fdisk. Use sfdisk to transfer the partition
> > > scheme to the other disks.
> > > 
> > > run mdadm --create /dev/md0 level=whatever you want --raid-
> > > devices=thenumberofdevices /dev/sdXY /dev/sdZY ...
> > > 
> > > mdadm --detail --scan >> /etc/mdadm.conf
> > > 
> > > done
> > 
> > OK, but there is active data on the disks, so I don't want to partition
> > them.  They should already partitioned, and running fdisk will erase the
> > data.
> 
> first rule:
> 
> always mount a scratch monkey
> 
> In your case: always backup data.
> 
No big deal.
99.9% of the data is backed up.  I was just hoping to recover the last
0.1% (picky huh?<g>).  Now that I know one of the main drawbacks of
fakeraid, I think I'll move ahead with software RAID.

OK, so I've partitioned the first disk as a single linux partition
(/dev/sdb1, ID 83, Linux).
How do I use sfdisk to transfer that partition scheme to the other
disks?  Is it not sufficient just to partition the other two disks in
the same way as the first?

Jeff

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Donnerstag, 5. Januar 2012, 20:13:04 schrieb Jeff Cranmer:
> On Thu, 2012-01-05 at 11:22 +0100, Volker Armin Hemmann wrote:
> > Am Mittwoch, 4. Januar 2012, 22:45:45 schrieb Jeff Cranmer:
> > > On Thu, 2012-01-05 at 04:01 +0100, Volker Armin Hemmann wrote:
> > > > the short one:
> > > > 
> > > > partition one disk with (c)fdisk. Use sfdisk to transfer the partition
> > > > scheme to the other disks.
> > > > 
> > > > run mdadm --create /dev/md0 level=whatever you want --raid-
> > > > devices=thenumberofdevices /dev/sdXY /dev/sdZY ...
> > > > 
> > > > mdadm --detail --scan >> /etc/mdadm.conf
> > > > 
> > > > done
> > > 
> > > OK, but there is active data on the disks, so I don't want to partition
> > > them.  They should already partitioned, and running fdisk will erase the
> > > data.
> > 
> > first rule:
> > 
> > always mount a scratch monkey
> > 
> > In your case: always backup data.
> 
> No big deal.
> 99.9% of the data is backed up.  I was just hoping to recover the last
> 0.1% (picky huh?<g>).  Now that I know one of the main drawbacks of
> fakeraid, I think I'll move ahead with software RAID.
> 
> OK, so I've partitioned the first disk as a single linux partition
> (/dev/sdb1, ID 83, Linux).

if you want to use kernel autodetection (nice but on the way out) you should 
change the type.

> How do I use sfdisk to transfer that partition scheme to the other
> disks?  Is it not sufficient just to partition the other two disks in
> the same way as the first?

sfdisk -d /dev/sda | sfdisk /dev/sdb 

is safe.

-- 
#163933

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Donnerstag, 5. Januar 2012, 20:13:04 schrieb Jeff Cranmer:
> On Thu, 2012-01-05 at 11:22 +0100, Volker Armin Hemmann wrote:
> > Am Mittwoch, 4. Januar 2012, 22:45:45 schrieb Jeff Cranmer:
> > > On Thu, 2012-01-05 at 04:01 +0100, Volker Armin Hemmann wrote:
> > > > the short one:
> > > > 
> > > > partition one disk with (c)fdisk. Use sfdisk to transfer the partition
> > > > scheme to the other disks.
> > > > 
> > > > run mdadm --create /dev/md0 level=whatever you want --raid-
> > > > devices=thenumberofdevices /dev/sdXY /dev/sdZY ...
> > > > 
> > > > mdadm --detail --scan >> /etc/mdadm.conf
> > > > 
> > > > done
> > > 
> > > OK, but there is active data on the disks, so I don't want to partition
> > > them.  They should already partitioned, and running fdisk will erase the
> > > data.
> > 
> > first rule:
> > 
> > always mount a scratch monkey
> > 
> > In your case: always backup data.
> 
> No big deal.
> 99.9% of the data is backed up.  I was just hoping to recover the last
> 0.1% (picky huh?<g>).  Now that I know one of the main drawbacks of
> fakeraid, I think I'll move ahead with software RAID.
> 
> OK, so I've partitioned the first disk as a single linux partition
> (/dev/sdb1, ID 83, Linux).
> How do I use sfdisk to transfer that partition scheme to the other
> disks?  Is it not sufficient just to partition the other two disks in
> the same way as the first?
> 
> Jeff

in your case 

sfdisk -d /dev/sdb | sfdisk /dev/sdc

of course ;)
 
-- 
#163933

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Fri, 2012-01-06 at 02:42 +0100, Volker Armin Hemmann wrote:

> in your case 
> 
> sfdisk -d /dev/sdb | sfdisk /dev/sdc
> 
> of course ;)
>  
One of the disks had a GPT partition table which I was eventually able
to get rid of with gdisk (emerge -av gptfdisk).   

I'm close.  I had a 2.7TiB RAID5 array using genkernal, comprising three
1.5TB disks, using the commands
mdadm --create /dev/md0 --level=5
--raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1

mdadm --detail --scan >> /etc/mdadm.conf

I formatted this array as an xfs filesystem.

After reboot, however, /dev/md0 is still there, but I get a 'can't read
superblock' error.

What am I missing?

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Donnerstag, 5. Januar 2012, 23:44:10 schrieb Jeff Cranmer:
> On Fri, 2012-01-06 at 02:42 +0100, Volker Armin Hemmann wrote:
> > in your case
> > 
> > sfdisk -d /dev/sdb | sfdisk /dev/sdc
> > 
> > of course ;)
> 
> One of the disks had a GPT partition table which I was eventually able
> to get rid of with gdisk (emerge -av gptfdisk).
> 
> I'm close.  I had a 2.7TiB RAID5 array using genkernal, comprising three
> 1.5TB disks, using the commands
> mdadm --create /dev/md0 --level=5
> --raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
> 
> mdadm --detail --scan >> /etc/mdadm.conf
> 
> I formatted this array as an xfs filesystem.
> 
> After reboot, however, /dev/md0 is still there, but I get a 'can't read
> superblock' error.
> 
> What am I missing?

have you set the type to linux raid autodetect?

have you tried mdadm --assemble? 

-- 
#163933

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Fri, 2012-01-06 at 13:36 +0100, Volker Armin Hemmann wrote:
> Am Donnerstag, 5. Januar 2012, 23:44:10 schrieb Jeff Cranmer:
> > On Fri, 2012-01-06 at 02:42 +0100, Volker Armin Hemmann wrote:
> > > in your case
> > > 
> > > sfdisk -d /dev/sdb | sfdisk /dev/sdc
> > > 
> > > of course ;)
> > 
> > One of the disks had a GPT partition table which I was eventually able
> > to get rid of with gdisk (emerge -av gptfdisk).
> > 
> > I'm close.  I had a 2.7TiB RAID5 array using genkernal, comprising three
> > 1.5TB disks, using the commands
> > mdadm --create /dev/md0 --level=5
> > --raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
> > 
> > mdadm --detail --scan >> /etc/mdadm.conf
> > 
> > I formatted this array as an xfs filesystem.
> > 
> > After reboot, however, /dev/md0 is still there, but I get a 'can't read
> > superblock' error.
> > 
> > What am I missing?
> 
> have you set the type to linux raid autodetect?
> 
> have you tried mdadm --assemble? 
> 
mdadm --assemble /dev/md0 didn't make any difference.
Where do I set the type?

Thanks

Jeff

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

> > > 
> > > What am I missing?
> > 
> > have you set the type to linux raid autodetect?
> > 
> > have you tried mdadm --assemble? 
> > 
> mdadm --assemble /dev/md0 didn't make any difference.
> Where do I set the type?
> 
after assembling,
results of cat/proc/mdstat
personalities : [linear] [raid0] [raid10] [raid6] [raid5] [raid4]
[multipath] [faulty]
md0 : inactive sdb1[0](S) sdd1[3](S) sdc1[1](S)
      4395409608 blocks super 1.2

unused devices: <none>

results of mdadm --detail /dev/md0
mdadm: md device /dev/md0 does not appear to be active.

results of /etc/init.d/mdadm status
 * status: started

fstab line
/dev/md0   /data   xfs   noatime   0 0

Is there a raid option I need to add to the fstab entry?
Is there another service that needs to run, other than mdam?

Thanks

Jeff

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Sat, 2012-01-07 at 10:11 -0500, Jeff Cranmer wrote:
> > > > 
> > > > What am I missing?
> > > 
> > > have you set the type to linux raid autodetect?
> > > 
> > > have you tried mdadm --assemble? 
> > > 
> > mdadm --assemble /dev/md0 didn't make any difference.
> > Where do I set the type?
> > 
> after assembling,
> results of cat/proc/mdstat
> personalities : [linear] [raid0] [raid10] [raid6] [raid5] [raid4]
> [multipath] [faulty]
> md0 : inactive sdb1[0](S) sdd1[3](S) sdc1[1](S)
>       4395409608 blocks super 1.2
> 
> unused devices: <none>
> 
> results of mdadm --detail /dev/md0
> mdadm: md device /dev/md0 does not appear to be active.
> 
> results of /etc/init.d/mdadm status
>  * status: started
> 
> fstab line
> /dev/md0   /data   xfs   noatime   0 0
> 
> Is there a raid option I need to add to the fstab entry?
> Is there another service that needs to run, other than mdam?
> 
> Thanks
> 
> Jeff
> 
> 
I tried changing the type of each array element in fdisk to fd (linux
raid autodetect.

The array is still not being recognised at boot, with the same 'cannot
read superblock' error.

I also tried re-running mdadm --create /dev/md0 --level=5
--raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
I get the error
mdadm: device /dev/sdb1 not suitable for any style of array.

What is going on here?

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Samstag, 7. Januar 2012, 12:20:08 schrieb Jeff Cranmer:
> On Sat, 2012-01-07 at 10:11 -0500, Jeff Cranmer wrote:
> > > > > What am I missing?
> > > > 
> > > > have you set the type to linux raid autodetect?
> > > > 
> > > > have you tried mdadm --assemble?
> > > 
> > > mdadm --assemble /dev/md0 didn't make any difference.
> > > Where do I set the type?
> > 
> > after assembling,
> > results of cat/proc/mdstat
> > personalities : [linear] [raid0] [raid10] [raid6] [raid5] [raid4]
> > [multipath] [faulty]
> > md0 : inactive sdb1[0](S) sdd1[3](S) sdc1[1](S)
> > 
> >       4395409608 blocks super 1.2
> > 
> > unused devices: <none>
> > 
> > results of mdadm --detail /dev/md0
> > mdadm: md device /dev/md0 does not appear to be active.
> > 
> > results of /etc/init.d/mdadm status
> > 
> >  * status: started
> > 
> > fstab line
> > /dev/md0   /data   xfs   noatime   0 0
> > 
> > Is there a raid option I need to add to the fstab entry?
> > Is there another service that needs to run, other than mdam?
> > 
> > Thanks
> > 
> > Jeff
> 
> I tried changing the type of each array element in fdisk to fd (linux
> raid autodetect.
> 
> The array is still not being recognised at boot, with the same 'cannot
> read superblock' error.
> 
> I also tried re-running mdadm --create /dev/md0 --level=5
> --raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
> I get the error
> mdadm: device /dev/sdb1 not suitable for any style of array.
> 
> What is going on here?

I am thinking ;)


-- 
#163933

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

> > 
> > I tried changing the type of each array element in fdisk to fd (linux
> > raid autodetect.
> > 
> > The array is still not being recognised at boot, with the same 'cannot
> > read superblock' error.
> > 
> > I also tried re-running mdadm --create /dev/md0 --level=5
> > --raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
> > I get the error
> > mdadm: device /dev/sdb1 not suitable for any style of array.
> > 
> > What is going on here?
> 
> I am thinking ;)
> 
> 
LOL!

Me too.

mdadm --detail /dev/md0 thinks that /dev/sdc1 is faulty.
I'm not sure whether it's really faulty, or just that my setup for RAID
is screwed up.

How do I get rid of an existing /dev/md0?

I'm thinking that I can try creating a RAID1 array using the two
allegedly good disks and see if I can make that work.

If that works, I'll get rid of it and try recreating the RAID1 with one
good disk and the one that mdadm thinks is faulty.

Hopefully that will show me whether I have a hardware problem or a
software one.

Jeff

Re: [gentoo-user] How to get raid

[Volker Armin Hemmann]

Am Samstag, 7. Januar 2012, 13:27:04 schrieb Jeff Cranmer:
> > > I tried changing the type of each array element in fdisk to fd (linux
> > > raid autodetect.
> > > 
> > > The array is still not being recognised at boot, with the same 'cannot
> > > read superblock' error.
> > > 
> > > I also tried re-running mdadm --create /dev/md0 --level=5
> > > --raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
> > > I get the error
> > > mdadm: device /dev/sdb1 not suitable for any style of array.
> > > 
> > > What is going on here?
> > 
> > I am thinking ;)
> 
> LOL!
> 
> Me too.
> 
> mdadm --detail /dev/md0 thinks that /dev/sdc1 is faulty.
> I'm not sure whether it's really faulty, or just that my setup for RAID
> is screwed up.
> 
> How do I get rid of an existing /dev/md0?

you stop it. Override the superblock with dd.. and lose all data on the disks.


> 
> I'm thinking that I can try creating a RAID1 array using the two
> allegedly good disks and see if I can make that work.

yeah

> 
> If that works, I'll get rid of it and try recreating the RAID1 with one
> good disk and the one that mdadm thinks is faulty.
> 

you don't have to. You can migrate a 2 disk raid1 to a 3 disk raid5. Howtos 
are availble via google.


just saying - box in suspend to ram. I change the cable (and connector on 
mobo) on a disk with two raid 1 partitions on it.

One came back after starting the box.

The other? Nothing I tried worked. At the end I dd'ed the partition.. and did 
a complete 'faulty disk/replacement' resync....

argl.


-- 
#163933

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

> > 
> > How do I get rid of an existing /dev/md0?
> 
> you stop it. Override the superblock with dd.. and lose all data on the disks.
> 
> 
> > 
> > I'm thinking that I can try creating a RAID1 array using the two
> > allegedly good disks and see if I can make that work.
> 
> yeah
> 
> > 
> > If that works, I'll get rid of it and try recreating the RAID1 with one
> > good disk and the one that mdadm thinks is faulty.
> > 
> 
> you don't have to. You can migrate a 2 disk raid1 to a 3 disk raid5. Howtos 
> are availble via google.
> 
> 
> just saying - box in suspend to ram. I change the cable (and connector on 
> mobo) on a disk with two raid 1 partitions on it.
> 
> One came back after starting the box.
> 
> The other? Nothing I tried worked. At the end I dd'ed the partition.. and did 
> a complete 'faulty disk/replacement' resync....
> 
> argl.
> 
> 
You're assuming I have more knowledge that I do.
Can you explain the steps more in layman's terms.  I've never used dd
before.

Jeff

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

> > 
> > Me too.
> > 
> > mdadm --detail /dev/md0 thinks that /dev/sdc1 is faulty.
> > I'm not sure whether it's really faulty, or just that my setup for RAID
> > is screwed up.
> > 
> > How do I get rid of an existing /dev/md0?
> 
> you stop it. Override the superblock with dd.. and lose all data on the disks.
> 
> 
> > 
> > I'm thinking that I can try creating a RAID1 array using the two
> > allegedly good disks and see if I can make that work.
> 
> yeah
> 
> > 
> > If that works, I'll get rid of it and try recreating the RAID1 with one
> > good disk and the one that mdadm thinks is faulty.
> > 
> 
> you don't have to. You can migrate a 2 disk raid1 to a 3 disk raid5. Howtos 
> are availble via google.
> 
> 
> just saying - box in suspend to ram. I change the cable (and connector on 
> mobo) on a disk with two raid 1 partitions on it.
> 
> One came back after starting the box.
> 
> The other? Nothing I tried worked. At the end I dd'ed the partition.. and did 
> a complete 'faulty disk/replacement' resync....
> 
> argl.
> 
> 
OK, so lesson learned.  Just because it builds correctly in a RAID1
array, that doesn't mean that the drive isn't toast.

I ran badblocks on the three drive components and, surprise,
surprise, /dev/sdc came up faulty.  I think I'll just build the two
non-faulty drives as a RAID0 array until the hard drive prices come back
down to pre-Thailand flood prices and backup regularly.

Thanks for all the help.

Jeff

Re: [gentoo-user] How to get raid

[Pandu Poluan]

[-- Attachment #1: Type: text/plain, Size: 1725 bytes --]

On Jan 10, 2012 8:48 AM, "Jeff Cranmer" <jeff@lotussevencars.com> wrote:
>
>
> > >
> > > Me too.
> > >
> > > mdadm --detail /dev/md0 thinks that /dev/sdc1 is faulty.
> > > I'm not sure whether it's really faulty, or just that my setup for
RAID
> > > is screwed up.
> > >
> > > How do I get rid of an existing /dev/md0?
> >
> > you stop it. Override the superblock with dd.. and lose all data on the
disks.
> >
> >
> > >
> > > I'm thinking that I can try creating a RAID1 array using the two
> > > allegedly good disks and see if I can make that work.
> >
> > yeah
> >
> > >
> > > If that works, I'll get rid of it and try recreating the RAID1 with
one
> > > good disk and the one that mdadm thinks is faulty.
> > >
> >
> > you don't have to. You can migrate a 2 disk raid1 to a 3 disk raid5.
Howtos
> > are availble via google.
> >
> >
> > just saying - box in suspend to ram. I change the cable (and connector
on
> > mobo) on a disk with two raid 1 partitions on it.
> >
> > One came back after starting the box.
> >
> > The other? Nothing I tried worked. At the end I dd'ed the partition..
and did
> > a complete 'faulty disk/replacement' resync....
> >
> > argl.
> >
> >
> OK, so lesson learned.  Just because it builds correctly in a RAID1
> array, that doesn't mean that the drive isn't toast.
>
> I ran badblocks on the three drive components and, surprise,
> surprise, /dev/sdc came up faulty.  I think I'll just build the two
> non-faulty drives as a RAID0 array until the hard drive prices come back
> down to pre-Thailand flood prices and backup regularly.
>
> Thanks for all the help.
>
> Jeff
>
>
>

RAID 0?!?!

Please reconsider.

With RAID 0, *any* single drive failure will result in *total* data loss.

Rgds,

[-- Attachment #2: Type: text/html, Size: 2474 bytes --]

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

This is true, however it's a temporary measure only, and I have backups.
Once the prices drop again, I'll buy another 1.5TB disk and convert back
to a RAID5.

On Tue, 2012-01-10 at 13:14 +0700, Pandu Poluan wrote:
> 
> On Jan 10, 2012 8:48 AM, "Jeff Cranmer" <jeff@lotussevencars.com>
> wrote:
> >
> >
> > > >
> > > > Me too.
> > > >
> > > > mdadm --detail /dev/md0 thinks that /dev/sdc1 is faulty.
> > > > I'm not sure whether it's really faulty, or just that my setup
> for RAID
> > > > is screwed up.
> > > >
> > > > How do I get rid of an existing /dev/md0?
> > >
> > > you stop it. Override the superblock with dd.. and lose all data
> on the disks.
> > >
> > >
> > > >
> > > > I'm thinking that I can try creating a RAID1 array using the two
> > > > allegedly good disks and see if I can make that work.
> > >
> > > yeah
> > >
> > > >
> > > > If that works, I'll get rid of it and try recreating the RAID1
> with one
> > > > good disk and the one that mdadm thinks is faulty.
> > > >
> > >
> > > you don't have to. You can migrate a 2 disk raid1 to a 3 disk
> raid5. Howtos
> > > are availble via google.
> > >
> > >
> > > just saying - box in suspend to ram. I change the cable (and
> connector on
> > > mobo) on a disk with two raid 1 partitions on it.
> > >
> > > One came back after starting the box.
> > >
> > > The other? Nothing I tried worked. At the end I dd'ed the
> partition.. and did
> > > a complete 'faulty disk/replacement' resync....
> > >
> > > argl.
> > >
> > >
> > OK, so lesson learned.  Just because it builds correctly in a RAID1
> > array, that doesn't mean that the drive isn't toast.
> >
> > I ran badblocks on the three drive components and, surprise,
> > surprise, /dev/sdc came up faulty.  I think I'll just build the two
> > non-faulty drives as a RAID0 array until the hard drive prices come
> back
> > down to pre-Thailand flood prices and backup regularly.
> >
> > Thanks for all the help.
> >
> > Jeff
> >
> >
> >
> 
> RAID 0?!?! 
> 
> Please reconsider. 
> 
> With RAID 0, *any* single drive failure will result in *total* data
> loss.
> 
> Rgds, 
> 
> 

Re: [gentoo-user] How to get raid

[Paul Hartman]

On 01/07/2012 11:20 AM, Jeff Cranmer wrote:
> On Sat, 2012-01-07 at 10:11 -0500, Jeff Cranmer wrote:
>>>>>
>>>>> What am I missing?
>>>>
>>>> have you set the type to linux raid autodetect?
>>>>
>>>> have you tried mdadm --assemble? 
>>>>
>>> mdadm --assemble /dev/md0 didn't make any difference.
>>> Where do I set the type?
>>>
>> after assembling,
>> results of cat/proc/mdstat
>> personalities : [linear] [raid0] [raid10] [raid6] [raid5] [raid4]
>> [multipath] [faulty]
>> md0 : inactive sdb1[0](S) sdd1[3](S) sdc1[1](S)
>>       4395409608 blocks super 1.2
>>
>> unused devices: <none>
>>
>> results of mdadm --detail /dev/md0
>> mdadm: md device /dev/md0 does not appear to be active.
>>
>> results of /etc/init.d/mdadm status
>>  * status: started
>>
>> fstab line
>> /dev/md0   /data   xfs   noatime   0 0
>>
>> Is there a raid option I need to add to the fstab entry?
>> Is there another service that needs to run, other than mdam?
>>
>> Thanks
>>
>> Jeff
>>
>>
> I tried changing the type of each array element in fdisk to fd (linux
> raid autodetect.
> 
> The array is still not being recognised at boot, with the same 'cannot
> read superblock' error.
> 
> I also tried re-running mdadm --create /dev/md0 --level=5
> --raid-devices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
> I get the error
> mdadm: device /dev/sdb1 not suitable for any style of array.
> 
> What is going on here?

(I didn't read this whole thread, sorry if I'm repeating someone else's
advice)

kernel autodetection only works on old superblock version 0.90, you're
using 1.2. Not a big deal, we use mdadm to do it.

Define your arrays in /etc/mdadm.conf and start /etc/init.d/mdadm in
your boot runscripts with "rc-update add mdadm boot", it will bring up
the array at boot time.

In my mdadm.conf i have a line like this:

ARRAY /dev/md1 metadata=1.01 name=black:1
UUID=8e653e72:9d5df6ba:bb66ea8b:02f1c317

(might be word-wrapped, should be all one line)

That's all that was needed to bring it up automatically at boot time.

Also AFAIR there was a "gotcha" about the hostname stored in the array's
metadata must match your machine's hostname or else mdadm auto-assemble
won't accept it (to protect you in case you're plugging disks from
another machine for recovery, you don't want it to use them as your main
drives), so in that case you must specify it explicitly or set the AUTO
parameter in mdadm.conf to accept this condition. If you created the
array from within a LiveCD or on another machine, the hostname might not
match your system.

See the mdadm manpage for more info.

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Sun, 2012-01-08 at 12:31 -0600, Paul Hartman wrote:
> > 
> > What is going on here?
> 
> (I didn't read this whole thread, sorry if I'm repeating someone else's
> advice)
> 
> kernel autodetection only works on old superblock version 0.90, you're
> using 1.2. Not a big deal, we use mdadm to do it.
> 
> Define your arrays in /etc/mdadm.conf and start /etc/init.d/mdadm in
> your boot runscripts with "rc-update add mdadm boot", it will bring up
> the array at boot time.
> 
> In my mdadm.conf i have a line like this:
> 
> ARRAY /dev/md1 metadata=1.01 name=black:1
> UUID=8e653e72:9d5df6ba:bb66ea8b:02f1c317
> 
> (might be word-wrapped, should be all one line)
> 
> That's all that was needed to bring it up automatically at boot time.
> 
> Also AFAIR there was a "gotcha" about the hostname stored in the array's
> metadata must match your machine's hostname or else mdadm auto-assemble
> won't accept it (to protect you in case you're plugging disks from
> another machine for recovery, you don't want it to use them as your main
> drives), so in that case you must specify it explicitly or set the AUTO
> parameter in mdadm.conf to accept this condition. If you created the
> array from within a LiveCD or on another machine, the hostname might not
> match your system.
> 
> See the mdadm manpage for more info.

mdadm was added to the default level, not boot.
My /etc/mdadm.conf file has two active lines
DEVICE /dev/sd[bcd]1
ARRAY dev/md0 metadata=1.2 spares=1 name=office-desktop:0
devices=/dev/sdb1,dev/sdc1,/dev/sdd1

It looks like I'm having trouble with a faulty /dev/sdc1, so what I'd
like to do is wipe out the existing array and try starting a RAID1 array
just with sdb1 and sdd1.

I got rid of the old array by using the commands
mdadm --manage --fail /dev/md0
mdadm  --manage --stop /dev/md0

I then used mdadm --verbose --create /dev/md0 --level=1
--raid-devices=2 /dev/sdb1 /dev/sdd1

The result of this command was
dadm: /dev/sdb1 appears to be part of a raid array:
    level=raid5 devices=3 ctime=Sat Jan  7 08:16:00 2012
mdadm: partition table exists on /dev/sdb1 but will be lost or
       meaningless after creating array
mdadm: Note: this array has metadata at the start and
    may not be suitable as a boot device.  If you plan to
    store '/boot' on this device please ensure that
    your boot-loader understands md/v1.x metadata, or use
    --metadata=0.90
mdadm: /dev/sdd1 appears to be part of a raid array:
    level=raid5 devices=3 ctime=Sat Jan  7 08:16:00 2012
mdadm: size set to 1465136400K
Continue creating array? y
mdadm: Defaulting to version 1.2 metadata
mdadm: array /dev/md0 started.

The results of cat /proc/mdstat are
Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5]
[raid4] [multipath] 
md0 : active raid1 sdd1[1] sdb1[0]
      1465136400 blocks super 1.2 [2/2] [UU]
      [>....................]  resync =  2.1% (31838144/1465136400)
finish=269.7min speed=88551K/sec
      
unused devices: <none>

Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5]
[raid4] [multipath] 
md0 : active raid1 sdd1[1] sdb1[0]
      1465136400 blocks super 1.2 [2/2] [UU]
      [>....................]  resync =  2.1% (31838144/1465136400)
finish=269.7min speed=88551K/sec
      
unused devices: <none>

The results of mdadm --detail /dev/md0 are
/dev/md0:
        Version : 1.2
  Creation Time : Sun Jan  8 14:47:43 2012
     Raid Level : raid1
     Array Size : 1465136400 (1397.26 GiB 1500.30 GB)
  Used Dev Size : 1465136400 (1397.26 GiB 1500.30 GB)
   Raid Devices : 2
  Total Devices : 2
    Persistence : Superblock is persistent

    Update Time : Sun Jan  8 14:48:54 2012
          State : active, resyncing
 Active Devices : 2
Working Devices : 2
 Failed Devices : 0
  Spare Devices : 0

 Rebuild Status : 2% complete

           Name : office-desktop:0  (local to host office-desktop)
           UUID : bfc16c6e:4e8cb910:96ff7ed2:6fec32bc
         Events : 1

    Number   Major   Minor   RaidDevice State
       0       8       17        0      active sync   /dev/sdb1
       1       8       49        1      active sync   /dev/sdd1

When I try to mount this drive, however, I get 
mount: /dev/md0: can't read superblock

What do I need to do to complete the process?

Thanks

Jeff

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

On Sun, 2012-01-08 at 15:03 -0500, Jeff Cranmer wrote:
> On Sun, 2012-01-08 at 12:31 -0600, Paul Hartman wrote:
> > > 
> > > What is going on here?
> > 
> > (I didn't read this whole thread, sorry if I'm repeating someone else's
> > advice)
> > 
> > kernel autodetection only works on old superblock version 0.90, you're
> > using 1.2. Not a big deal, we use mdadm to do it.
> > 
> > Define your arrays in /etc/mdadm.conf and start /etc/init.d/mdadm in
> > your boot runscripts with "rc-update add mdadm boot", it will bring up
> > the array at boot time.
> > 
> > In my mdadm.conf i have a line like this:
> > 
> > ARRAY /dev/md1 metadata=1.01 name=black:1
> > UUID=8e653e72:9d5df6ba:bb66ea8b:02f1c317
> > 
> > (might be word-wrapped, should be all one line)
> > 
> > That's all that was needed to bring it up automatically at boot time.
> > 
> > Also AFAIR there was a "gotcha" about the hostname stored in the array's
> > metadata must match your machine's hostname or else mdadm auto-assemble
> > won't accept it (to protect you in case you're plugging disks from
> > another machine for recovery, you don't want it to use them as your main
> > drives), so in that case you must specify it explicitly or set the AUTO
> > parameter in mdadm.conf to accept this condition. If you created the
> > array from within a LiveCD or on another machine, the hostname might not
> > match your system.
> > 
> > See the mdadm manpage for more info.
> 
> mdadm was added to the default level, not boot.
> My /etc/mdadm.conf file has two active lines
> DEVICE /dev/sd[bcd]1
> ARRAY dev/md0 metadata=1.2 spares=1 name=office-desktop:0
> devices=/dev/sdb1,dev/sdc1,/dev/sdd1
> 
> It looks like I'm having trouble with a faulty /dev/sdc1, so what I'd
> like to do is wipe out the existing array and try starting a RAID1 array
> just with sdb1 and sdd1.
> 
> I got rid of the old array by using the commands
> mdadm --manage --fail /dev/md0
> mdadm  --manage --stop /dev/md0
> 
> I then used mdadm --verbose --create /dev/md0 --level=1
> --raid-devices=2 /dev/sdb1 /dev/sdd1
> 
> The result of this command was
> dadm: /dev/sdb1 appears to be part of a raid array:
>     level=raid5 devices=3 ctime=Sat Jan  7 08:16:00 2012
> mdadm: partition table exists on /dev/sdb1 but will be lost or
>        meaningless after creating array
> mdadm: Note: this array has metadata at the start and
>     may not be suitable as a boot device.  If you plan to
>     store '/boot' on this device please ensure that
>     your boot-loader understands md/v1.x metadata, or use
>     --metadata=0.90
> mdadm: /dev/sdd1 appears to be part of a raid array:
>     level=raid5 devices=3 ctime=Sat Jan  7 08:16:00 2012
> mdadm: size set to 1465136400K
> Continue creating array? y
> mdadm: Defaulting to version 1.2 metadata
> mdadm: array /dev/md0 started.
> 
> The results of cat /proc/mdstat are
> Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5]
> [raid4] [multipath] 
> md0 : active raid1 sdd1[1] sdb1[0]
>       1465136400 blocks super 1.2 [2/2] [UU]
>       [>....................]  resync =  2.1% (31838144/1465136400)
> finish=269.7min speed=88551K/sec
>       
> unused devices: <none>
> 
> Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5]
> [raid4] [multipath] 
> md0 : active raid1 sdd1[1] sdb1[0]
>       1465136400 blocks super 1.2 [2/2] [UU]
>       [>....................]  resync =  2.1% (31838144/1465136400)
> finish=269.7min speed=88551K/sec
>       
> unused devices: <none>
> 
> The results of mdadm --detail /dev/md0 are
> /dev/md0:
>         Version : 1.2
>   Creation Time : Sun Jan  8 14:47:43 2012
>      Raid Level : raid1
>      Array Size : 1465136400 (1397.26 GiB 1500.30 GB)
>   Used Dev Size : 1465136400 (1397.26 GiB 1500.30 GB)
>    Raid Devices : 2
>   Total Devices : 2
>     Persistence : Superblock is persistent
> 
>     Update Time : Sun Jan  8 14:48:54 2012
>           State : active, resyncing
>  Active Devices : 2
> Working Devices : 2
>  Failed Devices : 0
>   Spare Devices : 0
> 
>  Rebuild Status : 2% complete
> 
>            Name : office-desktop:0  (local to host office-desktop)
>            UUID : bfc16c6e:4e8cb910:96ff7ed2:6fec32bc
>          Events : 1
> 
>     Number   Major   Minor   RaidDevice State
>        0       8       17        0      active sync   /dev/sdb1
>        1       8       49        1      active sync   /dev/sdd1
> 
> When I try to mount this drive, however, I get 
> mount: /dev/md0: can't read superblock
> 
> What do I need to do to complete the process?
> 
> Thanks
> 
> Jeff
> 
> 
> 
Success - I managed to get a raid1 device operating.  
I created the final filesystem by using mkfs.xfs -f /dev/md0, then
waited for the rebuild to complete before rebooting the system.

It appears to be created successfully.  Now I'll try the same sequence
with sdb and sdc to see if sdc is a good disk.  If that works, I'll
retry a raid5 array tomorrow night.

Jeff

Re: [gentoo-user] How to get raid

[Jeff Cranmer]

> > 
> > 
> > 
> Success - I managed to get a raid1 device operating.  
> I created the final filesystem by using mkfs.xfs -f /dev/md0, then
> waited for the rebuild to complete before rebooting the system.
> 
> It appears to be created successfully.  Now I'll try the same sequence
> with sdb and sdc to see if sdc is a good disk.  If that works, I'll
> retry a raid5 array tomorrow night.
> 
Hmm - it seems to be a bug in RAID5 creation.
I can successfully create a RAID1 array either 
/dev/sdb1 and /dev/sdc1 or 
/dev/sdb1 and /dev/sdd1

If, however, I try to create a RAID5 array with all three elements, I
get /dev/sdc reporting a failure.

cat /proc/mdstat fails with the following report.

Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5]
[raid4] [multipath] 
md0 : active raid5 sdd1[3](S) sdc1[1](F) sdb1[0]
      2930272256 blocks super 1.2 level 5, 512k chunk, algorithm 2 [3/1]
[U__]
      
unused devices: <none>

Has anyone else experienced similar problems?  Is there an extra
diagnostic procedure which I can use to validate the sdc drive?

Is there something extra I have to do when I go over the 2TB level which
could explain this goofy behaviour?