mdadm create size problem

mdadm create size problem

[Herta Van den Eynde]

Red Hat AS 2.1 mdadm 1.11
I have two devices that used to be part of a now corrupt RAID10 
configuration.  When I try to reuse them to create a RAID1, it only uses 
about 1 third of the diskspace available.
# fdisk -l /dev/sdg

Disk /dev/sdg: 255 heads, 63 sectors, 8924 cylinders
Units = cylinders of 16065 * 512 bytes

    Device Boot    Start       End    Blocks   Id  System
/dev/sdg1             1      8924  71681998+  fd  Linux raid autodetect

# fdisk -l /dev/sdn

Disk /dev/sdn: 255 heads, 63 sectors, 8924 cylinders
Units = cylinders of 16065 * 512 bytes

    Device Boot    Start       End    Blocks   Id  System
/dev/sdn1             1      8924  71681998+  fd  Linux raid autodetect

# mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 /dev/sdg 
/dev/sdn
mdadm: /dev/sdg appears to contain an ext2fs file system
     size=419520K  mtime=Mon May  2 17:04:04 2005
mdadm: /dev/sdn appears to contain an ext2fs file system
     size=419520K  mtime=Mon May  2 17:04:04 2005
mdadm: size set to 139904K
Continue creating array? n

If I try to add the --size option, even a size that is 139904K or less, 
it tells me the size is invalid:

# mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 /dev/sdg 
/dev/sdn --size=139904K
mdadm: invalid size: 139904K

I did reboot the system after fdisk'ing the devices.

Do I need to low-level format these devices to be able to reuse them?

Kind regards,

Herta

-- 
Herta Van den Eynde              -=- Toledo system management
K.U. Leuven - Ludit              -=- phone: +32 (0)16 322 166
                                  -=- 50°51'27" N 004°40'39" E

"I wish I were two little cats.  Then I could play together."
-
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

Re: mdadm create size problem

[Herta Van den Eynde]

OK, there's definitely something wrong here.  I ran a
# mdadm --zero-superblock --force /dev/sdX
followed by
# dd if=/dev/zero of=/dev/sdX bs=64k
where X=[g-in-p], and when I try to create a RAID 1 using two of these
devices, it still sets the size to 139904K.  When I create a RAID 0
using the same devices, mdadm does set the size correctly.

# mdadm --create --verbose /dev/md6 --level=1 --raid-devices=2 /dev/sdh
/dev/sdo
mdadm: size set to 139904K
mdadm: array /dev/md6 started.

# mdadm --create --verbose /dev/md6 --level=0 --raid-devices=2 /dev/sdh
/dev/sdo
mdadm: chunk size defaults to 64K
mdadm: array /dev/md6 started.
# mke2fs -j -v  /dev/md6
(...)
# mount /dev/md6 /mnt; df /mnt
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/md6             141124600     32828 133923052   1% /mnt

Is this a bug, or am I overlooking something?

Herta


Herta Van den Eynde wrote:
> Red Hat AS 2.1 mdadm 1.11
> I have two devices that used to be part of a now corrupt RAID10 
> configuration.  When I try to reuse them to create a RAID1, it only uses 
> about 1 third of the diskspace available.
> # fdisk -l /dev/sdg
> 
> Disk /dev/sdg: 255 heads, 63 sectors, 8924 cylinders
> Units = cylinders of 16065 * 512 bytes
> 
>    Device Boot    Start       End    Blocks   Id  System
> /dev/sdg1             1      8924  71681998+  fd  Linux raid autodetect
> 
> # fdisk -l /dev/sdn
> 
> Disk /dev/sdn: 255 heads, 63 sectors, 8924 cylinders
> Units = cylinders of 16065 * 512 bytes
> 
>    Device Boot    Start       End    Blocks   Id  System
> /dev/sdn1             1      8924  71681998+  fd  Linux raid autodetect
> 
> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 /dev/sdg 
> /dev/sdn
> mdadm: /dev/sdg appears to contain an ext2fs file system
>     size=419520K  mtime=Mon May  2 17:04:04 2005
> mdadm: /dev/sdn appears to contain an ext2fs file system
>     size=419520K  mtime=Mon May  2 17:04:04 2005
> mdadm: size set to 139904K
> Continue creating array? n
> 
> If I try to add the --size option, even a size that is 139904K or less, 
> it tells me the size is invalid:
> 
> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 /dev/sdg 
> /dev/sdn --size=139904K
> mdadm: invalid size: 139904K
> 
> I did reboot the system after fdisk'ing the devices.
> 
> Do I need to low-level format these devices to be able to reuse them?
> 
> Kind regards,
> 
> Herta
> 

-- 
Herta Van den Eynde              -=- Toledo system management
K.U. Leuven - Ludit              -=- phone: +32 (0)16 322 166
                                  -=- 50°51'27" N 004°40'39" E

"I wish I were two little cats.  Then I could play together."

-
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

Re: mdadm create size problem

[Herta Van den Eynde]

FWIIW, I reverted to mdadm-1.4.0-1, and that version creates the RAID1+0 
just fine.

Herta

Herta Van den Eynde wrote:
> OK, there's definitely something wrong here.  I ran a
> # mdadm --zero-superblock --force /dev/sdX
> followed by
> # dd if=/dev/zero of=/dev/sdX bs=64k
> where X=[g-in-p], and when I try to create a RAID 1 using two of these
> devices, it still sets the size to 139904K.  When I create a RAID 0
> using the same devices, mdadm does set the size correctly.
> 
> # mdadm --create --verbose /dev/md6 --level=1 --raid-devices=2 /dev/sdh
> /dev/sdo
> mdadm: size set to 139904K
> mdadm: array /dev/md6 started.
> 
> # mdadm --create --verbose /dev/md6 --level=0 --raid-devices=2 /dev/sdh
> /dev/sdo
> mdadm: chunk size defaults to 64K
> mdadm: array /dev/md6 started.
> # mke2fs -j -v  /dev/md6
> (...)
> # mount /dev/md6 /mnt; df /mnt
> Filesystem           1k-blocks      Used Available Use% Mounted on
> /dev/md6             141124600     32828 133923052   1% /mnt
> 
> Is this a bug, or am I overlooking something?
> 
> Herta
> 
> 
> Herta Van den Eynde wrote:
> 
>> Red Hat AS 2.1 mdadm 1.11
>> I have two devices that used to be part of a now corrupt RAID10 
>> configuration.  When I try to reuse them to create a RAID1, it only 
>> uses about 1 third of the diskspace available.
>> # fdisk -l /dev/sdg
>>
>> Disk /dev/sdg: 255 heads, 63 sectors, 8924 cylinders
>> Units = cylinders of 16065 * 512 bytes
>>
>>    Device Boot    Start       End    Blocks   Id  System
>> /dev/sdg1             1      8924  71681998+  fd  Linux raid autodetect
>>
>> # fdisk -l /dev/sdn
>>
>> Disk /dev/sdn: 255 heads, 63 sectors, 8924 cylinders
>> Units = cylinders of 16065 * 512 bytes
>>
>>    Device Boot    Start       End    Blocks   Id  System
>> /dev/sdn1             1      8924  71681998+  fd  Linux raid autodetect
>>
>> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 
>> /dev/sdg /dev/sdn
>> mdadm: /dev/sdg appears to contain an ext2fs file system
>>     size=419520K  mtime=Mon May  2 17:04:04 2005
>> mdadm: /dev/sdn appears to contain an ext2fs file system
>>     size=419520K  mtime=Mon May  2 17:04:04 2005
>> mdadm: size set to 139904K
>> Continue creating array? n
>>
>> If I try to add the --size option, even a size that is 139904K or 
>> less, it tells me the size is invalid:
>>
>> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 
>> /dev/sdg /dev/sdn --size=139904K
>> mdadm: invalid size: 139904K
>>
>> I did reboot the system after fdisk'ing the devices.
>>
>> Do I need to low-level format these devices to be able to reuse them?
>>
>> Kind regards,
>>
>> Herta
>>
> 

-- 
Herta Van den Eynde              -=- Toledo system management
K.U. Leuven - Ludit              -=- phone: +32 (0)16 322 166
                                  -=- 50°51'27" N 004°40'39" E

"I wish I were two little cats.  Then I could play together."
-
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

Re: mdadm create size problem

[Tyler]

What version were you trying?  1.9.0 and 1.11.0 are considered the 
latest stable versions i believe.

Regards,
Tyler.

Herta Van den Eynde wrote:

> FWIIW, I reverted to mdadm-1.4.0-1, and that version creates the 
> RAID1+0 just fine.
>
> Herta
>
> Herta Van den Eynde wrote:
>
>> OK, there's definitely something wrong here.  I ran a
>> # mdadm --zero-superblock --force /dev/sdX
>> followed by
>> # dd if=/dev/zero of=/dev/sdX bs=64k
>> where X=[g-in-p], and when I try to create a RAID 1 using two of these
>> devices, it still sets the size to 139904K.  When I create a RAID 0
>> using the same devices, mdadm does set the size correctly.
>>
>> # mdadm --create --verbose /dev/md6 --level=1 --raid-devices=2 /dev/sdh
>> /dev/sdo
>> mdadm: size set to 139904K
>> mdadm: array /dev/md6 started.
>>
>> # mdadm --create --verbose /dev/md6 --level=0 --raid-devices=2 /dev/sdh
>> /dev/sdo
>> mdadm: chunk size defaults to 64K
>> mdadm: array /dev/md6 started.
>> # mke2fs -j -v  /dev/md6
>> (...)
>> # mount /dev/md6 /mnt; df /mnt
>> Filesystem           1k-blocks      Used Available Use% Mounted on
>> /dev/md6             141124600     32828 133923052   1% /mnt
>>
>> Is this a bug, or am I overlooking something?
>>
>> Herta
>>
>>
>> Herta Van den Eynde wrote:
>>
>>> Red Hat AS 2.1 mdadm 1.11
>>> I have two devices that used to be part of a now corrupt RAID10 
>>> configuration.  When I try to reuse them to create a RAID1, it only 
>>> uses about 1 third of the diskspace available.
>>> # fdisk -l /dev/sdg
>>>
>>> Disk /dev/sdg: 255 heads, 63 sectors, 8924 cylinders
>>> Units = cylinders of 16065 * 512 bytes
>>>
>>>    Device Boot    Start       End    Blocks   Id  System
>>> /dev/sdg1             1      8924  71681998+  fd  Linux raid autodetect
>>>
>>> # fdisk -l /dev/sdn
>>>
>>> Disk /dev/sdn: 255 heads, 63 sectors, 8924 cylinders
>>> Units = cylinders of 16065 * 512 bytes
>>>
>>>    Device Boot    Start       End    Blocks   Id  System
>>> /dev/sdn1             1      8924  71681998+  fd  Linux raid autodetect
>>>
>>> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 
>>> /dev/sdg /dev/sdn
>>> mdadm: /dev/sdg appears to contain an ext2fs file system
>>>     size=419520K  mtime=Mon May  2 17:04:04 2005
>>> mdadm: /dev/sdn appears to contain an ext2fs file system
>>>     size=419520K  mtime=Mon May  2 17:04:04 2005
>>> mdadm: size set to 139904K
>>> Continue creating array? n
>>>
>>> If I try to add the --size option, even a size that is 139904K or 
>>> less, it tells me the size is invalid:
>>>
>>> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 
>>> /dev/sdg /dev/sdn --size=139904K
>>> mdadm: invalid size: 139904K
>>>
>>> I did reboot the system after fdisk'ing the devices.
>>>
>>> Do I need to low-level format these devices to be able to reuse them?
>>>
>>> Kind regards,
>>>
>>> Herta
>>

Re: mdadm create size problem

[Herta Van den Eynde]

I was trying mdadm-1.11.0 which I picked up from
http://www.be.kernel.org/pub/linux/utils/raid/mdadm/mdadm-1.11.0.tar.gz
on Red Hat Advanced Server 2.1 most recent kernel update:
# cat /proc/version
Linux version 2.4.9-e.62enterprise (bhcompile@tweety.build.redhat.com) 
(gcc version 2.96 20000731 (Red Hat Linux 7.2 2.96-129.7.2)) #1 SMP Fri 
Apr 8 18:46:17 EDT 2005

Kind regards,

Herta

Tyler wrote:
> What version were you trying?  1.9.0 and 1.11.0 are considered the 
> latest stable versions i believe.
> 
> Regards,
> Tyler.
> 
> Herta Van den Eynde wrote:
> 
>> FWIIW, I reverted to mdadm-1.4.0-1, and that version creates the 
>> RAID1+0 just fine.
>>
>> Herta
>>
>> Herta Van den Eynde wrote:
>>
>>> OK, there's definitely something wrong here.  I ran a
>>> # mdadm --zero-superblock --force /dev/sdX
>>> followed by
>>> # dd if=/dev/zero of=/dev/sdX bs=64k
>>> where X=[g-in-p], and when I try to create a RAID 1 using two of these
>>> devices, it still sets the size to 139904K.  When I create a RAID 0
>>> using the same devices, mdadm does set the size correctly.
>>>
>>> # mdadm --create --verbose /dev/md6 --level=1 --raid-devices=2 /dev/sdh
>>> /dev/sdo
>>> mdadm: size set to 139904K
>>> mdadm: array /dev/md6 started.
>>>
>>> # mdadm --create --verbose /dev/md6 --level=0 --raid-devices=2 /dev/sdh
>>> /dev/sdo
>>> mdadm: chunk size defaults to 64K
>>> mdadm: array /dev/md6 started.
>>> # mke2fs -j -v  /dev/md6
>>> (...)
>>> # mount /dev/md6 /mnt; df /mnt
>>> Filesystem           1k-blocks      Used Available Use% Mounted on
>>> /dev/md6             141124600     32828 133923052   1% /mnt
>>>
>>> Is this a bug, or am I overlooking something?
>>>
>>> Herta
>>>
>>>
>>> Herta Van den Eynde wrote:
>>>
>>>> Red Hat AS 2.1 mdadm 1.11
>>>> I have two devices that used to be part of a now corrupt RAID10 
>>>> configuration.  When I try to reuse them to create a RAID1, it only 
>>>> uses about 1 third of the diskspace available.
>>>> # fdisk -l /dev/sdg
>>>>
>>>> Disk /dev/sdg: 255 heads, 63 sectors, 8924 cylinders
>>>> Units = cylinders of 16065 * 512 bytes
>>>>
>>>>    Device Boot    Start       End    Blocks   Id  System
>>>> /dev/sdg1             1      8924  71681998+  fd  Linux raid autodetect
>>>>
>>>> # fdisk -l /dev/sdn
>>>>
>>>> Disk /dev/sdn: 255 heads, 63 sectors, 8924 cylinders
>>>> Units = cylinders of 16065 * 512 bytes
>>>>
>>>>    Device Boot    Start       End    Blocks   Id  System
>>>> /dev/sdn1             1      8924  71681998+  fd  Linux raid autodetect
>>>>
>>>> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 
>>>> /dev/sdg /dev/sdn
>>>> mdadm: /dev/sdg appears to contain an ext2fs file system
>>>>     size=419520K  mtime=Mon May  2 17:04:04 2005
>>>> mdadm: /dev/sdn appears to contain an ext2fs file system
>>>>     size=419520K  mtime=Mon May  2 17:04:04 2005
>>>> mdadm: size set to 139904K
>>>> Continue creating array? n
>>>>
>>>> If I try to add the --size option, even a size that is 139904K or 
>>>> less, it tells me the size is invalid:
>>>>
>>>> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 
>>>> /dev/sdg /dev/sdn --size=139904K
>>>> mdadm: invalid size: 139904K
>>>>
>>>> I did reboot the system after fdisk'ing the devices.
>>>>
>>>> Do I need to low-level format these devices to be able to reuse them?
>>>>
>>>> Kind regards,
>>>>
>>>> Herta
>>>
>>>
> -
> 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
> 

-- 
Herta Van den Eynde              -=- Toledo system management
K.U. Leuven - Ludit              -=- phone: +32 (0)16 322 166
                                  -=- 50°51'27" N 004°40'39" E

"I wish I were two little cats.  Then I could play together."
-
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

Re: mdadm create size problem

[Neil Brown]

On Tuesday May 3, herta.vandeneynde@cc.kuleuven.be wrote:
(sorry for the delay in looking at this).

> Red Hat AS 2.1 mdadm 1.11
> I have two devices that used to be part of a now corrupt RAID10 
> configuration.  When I try to reuse them to create a RAID1, it only uses 
> about 1 third of the diskspace available.
> # fdisk -l /dev/sdg
> 
> Disk /dev/sdg: 255 heads, 63 sectors, 8924 cylinders
> Units = cylinders of 16065 * 512 bytes
...
> mdadm: size set to 139904K
> Continue creating array? n

Yep, there is a bug in 1.11.0.
In Create.c, line 209,
		else {
			ldsize = dsize;
			dsize <<= 9;
		}
should be
		else {
			ldsize = dsize;
			ldsize <<= 9;
		}

It only affects systems where ioctl(BLKGETSIZE64) doesn't exist or
fails.

> 
> If I try to add the --size option, even a size that is 139904K or less, 
> it tells me the size is invalid:
> 
> # mdadm --create --verbose /dev/md5 --level=1 --raid-devices=2 /dev/sdg 
> /dev/sdn --size=139904K
> mdadm: invalid size: 139904K

size should be a simple integer, no suffix, and it wouldn't let you
create an array using more space than it thinks exists anyway.


I guess it is time for a 1.12 ??

NeilBrown

Re: mdadm create size problem

[Luca Berra]

On Tue, May 24, 2005 at 09:43:20AM +1000, Neil Brown wrote:
>I guess it is time for a 1.12 ??
>
in this case please recheck the ordering issue, since mdadm -Ss does not
yet work properly.

L.

-- 
Luca Berra -- bluca@comedia.it
        Communication Media & Services S.r.l.
 /"\
 \ /     ASCII RIBBON CAMPAIGN
  X        AGAINST HTML MAIL
 / \

Re: mdadm create size problem

[Neil Brown]

On Tuesday May 24, bluca@comedia.it wrote:
> On Tue, May 24, 2005 at 09:43:20AM +1000, Neil Brown wrote:
> >I guess it is time for a 1.12 ??
> >
> in this case please recheck the ordering issue, since mdadm -Ss does not
> yet work properly.

Thanks for the reminder.
I think I'll just do it as a mutli-pass.
i.e. try to stop everything quietly in any order and as long as
something succeeded and something failed, try again.
Then one last try in non-quite mode to get the right error messages.

NeilBrown

Re: mdadm create size problem

[Luca Berra]

On Tue, May 24, 2005 at 08:51:36PM +1000, Neil Brown wrote:
>On Tuesday May 24, bluca@comedia.it wrote:
>> On Tue, May 24, 2005 at 09:43:20AM +1000, Neil Brown wrote:
>> >I guess it is time for a 1.12 ??
>> >
>> in this case please recheck the ordering issue, since mdadm -Ss does not
>> yet work properly.
>
>Thanks for the reminder.
>I think I'll just do it as a mutli-pass.
>i.e. try to stop everything quietly in any order and as long as
>something succeeded and something failed, try again.
>Then one last try in non-quite mode to get the right error messages.
>
this is a good idea, i'd do the same for start, so we do not depend on
correct ordering in /etc/mdadm.conf

L.

-- 
Luca Berra -- bluca@comedia.it
        Communication Media & Services S.r.l.
 /"\
 \ /     ASCII RIBBON CAMPAIGN
  X        AGAINST HTML MAIL
 / \